fetch config once

* Add some pool settings

* fmt

.circleci/config.yml

@@ -9,39 +9,50 @@ jobs:
     # Specify the execution environment. You can specify an image from Dockerhub or use one of our Convenience Images from CircleCI's Developer Hub.
     # See: https://circleci.com/docs/2.0/configuration-reference/#docker-machine-macos-windows-executor
     docker:
-      - image: levkk/pgcat-ci:latest
+      - image: ghcr.io/levkk/pgcat-ci:1.67
         environment:
           RUST_LOG: info
-          RUSTFLAGS: "-C instrument-coverage"
-          LLVM_PROFILE_FILE: "pgcat-%m.profraw"
+          LLVM_PROFILE_FILE: /tmp/pgcat-%m-%p.profraw
+          RUSTC_BOOTSTRAP: 1
+          CARGO_INCREMENTAL: 0
+          RUSTFLAGS: "-Zprofile -Ccodegen-units=1 -Copt-level=0 -Clink-dead-code -Coverflow-checks=off -Zpanic_abort_tests -Cpanic=abort -Cinstrument-coverage"
+          RUSTDOCFLAGS: "-Cpanic=abort"
       - image: postgres:14
-        command: ["postgres", "-p", "5432", "-c", "shared_preload_libraries=pg_stat_statements"]
+        command: ["postgres", "-p", "5432", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-c", "pg_stat_statements.max=100000"]
         environment:
           POSTGRES_USER: postgres
           POSTGRES_DB: postgres
           POSTGRES_PASSWORD: postgres
-          POSTGRES_HOST_AUTH_METHOD: scram-sha-256
+          POSTGRES_INITDB_ARGS: --auth-local=md5 --auth-host=md5 --auth=md5
       - image: postgres:14
-        command: ["postgres", "-p", "7432", "-c", "shared_preload_libraries=pg_stat_statements"]
+        command: ["postgres", "-p", "7432", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-c", "pg_stat_statements.max=100000"]
         environment:
           POSTGRES_USER: postgres
           POSTGRES_DB: postgres
           POSTGRES_PASSWORD: postgres
-          POSTGRES_HOST_AUTH_METHOD: scram-sha-256
+          POSTGRES_INITDB_ARGS: --auth-local=scram-sha-256 --auth-host=scram-sha-256 --auth=scram-sha-256
       - image: postgres:14
-        command: ["postgres", "-p", "8432", "-c", "shared_preload_libraries=pg_stat_statements"]
+        command: ["postgres", "-p", "8432", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-c", "pg_stat_statements.max=100000"]
         environment:
           POSTGRES_USER: postgres
           POSTGRES_DB: postgres
           POSTGRES_PASSWORD: postgres
-          POSTGRES_HOST_AUTH_METHOD: scram-sha-256
+          POSTGRES_INITDB_ARGS: --auth-local=scram-sha-256 --auth-host=scram-sha-256 --auth=scram-sha-256
       - image: postgres:14
-        command: ["postgres", "-p", "9432", "-c", "shared_preload_libraries=pg_stat_statements"]
+        command: ["postgres", "-p", "9432", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-c", "pg_stat_statements.max=100000"]
         environment:
           POSTGRES_USER: postgres
           POSTGRES_DB: postgres
           POSTGRES_PASSWORD: postgres
-          POSTGRES_HOST_AUTH_METHOD: scram-sha-256
+          POSTGRES_INITDB_ARGS: --auth-local=scram-sha-256 --auth-host=scram-sha-256 --auth=scram-sha-256
+
+      - image: postgres:14
+        command: ["postgres", "-p", "10432", "-c", "shared_preload_libraries=pg_stat_statements"]
+        environment:
+          POSTGRES_USER: postgres
+          POSTGRES_DB: postgres
+          POSTGRES_PASSWORD: postgres
+          POSTGRES_INITDB_ARGS: --auth-local=md5 --auth-host=md5 --auth=md5
 
     # Add steps to the job
     # See: https://circleci.com/docs/2.0/configuration-reference/#steps
@@ -52,18 +63,9 @@ jobs:
       - run:
           name: "Lint"
           command: "cargo fmt --check"
-      - run:
-          name: "Install dependencies"
-          command: "sudo apt-get update && sudo apt-get install -y psmisc postgresql-contrib-12 postgresql-client-12 ruby ruby-dev libpq-dev python3 python3-pip lcov llvm-11 && sudo apt-get upgrade curl"
-      - run:
-          name: "Install rust tools"
-          command: "cargo install cargo-binutils rustfilt && rustup component add llvm-tools-preview"
-      - run:
-          name: "Build"
-          command: "cargo build"
       - run:
           name: "Tests"
-          command: "cargo test && bash .circleci/run_tests.sh && .circleci/generate_coverage.sh"
+          command: "cargo clean && cargo build && cargo test && bash .circleci/run_tests.sh && .circleci/generate_coverage.sh"
       - store_artifacts:
           path: /tmp/cov
           destination: coverage-data

.circleci/generate_coverage.sh

@@ -1,7 +1,15 @@
 #!/bin/bash
 
-rust-profdata merge -sparse pgcat-*.profraw -o pgcat.profdata
+# inspired by https://doc.rust-lang.org/rustc/instrument-coverage.html#tips-for-listing-the-binaries-automatically
+TEST_OBJECTS=$( \
+    for file in $(cargo test --no-run 2>&1 | grep "target/debug/deps/pgcat-[[:alnum:]]\+" -o); \
+    do \
+        printf "%s %s " --object $file; \
+    done \
+)
 
-rust-cov export -ignore-filename-regex="rustc|registry" -Xdemangler=rustfilt -instr-profile=pgcat.profdata --object ./target/debug/pgcat --format lcov > ./lcov.info
+rust-profdata merge -sparse /tmp/pgcat-*.profraw -o /tmp/pgcat.profdata
 
-genhtml lcov.info --output-directory /tmp/cov --prefix $(pwd)
+bash -c "rust-cov export -ignore-filename-regex='rustc|registry' -Xdemangler=rustfilt -instr-profile=/tmp/pgcat.profdata $TEST_OBJECTS --object ./target/debug/pgcat --format lcov > ./lcov.info"
+
+genhtml lcov.info  --title "PgCat Code Coverage" --css-file ./cov-style.css --no-function-coverage --highlight --ignore-errors source --legend  --output-directory /tmp/cov --prefix $(pwd)

.circleci/pgcat.toml

@@ -18,10 +18,10 @@ enable_prometheus_exporter = true
 prometheus_exporter_port = 9930
 
 # How long to wait before aborting a server connection (ms).
-connect_timeout = 100
+connect_timeout = 1000
 
 # How much time to give the health check query to return with a result (ms).
-healthcheck_timeout = 100
+healthcheck_timeout = 1000
 
 # How long to keep connection available for immediate re-use, without running a healthcheck query on it
 healthcheck_delay = 30000
@@ -32,8 +32,14 @@ shutdown_timeout = 5000
 # For how long to ban a server if it fails a health check (seconds).
 ban_time = 60 # Seconds
 
+# If we should log client connections
+log_client_connections = false
+
+# If we should log client disconnections
+log_client_disconnections = false
+
 # Reload config automatically if it changes.
-autoreload = true
+autoreload = 15000
 
 # TLS
 tls_certificate = ".circleci/server.cert"

.circleci/run_tests.sh

@@ -19,15 +19,12 @@ PGPASSWORD=postgres psql -e -h 127.0.0.1 -p 5432 -U postgres -f tests/sharding/q
 PGPASSWORD=postgres psql -e -h 127.0.0.1 -p 7432 -U postgres -f tests/sharding/query_routing_setup.sql
 PGPASSWORD=postgres psql -e -h 127.0.0.1 -p 8432 -U postgres -f tests/sharding/query_routing_setup.sql
 PGPASSWORD=postgres psql -e -h 127.0.0.1 -p 9432 -U postgres -f tests/sharding/query_routing_setup.sql
+PGPASSWORD=postgres psql -e -h 127.0.0.1 -p 10432 -U postgres -f tests/sharding/query_routing_setup.sql
 
 PGPASSWORD=sharding_user pgbench -h 127.0.0.1 -U sharding_user shard0 -i
 PGPASSWORD=sharding_user pgbench -h 127.0.0.1 -U sharding_user shard1 -i
 PGPASSWORD=sharding_user pgbench -h 127.0.0.1 -U sharding_user shard2 -i
 
-# Install Toxiproxy to simulate a downed/slow database
-wget -O toxiproxy-2.4.0.deb https://github.com/Shopify/toxiproxy/releases/download/v2.4.0/toxiproxy_2.4.0_linux_$(dpkg --print-architecture).deb
-sudo dpkg -i toxiproxy-2.4.0.deb
-
 # Start Toxiproxy
 LOG_LEVEL=error toxiproxy-server &
 sleep 1
@@ -56,6 +53,17 @@ psql -U sharding_user -h 127.0.0.1 -p 6432 -c 'COPY (SELECT * FROM pgbench_accou
 sleep 1
 killall psql -s SIGINT
 
+# Pause/resume test.
+# Running benches before, during, and after pause/resume.
+pgbench -U sharding_user -t 500 -c 2 -h 127.0.0.1 -p 6432 --protocol extended &
+BENCH_ONE=$!
+PGPASSWORD=admin_pass psql -U admin_user -h 127.0.0.1 -p 6432 -d pgbouncer -c 'PAUSE sharded_db,sharding_user'
+pgbench -U sharding_user -h 127.0.0.1 -p 6432 -t 500 -c 2 --protocol extended &
+BENCH_TWO=$!
+PGPASSWORD=admin_pass psql -U admin_user -h 127.0.0.1 -p 6432 -d pgbouncer -c 'RESUME sharded_db,sharding_user'
+wait ${BENCH_ONE}
+wait ${BENCH_TWO}
+
 # Reload pool (closing unused server connections)
 PGPASSWORD=admin_pass psql -U admin_user -h 127.0.0.1 -p 6432 -d pgbouncer -c 'RELOAD'
 
@@ -85,13 +93,12 @@ sed -i 's/statement_timeout = 100/statement_timeout = 0/' .circleci/pgcat.toml
 kill -SIGHUP $(pgrep pgcat) # Reload config again
 
 #
-# ActiveRecord tests
+# Integration tests and ActiveRecord tests
 #
 cd tests/ruby
-sudo gem install bundler
-bundle install
-bundle exec ruby tests.rb || exit 1
-bundle exec rspec *_spec.rb || exit 1
+sudo bundle install
+bundle exec ruby tests.rb --format documentation || exit 1
+bundle exec rspec *_spec.rb --format documentation || exit 1
 cd ../..
 
 #

.dockerignore

@@ -2,3 +2,5 @@ target/
 tests/
 tracing/
 .circleci/
+.git/
+dev/

.editorconfig

@@ -0,0 +1,14 @@
+root = true
+
+[*]
+trim_trailing_whitespace = true
+insert_final_newline = true
+
+[*.rs]
+indent_style = space
+indent_size = 4
+max_line_length = 120
+
+[*.toml]
+indent_style = space
+indent_size = 2

.github/dependabot.yml

@@ -0,0 +1,12 @@
+version: 2
+updates:
+  - package-ecosystem: "cargo"
+    directory: "/"
+    schedule:
+      interval: "daily"
+      time: "04:00" # UTC
+    labels:
+      - "dependencies"
+    commit-message:
+      prefix: "chore(deps)"
+    open-pull-requests-limit: 10

.github/workflows/build-and-push.yaml

@@ -0,0 +1,54 @@
+name: Build and Push
+
+on: push
+
+env:
+  registry: ghcr.io
+  image-name: ${{ github.repository }}
+
+jobs:
+  build-and-push:
+    runs-on: ubuntu-latest
+
+    permissions:
+      contents: read
+      packages: write
+
+    steps:
+      - name: Checkout Repository
+        uses: actions/checkout@v3
+
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@v2
+
+      - name: Determine tags
+        id: metadata
+        uses: docker/metadata-action@v4
+        with:
+          images: ${{ env.registry }}/${{ env.image-name }}
+          tags: |
+            type=sha,prefix=,format=long
+            type=schedule
+            type=ref,event=branch
+            type=ref,event=pr
+            type=raw,value=latest,enable={{ is_default_branch }}
+
+      - name: Log in to the Container registry
+        uses: docker/login-action@v2.1.0
+        with:
+          registry: ${{ env.registry }}
+          username: ${{ github.actor }}
+          password: ${{ secrets.GITHUB_TOKEN }}
+
+      - name: Build and push ${{ env.image-name }}
+        uses: docker/build-push-action@v3
+        with:
+          push: true
+          tags: ${{ steps.metadata.outputs.tags }}
+          labels: ${{ steps.metadata.outputs.labels }}
+          cache-from: type=gha
+          cache-to: type=gha,mode=max
+
+concurrency:
+  group: ${{ github.ref }}
+  cancel-in-progress: true

.github/workflows/publish-ci-docker-image.yml

@@ -0,0 +1,20 @@
+name: publish-ci-docker-image
+on:
+  push:
+    branches: [ main ]
+jobs:
+  publish-ci-docker-image:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v2
+      - name: Login to GitHub Container Registry
+        uses: docker/login-action@v1
+        with:
+          registry: ghcr.io
+          username: ${{ github.actor }}
+          password: ${{ secrets.GITHUB_TOKEN }}
+      - name: Build CI Docker image
+        run: |
+          docker build . -f Dockerfile.ci --tag ghcr.io/postgresml/pgcat-ci:latest
+          docker run ghcr.io/postgresml/pgcat-ci:latest
+          docker push ghcr.io/postgresml/pgcat-ci:latest

.gitignore

@@ -1,4 +1,12 @@
 .idea
 /target
 *.deb
-.vscode
+.vscode
+*.profraw
+cov/
+lcov.info
+
+# Dev
+dev/.bash_history
+dev/cache
+!dev/cache/.keepme

.rustfmt.toml

@@ -0,0 +1,2 @@
+edition = "2021"
+hard_tabs = false

CONFIG.md

@@ -0,0 +1,409 @@
+# PgCat Configurations 
+## `general` Section
+
+### host
+```
+path: general.host
+default: "0.0.0.0"
+```
+
+What IP to run on, 0.0.0.0 means accessible from everywhere.
+
+### port
+```
+path: general.port
+default: 6432
+```
+
+Port to run on, same as PgBouncer used in this example.
+
+### enable_prometheus_exporter
+```
+path: general.enable_prometheus_exporter
+default: true
+```
+
+Whether to enable prometheus exporter or not.
+
+### prometheus_exporter_port
+```
+path: general.prometheus_exporter_port
+default: 9930
+```
+
+Port at which prometheus exporter listens on.
+
+### connect_timeout
+```
+path: general.connect_timeout
+default: 5000 # milliseconds
+```
+
+How long to wait before aborting a server connection (ms).
+
+### idle_timeout
+```
+path: general.idle_timeout
+default: 30000 # milliseconds
+```
+
+How long an idle connection with a server is left open (ms).
+
+### server_lifetime
+```
+path: general.server_lifetime
+default: 86400000 # 24 hours
+```
+
+Max connection lifetime before it's closed, even if actively used.
+
+### idle_client_in_transaction_timeout
+```
+path: general.idle_client_in_transaction_timeout
+default: 0 # milliseconds
+```
+
+How long a client is allowed to be idle while in a transaction (ms).
+
+### healthcheck_timeout
+```
+path: general.healthcheck_timeout
+default: 1000 # milliseconds
+```
+
+How much time to give the health check query to return with a result (ms).
+
+### healthcheck_delay
+```
+path: general.healthcheck_delay
+default: 30000 # milliseconds
+```
+
+How long to keep connection available for immediate re-use, without running a healthcheck query on it
+
+### shutdown_timeout
+```
+path: general.shutdown_timeout
+default: 60000 # milliseconds
+```
+
+How much time to give clients during shutdown before forcibly killing client connections (ms).
+
+### ban_time
+```
+path: general.ban_time
+default: 60 # seconds
+```
+
+How long to ban a server if it fails a health check (seconds).
+
+### log_client_connections
+```
+path: general.log_client_connections
+default: false
+```
+
+If we should log client connections
+
+### log_client_disconnections
+```
+path: general.log_client_disconnections
+default: false
+```
+
+If we should log client disconnections
+
+### autoreload
+```
+path: general.autoreload
+default: 15000
+```
+
+When set to true, PgCat reloads configs if it detects a change in the config file.
+
+### worker_threads
+```
+path: general.worker_threads
+default: 5
+```
+
+Number of worker threads the Runtime will use (4 by default).
+
+### tcp_keepalives_idle
+```
+path: general.tcp_keepalives_idle
+default: 5
+```
+
+Number of seconds of connection idleness to wait before sending a keepalive packet to the server.
+
+### tcp_keepalives_count
+```
+path: general.tcp_keepalives_count
+default: 5
+```
+
+Number of unacknowledged keepalive packets allowed before giving up and closing the connection.
+
+### tcp_keepalives_interval
+```
+path: general.tcp_keepalives_interval
+default: 5
+```
+
+Number of seconds between keepalive packets.
+
+### tls_certificate
+```
+path: general.tls_certificate
+default: <UNSET>
+example: "server.cert"
+```
+
+Path to TLS Certificate file to use for TLS connections
+
+### tls_private_key
+```
+path: general.tls_private_key
+default: <UNSET>
+example: "server.key"
+```
+
+Path to TLS private key file to use for TLS connections
+
+### admin_username
+```
+path: general.admin_username
+default: "admin_user"
+```
+
+User name to access the virtual administrative database (pgbouncer or pgcat)
+Connecting to that database allows running commands like `SHOW POOLS`, `SHOW DATABASES`, etc..
+
+### admin_password
+```
+path: general.admin_password
+default: "admin_pass"
+```
+
+Password to access the virtual administrative database
+
+## `pools.<pool_name>` Section
+
+### pool_mode
+```
+path: pools.<pool_name>.pool_mode
+default: "transaction"
+```
+
+Pool mode (see PgBouncer docs for more).
+`session` one server connection per connected client
+`transaction` one server connection per client transaction
+
+### load_balancing_mode
+```
+path: pools.<pool_name>.load_balancing_mode
+default: "random"
+```
+
+Load balancing mode
+`random` selects the server at random
+`loc` selects the server with the least outstanding busy conncetions
+
+### default_role
+```
+path: pools.<pool_name>.default_role
+default: "any"
+```
+
+If the client doesn't specify, PgCat routes traffic to this role by default.
+`any` round-robin between primary and replicas,
+`replica` round-robin between replicas only without touching the primary,
+`primary` all queries go to the primary unless otherwise specified.
+
+### query_parser_enabled
+```
+path: pools.<pool_name>.query_parser_enabled
+default: true
+```
+
+If Query Parser is enabled, we'll attempt to parse
+every incoming query to determine if it's a read or a write.
+If it's a read query, we'll direct it to a replica. Otherwise, if it's a write,
+we'll direct it to the primary.
+
+### primary_reads_enabled
+```
+path: pools.<pool_name>.primary_reads_enabled
+default: true
+```
+
+If the query parser is enabled and this setting is enabled, the primary will be part of the pool of databases used for
+load balancing of read queries. Otherwise, the primary will only be used for write
+queries. The primary can always be explicitly selected with our custom protocol.
+
+### sharding_key_regex
+```
+path: pools.<pool_name>.sharding_key_regex
+default: <UNSET>
+example: '/\* sharding_key: (\d+) \*/'
+```
+
+Allow sharding commands to be passed as statement comments instead of
+separate commands. If these are unset this functionality is disabled.
+
+### sharding_function
+```
+path: pools.<pool_name>.sharding_function
+default: "pg_bigint_hash"
+```
+
+So what if you wanted to implement a different hashing function,
+or you've already built one and you want this pooler to use it?
+Current options:
+`pg_bigint_hash`: PARTITION BY HASH (Postgres hashing function)
+`sha1`: A hashing function based on SHA1
+
+### auth_query
+```
+path: pools.<pool_name>.auth_query
+default: <UNSET>
+example: "SELECT $1"
+```
+
+Query to be sent to servers to obtain the hash used for md5 authentication. The connection will be
+established using the database configured in the pool. This parameter is inherited by every pool
+and can be redefined in pool configuration.
+
+### auth_query_user
+```
+path: pools.<pool_name>.auth_query_user
+default: <UNSET>
+example: "sharding_user"
+```
+
+User to be used for connecting to servers to obtain the hash used for md5 authentication by sending the query
+specified in `auth_query_user`. The connection will be established using the database configured in the pool.
+This parameter is inherited by every pool and can be redefined in pool configuration.
+
+### auth_query_password
+```
+path: pools.<pool_name>.auth_query_password
+default: <UNSET>
+example: "sharding_user"
+```
+
+Password to be used for connecting to servers to obtain the hash used for md5 authentication by sending the query
+specified in `auth_query_user`. The connection will be established using the database configured in the pool.
+This parameter is inherited by every pool and can be redefined in pool configuration.
+
+### automatic_sharding_key
+```
+path: pools.<pool_name>.automatic_sharding_key
+default: <UNSET>
+example: "data.id"
+```
+
+Automatically parse this from queries and route queries to the right shard!
+
+### idle_timeout
+```
+path: pools.<pool_name>.idle_timeout
+default: 40000
+```
+
+Idle timeout can be overwritten in the pool
+
+### connect_timeout
+```
+path: pools.<pool_name>.connect_timeout
+default: 3000
+```
+
+Connect timeout can be overwritten in the pool
+
+## `pools.<pool_name>.users.<user_index>` Section
+
+### username
+```
+path: pools.<pool_name>.users.<user_index>.username
+default: "sharding_user"
+```
+
+PostgreSQL username used to authenticate the user and connect to the server
+if `server_username` is not set.
+
+### password
+```
+path: pools.<pool_name>.users.<user_index>.password
+default: "sharding_user"
+```
+
+PostgreSQL password used to authenticate the user and connect to the server
+if `server_password` is not set.
+
+### server_username
+```
+path: pools.<pool_name>.users.<user_index>.server_username
+default: <UNSET>
+example: "another_user"
+```
+
+PostgreSQL username used to connect to the server.
+
+### server_password
+```
+path: pools.<pool_name>.users.<user_index>.server_password
+default: <UNSET>
+example: "another_password"
+```
+
+PostgreSQL password used to connect to the server.
+
+### pool_size
+```
+path: pools.<pool_name>.users.<user_index>.pool_size
+default: 9
+```
+
+Maximum number of server connections that can be established for this user
+The maximum number of connection from a single Pgcat process to any database in the cluster
+is the sum of pool_size across all users.
+
+### statement_timeout
+```
+path: pools.<pool_name>.users.<user_index>.statement_timeout
+default: 0
+```
+
+Maximum query duration. Dangerous, but protects against DBs that died in a non-obvious way.
+0 means it is disabled.
+
+## `pools.<pool_name>.shards.<shard_index>` Section
+
+### servers
+```
+path: pools.<pool_name>.shards.<shard_index>.servers
+default: [["127.0.0.1", 5432, "primary"], ["localhost", 5432, "replica"]]
+```
+
+Array of servers in the shard, each server entry is an array of `[host, port, role]`
+
+### mirrors
+```
+path: pools.<pool_name>.shards.<shard_index>.mirrors
+default: <UNSET>
+example: [["1.2.3.4", 5432, 0], ["1.2.3.4", 5432, 1]]
+```
+
+Array of mirrors for the shard, each mirror entry is an array of `[host, port, index of server in servers array]`
+Traffic hitting the server identified by the index will be sent to the mirror.
+
+### database
+```
+path: pools.<pool_name>.shards.<shard_index>.database
+default: "shard0"
+```
+
+Database name (e.g. "postgres")
+

Cargo.toml

@@ -1,10 +1,9 @@
 [package]
 name = "pgcat"
-version = "0.6.0-alpha1"
+version = "1.0.1"
 edition = "2021"
 
 # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
-
 [dependencies]
 tokio = { version = "1", features = ["full"] }
 bytes = "1"
@@ -14,23 +13,35 @@ async-trait = "0.1"
 rand = "0.8"
 chrono = "0.4"
 sha-1 = "0.10"
-toml = "0.5"
+toml = "0.7"
 serde = "1"
 serde_derive = "1"
 regex = "1"
 num_cpus = "1"
 once_cell = "1"
-sqlparser = "0.23.0"
+sqlparser = "0.33.0"
 log = "0.4"
 arc-swap = "1"
-env_logger = "0.9"
+env_logger = "0.10"
 parking_lot = "0.12.1"
 hmac = "0.12"
 sha2 = "0.10"
-base64 = "0.13"
+base64 = "0.21"
 stringprep = "0.1"
-tokio-rustls = "0.23"
+tokio-rustls = "0.24"
 rustls-pemfile = "1"
 hyper = { version = "0.14", features = ["full"] }
 phf = { version = "0.11.1", features = ["macros"] }
 exitcode = "1.1.2"
+futures = "0.3"
+socket2 = { version = "0.4.7", features = ["all"] }
+nix = "0.26.2"
+atomic_enum = "0.2.0"
+postgres-protocol = "0.6.5"
+fallible-iterator = "0.2"
+pin-project = "1"
+webpki-roots = "0.23"
+rustls = { version = "0.21", features = ["dangerous_configuration"] }
+
+[target.'cfg(not(target_env = "msvc"))'.dependencies]
+jemallocator = "0.5.0"

Dockerfile.ci

@@ -1,8 +1,12 @@
-FROM cimg/rust:1.62.0
+FROM cimg/rust:1.67.1
 RUN sudo apt-get update && \
-	sudo apt-get install -y psmisc postgresql-contrib-12 postgresql-client-12 ruby ruby-dev libpq-dev python3 python3-pip lcov llvm-11 && \
-	sudo apt-get upgrade curl
-RUN cargo install cargo-binutils rustfilt && \
-	rustup component add llvm-tools-preview
-RUN pip3 install psycopg2 && \
-	sudo gem install bundler
+	sudo apt-get install -y \
+		psmisc postgresql-contrib-14 postgresql-client-14 libpq-dev \
+		ruby ruby-dev python3 python3-pip \
+		lcov llvm-11 iproute2 && \
+	sudo apt-get upgrade curl && \
+	cargo install cargo-binutils rustfilt && \
+	rustup component add llvm-tools-preview && \
+ 	pip3 install psycopg2 && sudo gem install bundler && \
+	wget -O /tmp/toxiproxy-2.4.0.deb https://github.com/Shopify/toxiproxy/releases/download/v2.4.0/toxiproxy_2.4.0_linux_$(dpkg --print-architecture).deb && \
+	sudo dpkg -i /tmp/toxiproxy-2.4.0.deb

LICENSE

@@ -1,4 +1,4 @@
-Copyright (c) 2022 Lev Kokotov <lev@levthe.dev>
+Copyright (c) 2023 PgCat Contributors
 
 Permission is hereby granted, free of charge, to any person obtaining
 a copy of this software and associated documentation files (the

README.md

@@ -1,77 +1,98 @@
-![PgCat](./pgcat3.png)
+## PgCat: Nextgen PostgreSQL Pooler
 
-##### PgCat: PostgreSQL at petabyte scale
-
-[![CircleCI](https://circleci.com/gh/levkk/pgcat/tree/main.svg?style=svg)](https://circleci.com/gh/levkk/pgcat/tree/main)
+[![CircleCI](https://circleci.com/gh/postgresml/pgcat/tree/main.svg?style=svg)](https://circleci.com/gh/postgresml/pgcat/tree/main)
 <a href="https://discord.gg/DmyJP3qJ7U" target="_blank">
     <img src="https://img.shields.io/discord/1013868243036930099" alt="Join our Discord!" />
 </a>
 
-PostgreSQL pooler (like PgBouncer) with sharding, load balancing and failover support.
-
-**Beta**: looking for beta testers, see [#35](https://github.com/levkk/pgcat/issues/35).
+PostgreSQL pooler and proxy (like PgBouncer) with support for sharding, load balancing, failover and mirroring.
 
 ## Features
-| **Feature**                    | **Status**                  | **Comments**                                                                                                                                          |
-|--------------------------------|-----------------------------|-------------------------------------------------------------------------------------------------------------------------------------------------------|
-| Transaction pooling            | :white_check_mark:          | Identical to PgBouncer.                                                                                                                               |
-| Session pooling                | :white_check_mark:          | Identical to PgBouncer.                                                                                                                               |
-| `COPY` support                 | :white_check_mark:          | Both `COPY TO` and `COPY FROM` are supported.                                                                                                         |
-| Query cancellation             | :white_check_mark:          | Supported both in transaction and session pooling modes.                                                                                              |
-| Load balancing of read queries | :white_check_mark:          | Using random between replicas. Primary is included when `primary_reads_enabled` is enabled (default).                                            |
-| Sharding                       | :white_check_mark:          | Transactions are sharded using `SET SHARD TO` and `SET SHARDING KEY TO` syntax extensions; see examples below.                                        |
-| Failover                       | :white_check_mark:          | Replicas are tested with a health check. If a health check fails, remaining replicas are attempted; see below for algorithm description and examples. |
-| Statistics                     | :white_check_mark:          | Statistics available in the admin database (`pgcat` and `pgbouncer`) with `SHOW STATS`, `SHOW POOLS` and others.                                      |
-| Live configuration reloading   | :white_check_mark:          | Reload supported settings with a `SIGHUP` to the process, e.g. `kill -s SIGHUP $(pgrep pgcat)` or `RELOAD` query issued to the admin database.        |
-| Client authentication          | :white_check_mark: :wrench: | MD5 password authentication is supported, SCRAM is on the roadmap; one user is used to connect to Postgres with both SCRAM and MD5 supported.         |
-| Admin database                 | :white_check_mark:          | The admin database, similar to PgBouncer's, allows to query for statistics and reload the configuration.                                              |
+
+| **Feature** | **Status** | **Comments** |
+|-------------|------------|--------------|
+| Transaction pooling | **Stable** | Identical to PgBouncer with notable improvements for handling bad clients and abandoned transactions. |
+| Session pooling | **Stable** | Identical to PgBouncer. |
+| Multi-threaded runtime | **Stable** | Using Tokio asynchronous runtime, the pooler takes advantage of multicore machines. |
+| Load balancing of read queries | **Stable** | Queries are automatically load balanced between replicas and the primary. |
+| Failover | **Stable** | Queries are automatically rerouted around broken replicas, validated by regular health checks. |
+| Admin database statistics | **Stable** | Pooler statistics and administration via the `pgbouncer` and `pgcat` databases. |
+| Prometheus statistics | **Stable** | Statistics are reported via a HTTP endpoint for Prometheus. |
+| Client TLS | **Stable** | Clients can connect to the pooler using TLS/SSL. |
+| Client/Server authentication | **Stable** | Clients can connect using MD5 authentication, supported by `libpq` and all Postgres client drivers. PgCat can connect to Postgres using MD5 and SCRAM-SHA-256. |
+| Live configuration reloading | **Stable** | Identical to PgBouncer; all settings can be reloaded dynamically (except `host` and `port`). |
+| Auth passthrough | **Stable** | MD5 password authentication can be configured to use an `auth_query` so no cleartext passwords are needed in the config file.|
+| Sharding using extended SQL syntax | **Experimental** | Clients can dynamically configure the pooler to route queries to specific shards. |
+| Sharding using comments parsing/Regex | **Experimental** | Clients can include shard information (sharding key, shard ID) in the query comments. |
+| Automatic sharding | **Experimental** | PgCat can parse queries, detect sharding keys automatically, and route queries to the correct shard. |
+| Mirroring | **Experimental** | Mirror queries between multiple databases in order to test servers with realistic production traffic. |
+
+
+## Status
+
+PgCat is stable and used in production to serve hundreds of thousands of queries per second.
+
+<table>
+  <tr>
+    <td>
+      <a href="https://tech.instacart.com/adopting-pgcat-a-nextgen-postgres-proxy-3cf284e68c2f">
+        <img src="./images/instacart.webp" height="70" width="auto">
+      </a>
+    </td>
+    <td>
+      <a href="https://postgresml.org/blog/scaling-postgresml-to-one-million-requests-per-second">
+        <img src="./images/postgresml.webp" height="70" width="auto">
+      </a>
+    </td>
+    <td>
+      <a href="https://onesignal.com">
+        <img src="./images/one_signal.webp" height="70" width="auto">
+      </a>
+    </td>
+  </tr>
+  <tr>
+    <td>
+      <a href="https://tech.instacart.com/adopting-pgcat-a-nextgen-postgres-proxy-3cf284e68c2f">
+        Instacart
+      </a>
+    </td>
+    <td>
+      <a href="https://postgresml.org/blog/scaling-postgresml-to-one-million-requests-per-second">
+        PostgresML
+      </a>
+    </td>
+    <td>
+      OneSignal
+    </td>
+  </tr>
+</table>
+
+Some features remain experimental and are being actively developed. They are optional and can be enabled through configuration.
 
 ## Deployment
 
 See `Dockerfile` for example deployment using Docker. The pooler is configured to spawn 4 workers so 4 CPUs are recommended for optimal performance. That setting can be adjusted to spawn as many (or as little) workers as needed.
 
+A Docker image is available from `docker pull ghcr.io/postgresml/pgcat:latest`. See our [Github packages repository](https://github.com/postgresml/pgcat/pkgs/container/pgcat).
+
 For quick local example, use the Docker Compose environment provided:
 
 ```bash
 docker-compose up
 
 # In a new terminal:
-psql -h 127.0.0.1 -p 6432 -c 'SELECT 1'
+PGPASSWORD=postgres psql -h 127.0.0.1 -p 6432 -U postgres -c 'SELECT 1'
 ```
 
 ### Config
 
-| **Name**                     | **Description**                                                                                                                            | **Examples**                     |
-|------------------------------|--------------------------------------------------------------------------------------------------------------------------------------------|----------------------------------|
-| **`general`**                |                                                                                                                                            |                                  |
-| `host`                       | The pooler will run on this host, 0.0.0.0 means accessible from everywhere.                                                                | `0.0.0.0`                        |
-| `port`                       | The pooler will run on this port.                                                                                                          | `6432`                           |
-| `enable_prometheus_exporter` | Enable prometheus exporter which will export metrics in prometheus exposition format.                                                      | `true`                           |
-| `prometheus_exporter_port`   | Port at which prometheus exporter listens on.                                                                                              | `9930`                           |
-| `pool_size`                  | Maximum allowed server connections per pool. Pools are separated for each user/shard/server role. The connections are allocated as needed. | `15`                             |
-| `pool_mode`                  | The pool mode to use, i.e. `session` or `transaction`.                                                                                     | `transaction`                    |
-| `connect_timeout`            | Maximum time to establish a connection to a server (milliseconds). If reached, the server is banned and the next target is attempted.      | `5000`                           |
-| `healthcheck_timeout`        | Maximum time to pass a health check (`SELECT 1`, milliseconds). If reached, the server is banned and the next target is attempted.         | `1000`                           |
-| `shutdown_timeout`           | Maximum time to give clients during shutdown before forcibly killing client connections (ms).                                              | `60000`                          |
-| `healthcheck_delay`          | How long to keep connection available for immediate re-use, without running a healthcheck query on it                                      | `30000`                          |
-| `ban_time`                   | Ban time for a server (seconds). It won't be allowed to serve transactions until the ban expires; failover targets will be used instead.   | `60`                             |
-| `autoreload`                 | Enable auto-reload of config after fixed time-interval.                                                                                    | `false`                          |
-|                              |                                                                                                                                            |                                  |
-| **`user`**                   |                                                                                                                                            |                                  |
-| `name`                       | The user name.                                                                                                                             | `sharding_user`                  |
-| `password`                   | The user password in plaintext.                                                                                                            | `hunter2`                        |
-| `statement_timeout` | Timeout in milliseconds for how long a query takes to execute | `0 (disabled)`
-|                              |                                                                                                                                            |                                  |
-| **`shards`**                 | Shards are numerically numbered starting from 0; the order in the config is preserved by the pooler to route queries accordingly.          | `[shards.0]`                     |
-| `servers`                    | List of servers to connect to and their roles. A server is: `[host, port, role]`, where `role` is either `primary` or `replica`.           | `["127.0.0.1", 5432, "primary"]` |
-| `database`                   | The name of the database to connect to. This is the same on all servers that are part of one shard.                                        |                                  |
-|                              |                                                                                                                                            |                                  |
-| **`query_router`**           |                                                                                                                                            |                                  |
-| `default_role`               | Traffic is routed to this role by default (random), unless the client specifies otherwise. Default is `any`, for any role available.  | `any`, `primary`, `replica`      |
-| `query_parser_enabled`       | Enable the query parser which will inspect incoming queries and route them to a primary or replicas.                                       | `false`                          |
-| `primary_reads_enabled`      | Enable this to allow read queries on the primary; otherwise read queries are routed to the replicas.                                       | `true`                           |
+See **[Configuration](https://github.com/levkk/pgcat/blob/main/CONFIG.md)**.
 
-## Local development
+## Contributing
+
+The project is being actively developed and looking for additional contributors and production deployments.
+
+### Local development
 
 1. Install Rust (latest stable will work great).
 2. `cargo build --release` (to get better benchmarks).
@@ -81,7 +102,7 @@ psql -h 127.0.0.1 -p 6432 -c 'SELECT 1'
 
 ### Tests
 
-Quickest way to test your changes is to use pgbench:
+When making substantial modifications to the protocol implementation, make sure to test them with pgbench:
 
 ```
 pgbench -i -h 127.0.0.1 -p 6432 && \
@@ -91,25 +112,26 @@ pgbench -t 1000 -p 6432 -h 127.0.0.1 --protocol extended
 
 See [sharding README](./tests/sharding/README.md) for sharding logic testing.
 
-Run `cargo test` to run Rust tests.
+Additionally, all features are tested with Ruby, Python, and Rust unit and integration tests.
+
+Run `cargo test` to run Rust unit tests.
+
+Run the following commands to run Ruby and Python integration tests:
 
-Run the following commands to run Integration tests locally.
 ```
 cd tests/docker/
 docker compose up --exit-code-from main # This will also produce coverage report under ./cov/
 ```
 
-| **Feature**           | **Tested in CI**   | **Tested manually** | **Comments**                                                                                                             |
-|-----------------------|--------------------|---------------------|--------------------------------------------------------------------------------------------------------------------------|
-| Transaction pooling   | :white_check_mark: | :white_check_mark:  | Used by default for all tests.                                                                                           |
-| Session pooling       | :white_check_mark: | :white_check_mark:  | Tested by running pgbench with `--protocol prepared` which only works in session mode.                                   |
-| `COPY`                | :white_check_mark: | :white_check_mark:  | `pgbench -i` uses `COPY`. `COPY FROM` is tested as well.                                                                 |
-| Query cancellation    | :white_check_mark: | :white_check_mark:  | `psql -c 'SELECT pg_sleep(1000);'` and press `Ctrl-C`.                                                                   |
-| Load balancing        | :white_check_mark: | :white_check_mark:  | We could test this by emitting statistics for each replica and compare them.                                             |
-| Failover              | :white_check_mark: | :white_check_mark:  | Misconfigure a replica in `pgcat.toml` and watch it forward queries to spares. CI testing is using Toxiproxy.            |
-| Sharding              | :white_check_mark: | :white_check_mark:  | See `tests/sharding` and `tests/ruby` for an Rails/ActiveRecord example.                                                 |
-| Statistics            | :white_check_mark: | :white_check_mark:  | Query the admin database with `psql -h 127.0.0.1 -p 6432 -d pgbouncer -c 'SHOW STATS'`.                                  |
-| Live config reloading | :white_check_mark: | :white_check_mark:  | Run `kill -s SIGHUP $(pgrep pgcat)` and watch the config reload.                                                         |
+### Docker-based local development
+
+You can open a Docker development environment where you can debug tests easier. Run the following command to spin it up:
+
+```
+./dev/script/console
+```
+
+This will open a terminal in an environment similar to that used in tests. In there, you can compile the pooler, run tests, do some debugging with the test environment, etc. Objects compiled inside the container (and bundled gems) will be placed in `dev/cache` so they don't interfere with what you have on your machine.
 
 ## Usage
 
@@ -124,11 +146,9 @@ In transaction mode, a client talks to one server for the duration of a single t
 This mode is enabled by default.
 
 ### Load balancing of read queries
-All queries are load balanced against the configured servers using the random algorithm. The most straight forward configuration example would be to put this pooler in front of several replicas and let it load balance all queries.
+All queries are load balanced against the configured servers using either the random or least open connections algorithms. The most straightforward configuration example would be to put this pooler in front of several replicas and let it load balance all queries.
 
-If the configuration includes a primary and replicas, the queries can be separated with the built-in query parser. The query parser will interpret the query and route all `SELECT` queries to a replica, while all other queries including explicit transactions will be routed to the primary.
-
-The query parser is disabled by default.
+If the configuration includes a primary and replicas, the queries can be separated with the built-in query parser. The query parser, implemented with the `sqlparser` crate, will interpret the query and route all `SELECT` queries to a replica, while all other queries including explicit transactions will be routed to the primary.
 
 #### Query parser
 The query parser will do its best to determine where the query should go, but sometimes that's not possible. In that case, the client can select which server it wants using this custom SQL syntax:
@@ -155,42 +175,18 @@ The setting will persist until it's changed again or the client disconnects.
 By default, all queries are routed to the first available server; `default_role` setting controls this behavior.
 
 ### Failover
-All servers are checked with a `SELECT 1` query before being given to a client. If the server is not reachable, it will be banned and cannot serve any more transactions for the duration of the ban. The queries are routed to the remaining servers. If all servers become banned, the ban list is cleared: this is a safety precaution against false positives. The primary can never be banned.
+All servers are checked with a `;` (very fast) query before being given to a client. Additionally, the server health is monitored with every client query that it processes. If the server is not reachable, it will be banned and cannot serve any more transactions for the duration of the ban. The queries are routed to the remaining servers. If all servers become banned, the ban list is cleared: this is a safety precaution against false positives. The primary can never be banned.
 
 The ban time can be changed with `ban_time`. The default is 60 seconds.
 
-Failover behavior can get pretty interesting (read complex) when multiple configurations and factors are involved. The table below will try to explain what PgCat does in each scenario:
-
-| **Query**                 | **`SET SERVER ROLE TO`** | **`query_parser_enabled`** | **`primary_reads_enabled`** | **Target state** | **Outcome**                                                                                                                                                          |
-|---------------------------|--------------------------|----------------------------|-----------------------------|------------------|----------------------------------------------------------------------------------------------------------------------------------------------------------------------|
-| Read query, i.e. `SELECT` | unset (any)              | false                      | false                       | up               | Query is routed to the first instance in the random loop.                                                                                                       |
-| Read query                | unset (any)              | true                       | false                       | up               | Query is routed to the first replica instance in the random loop.                                                                                               |
-| Read query                | unset (any)              | true                       | true                        | up               | Query is routed to the first instance in the random loop.                                                                                                       |
-| Read query                | replica                  | false                      | false                       | up               | Query is routed to the first replica instance in the random loop.                                                                                               |
-| Read query                | primary                  | false                      | false                       | up               | Query is routed to the primary.                                                                                                                                      |
-| Read query                | unset (any)              | false                      | false                       | down             | First instance is banned for reads. Next target in the random loop is attempted.                                                                                |
-| Read query                | unset (any)              | true                       | false                       | down             | First replica instance is banned. Next replica instance is attempted in the random loop.                                                                        |
-| Read query                | unset (any)              | true                       | true                        | down             | First instance (even if primary) is banned for reads. Next instance is attempted in the random loop.                                                            |
-| Read query                | replica                  | false                      | false                       | down             | First replica instance is banned. Next replica instance is attempted in the random loop.                                                                        |
-| Read query                | primary                  | false                      | false                       | down             | The query is attempted against the primary and fails. The client receives an error.                                                                                  |
-|                           |                          |                            |                             |                  |                                                                                                                                                                      |
-| Write query e.g. `INSERT` | unset (any)              | false                      | false                       | up               | The query is attempted against the first available instance in the random loop. If the instance is a replica, the query fails and the client receives an error. |
-| Write query               | unset (any)              | true                       | false                       | up               | The query is routed to the primary.                                                                                                                                  |
-| Write query               | unset (any)              | true                       | true                        | up               | The query is routed to the primary.                                                                                                                                  |
-| Write query               | primary                  | false                      | false                       | up               | The query is routed to the primary.                                                                                                                                  |
-| Write query               | replica                  | false                      | false                       | up               | The query is routed to the replica and fails. The client receives an error.                                                                                          |
-| Write query               | unset (any)              | true                       | false                       | down             | The query is routed to the primary and fails. The client receives an error.                                                                                          |
-| Write query               | unset (any)              | true                       | true                        | down             | The query is routed to the primary and fails. The client receives an error.                                                                                          |
-| Write query               | primary                  | false                      | false                       | down             | The query is routed to the primary and fails. The client receives an error.                                                                                          |
-|                           |                          |                            |                             |                  |                                                                                                                                                                      |
-
 ### Sharding
 We use the `PARTITION BY HASH` hashing function, the same as used by Postgres for declarative partitioning. This allows to shard the database using Postgres partitions and place the partitions on different servers (shards). Both read and write queries can be routed to the shards using this pooler.
 
+#### Extended syntax
 To route queries to a particular shard, we use this custom SQL syntax:
 
 ```sql
--- To talk to a shard explicitely
+-- To talk to a shard explicitly
 SET SHARD TO '1';
 
 -- To let the pooler choose based on a value
@@ -201,7 +197,8 @@ The active shard will last until it's changed again or the client disconnects. B
 
 For hash function implementation, see `src/sharding.rs` and `tests/sharding/partition_hash_test_setup.sql`.
 
-#### ActiveRecord/Rails
+
+##### ActiveRecord/Rails
 
 ```ruby
 class User < ActiveRecord::Base
@@ -229,7 +226,7 @@ User.connection.execute "SET SERVER ROLE TO 'auto'"
 User.find_by_email("test@example.com")
 ```
 
-#### Raw SQL
+##### Raw SQL
 
 ```sql
 -- Grab a bunch of users from shard 1
@@ -249,268 +246,45 @@ SET SERVER ROLE TO 'auto'; -- let the query router figure out where the query sh
 SELECT * FROM users WHERE email = 'test@example.com'; -- shard setting lasts until set again; we are reading from the primary
 ```
 
+#### With comments
+Issuing queries to the pooler can cause additional latency. To reduce its impact, it's possible to include sharding information inside SQL comments sent via the query. This is reasonably easy to implement with ORMs like [ActiveRecord](https://api.rubyonrails.org/classes/ActiveRecord/QueryMethods.html#method-i-annotate) and [SQLAlchemy](https://docs.sqlalchemy.org/en/20/core/events.html#sql-execution-and-connection-events).
+
+```
+/* shard_id: 5 */ SELECT * FROM foo WHERE id = 1234;
+
+/* sharding_key: 1234 */ SELECT * FROM foo WHERE id = 1234;
+```
+
+#### Automatic query parsing
+PgCat can use the `sqlparser` crate to parse SQL queries and extract the sharding key. This is configurable with the `automatic_sharding_key` setting. This feature is still experimental, but it's the ideal implementation for sharding, requiring no client modifications.
+
 ### Statistics reporting
 
-The stats are very similar to what Pgbouncer reports and the names are kept to be comparable. They are accessible by querying the admin database `pgcat`, and `pgbouncer` for compatibility.
+The stats are very similar to what PgBouncer reports and the names are kept to be comparable. They are accessible by querying the admin database `pgcat`, and `pgbouncer` for compatibility.
 
 ```
 psql -h 127.0.0.1 -p 6432 -d pgbouncer -c 'SHOW DATABASES'
 ```
 
+Additionally, Prometheus statistics are available at `/metrics` via HTTP.
+
 ### Live configuration reloading
 
-The config can be reloaded by sending a `kill -s SIGHUP` to the process or by querying `RELOAD` to the admin database. Not all settings are currently supported by live reload:
+The config can be reloaded by sending a `kill -s SIGHUP` to the process or by querying `RELOAD` to the admin database. All settings except the `host` and `port` can be reloaded without restarting the pooler, including sharding and replicas configurations.
 
-| **Config**              | **Requires restart** |
-|-------------------------|----------------------|
-| `host`                  | yes                  |
-| `port`                  | yes                  |
-| `pool_mode`             | no                   |
-| `connect_timeout`       | yes                  |
-| `healthcheck_timeout`   | no                   |
-| `shutdown_timeout`      | no                   |
-| `healthcheck_delay`     | no                   |
-| `ban_time`              | no                   |
-| `user`                  | yes                  |
-| `shards`                | yes                  |
-| `default_role`          | no                   |
-| `primary_reads_enabled` | no                   |
-| `query_parser_enabled`  | no                   |
+### Mirroring
 
+Mirroring allows to route queries to multiple databases at the same time. This is useful for prewarning replicas before placing them into the active configuration, or for testing different versions of Postgres with live traffic.
 
-## Benchmarks
+## License
 
-You can setup PgBench locally through PgCat:
+PgCat is free and open source, released under the MIT license.
 
-```
-pgbench -h 127.0.0.1 -p 6432 -i
-```
+## Contributors
 
-Coincidenly, this uses `COPY` so you can test if that works. Additionally, we'll be running the following PgBench configurations:
+Many thanks to our amazing contributors!
 
-1. 16 clients, 2 threads
-2. 32 clients, 2 threads
-3. 64 clients, 2 threads
-4. 128 clients, 2 threads
+<a href = "https://github.com/postgresml/pgcat/graphs/contributors">
+  <img src = "https://contrib.rocks/image?repo=postgresml/pgcat"/>
+</a>
 
-All queries will be `SELECT` only (`-S`) just so disks don't get in the way, since the dataset will be effectively all in RAM.
-
-My setup:
-
-- 8 cores, 16 hyperthreaded (AMD Ryzen 5800X)
-- 32GB RAM (doesn't matter for this benchmark, except to prove that Postgres will fit the whole dataset into RAM)
-
-### PgBouncer
-
-#### Config
-
-```ini
-[databases]
-shard0 = host=localhost port=5432 user=sharding_user password=sharding_user
-
-[pgbouncer]
-pool_mode = transaction
-max_client_conn = 1000
-```
-
-Everything else stays default.
-
-#### Runs
-
-
-```
-$ pgbench -t 1000 -c 16 -j 2 -p 6432 -h 127.0.0.1 -S --protocol extended shard0
-
-starting vacuum...end.
-transaction type: <builtin: select only>
-scaling factor: 1
-query mode: extended
-number of clients: 16
-number of threads: 2
-number of transactions per client: 1000
-number of transactions actually processed: 16000/16000
-latency average = 0.155 ms
-tps = 103417.377469 (including connections establishing)
-tps = 103510.639935 (excluding connections establishing)
-
-
-$ pgbench -t 1000 -c 32 -j 2 -p 6432 -h 127.0.0.1 -S --protocol extended shard0
-
-starting vacuum...end.
-transaction type: <builtin: select only>
-scaling factor: 1
-query mode: extended
-number of clients: 32
-number of threads: 2
-number of transactions per client: 1000
-number of transactions actually processed: 32000/32000
-latency average = 0.290 ms
-tps = 110325.939785 (including connections establishing)
-tps = 110386.513435 (excluding connections establishing)
-
-
-$ pgbench -t 1000 -c 64 -j 2 -p 6432 -h 127.0.0.1 -S --protocol extended shard0
-
-starting vacuum...end.
-transaction type: <builtin: select only>
-scaling factor: 1
-query mode: extended
-number of clients: 64
-number of threads: 2
-number of transactions per client: 1000
-number of transactions actually processed: 64000/64000
-latency average = 0.692 ms
-tps = 92470.427412 (including connections establishing)
-tps = 92618.389350 (excluding connections establishing)
-
-$ pgbench -t 1000 -c 128 -j 2 -p 6432 -h 127.0.0.1 -S --protocol extended shard0
-
-starting vacuum...end.
-transaction type: <builtin: select only>
-scaling factor: 1
-query mode: extended
-number of clients: 128
-number of threads: 2
-number of transactions per client: 1000
-number of transactions actually processed: 128000/128000
-latency average = 1.406 ms
-tps = 91013.429985 (including connections establishing)
-tps = 91067.583928 (excluding connections establishing)
-```
-
-### PgCat
-
-#### Config
-
-The only thing that matters here is the number of workers in the Tokio pool. Make sure to set it to < than the number of your CPU cores.
-Also account for hyper-threading, so if you have that, take the number you got above and divide it by two, that way only "real" cores serving
-requests.
-
-My setup is 16 threads, 8 cores (`htop` shows as 16 CPUs), so I set the `max_workers` in Tokio to 4. Too many, and it starts conflicting with PgBench
-which is also running on the same system.
-
-#### Runs
-
-
-```
-$ pgbench -t 1000 -c 16 -j 2 -p 6432 -h 127.0.0.1 -S --protocol extended
-starting vacuum...end.
-transaction type: <builtin: select only>
-scaling factor: 1
-query mode: extended
-number of clients: 16
-number of threads: 2
-number of transactions per client: 1000
-number of transactions actually processed: 16000/16000
-latency average = 0.164 ms
-tps = 97705.088232 (including connections establishing)
-tps = 97872.216045 (excluding connections establishing)
-
-
-$ pgbench -t 1000 -c 32 -j 2 -p 6432 -h 127.0.0.1 -S --protocol extended
-
-starting vacuum...end.
-transaction type: <builtin: select only>
-scaling factor: 1
-query mode: extended
-number of clients: 32
-number of threads: 2
-number of transactions per client: 1000
-number of transactions actually processed: 32000/32000
-latency average = 0.288 ms
-tps = 111300.488119 (including connections establishing)
-tps = 111413.107800 (excluding connections establishing)
-
-
-$ pgbench -t 1000 -c 64 -j 2 -p 6432 -h 127.0.0.1 -S --protocol extended
-
-starting vacuum...end.
-transaction type: <builtin: select only>
-scaling factor: 1
-query mode: extended
-number of clients: 64
-number of threads: 2
-number of transactions per client: 1000
-number of transactions actually processed: 64000/64000
-latency average = 0.556 ms
-tps = 115190.496139 (including connections establishing)
-tps = 115247.521295 (excluding connections establishing)
-
-$ pgbench -t 1000 -c 128 -j 2 -p 6432 -h 127.0.0.1 -S --protocol extended
-
-starting vacuum...end.
-transaction type: <builtin: select only>
-scaling factor: 1
-query mode: extended
-number of clients: 128
-number of threads: 2
-number of transactions per client: 1000
-number of transactions actually processed: 128000/128000
-latency average = 1.135 ms
-tps = 112770.562239 (including connections establishing)
-tps = 112796.502381 (excluding connections establishing)
-```
-
-### Direct Postgres
-
-Always good to have a base line.
-
-#### Runs
-
-```
-$ pgbench -t 1000 -c 16 -j 2 -p 5432 -h 127.0.0.1 -S --protocol extended shard0
-Password:
-starting vacuum...end.
-transaction type: <builtin: select only>
-scaling factor: 1
-query mode: extended
-number of clients: 16
-number of threads: 2
-number of transactions per client: 1000
-number of transactions actually processed: 16000/16000
-latency average = 0.115 ms
-tps = 139443.955722 (including connections establishing)
-tps = 142314.859075 (excluding connections establishing)
-
-$ pgbench -t 1000 -c 32 -j 2 -p 5432 -h 127.0.0.1 -S --protocol extended shard0
-Password:
-starting vacuum...end.
-transaction type: <builtin: select only>
-scaling factor: 1
-query mode: extended
-number of clients: 32
-number of threads: 2
-number of transactions per client: 1000
-number of transactions actually processed: 32000/32000
-latency average = 0.212 ms
-tps = 150644.840891 (including connections establishing)
-tps = 152218.499430 (excluding connections establishing)
-
-$ pgbench -t 1000 -c 64 -j 2 -p 5432 -h 127.0.0.1 -S --protocol extended shard0
-Password:
-starting vacuum...end.
-transaction type: <builtin: select only>
-scaling factor: 1
-query mode: extended
-number of clients: 64
-number of threads: 2
-number of transactions per client: 1000
-number of transactions actually processed: 64000/64000
-latency average = 0.420 ms
-tps = 152517.663404 (including connections establishing)
-tps = 153319.188482 (excluding connections establishing)
-
-$ pgbench -t 1000 -c 128 -j 2 -p 5432 -h 127.0.0.1 -S --protocol extended shard0
-Password:
-starting vacuum...end.
-transaction type: <builtin: select only>
-scaling factor: 1
-query mode: extended
-number of clients: 128
-number of threads: 2
-number of transactions per client: 1000
-number of transactions actually processed: 128000/128000
-latency average = 0.854 ms
-tps = 149818.594087 (including connections establishing)
-tps = 150200.603049 (excluding connections establishing)
-```

cov-style.css

@@ -0,0 +1,158 @@
+/*
+ * Copyright 2021 Collabora, Ltd.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial
+ * portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT.  IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+body {
+	background-color: #f2f2f2;
+	font-family: -apple-system, BlinkMacSystemFont, "Segoe UI", Roboto,
+		"Noto Sans", Ubuntu, Cantarell, "Helvetica Neue", sans-serif,
+		"Apple Color Emoji", "Segoe UI Emoji", "Segoe UI Symbol",
+		"Noto Color Emoji";
+}
+
+.sourceHeading, .source, .coverFn,
+.testName, .testPer, .testNum,
+.coverLegendCovLo, .headerCovTableEntryLo, .coverPerLo, .coverNumLo,
+.coverLegendCovMed, .headerCovTableEntryMed, .coverPerMed, .coverNumMed,
+.coverLegendCovHi, .headerCovTableEntryHi, .coverPerHi, .coverNumHi,
+.coverFile {
+	font-family: "Menlo", "DejaVu Sans Mono", "Liberation Mono",
+		"Consolas", "Ubuntu Mono", "Courier New", "andale mono",
+		"lucida console", monospace;
+}
+
+pre {
+	font-size: 0.7875rem;
+}
+
+.headerCovTableEntry, .testPer, .testNum, .testName,
+.coverLegendCovLo, .headerCovTableEntryLo, .coverPerLo, .coverNumLo,
+.coverLegendCovMed, .headerCovTableEntryMed, .coverPerMed, .coverNumMed,
+.coverLegendCovHi, .headerCovTableEntryHi, .coverPerHi, .coverNumHi {
+	text-align: right;
+	white-space: nowrap;
+}
+
+.coverPerLo, .coverPerMed, .coverPerHi, .testPer {
+/*	font-weight: bold;*/
+}
+
+.coverNumLo, .coverNumMed, .coverNumHi, .testNum {
+	font-style: italic;
+	font-size: 90%;
+	padding-left: 1em;
+}
+
+.title {
+	font-size: 200%;
+}
+
+.tableHead {
+	text-align: center;
+	font-weight: bold;
+	background-color: #bfbfbf;
+}
+
+.coverFile, .coverBar, .coverFn {
+	background-color: #d9d9d9;
+}
+
+.headerCovTableHead {
+	font-weight: bold;
+	text-align: right;
+}
+
+.headerCovTableEntry {
+	background-color: #d9d9d9;
+}
+
+.coverFnLo,
+.coverLegendCovLo, .headerCovTableEntryLo, .coverPerLo, .coverNumLo {
+	background-color: #f2dada;
+}
+
+.coverFnHi,
+.coverLegendCovMed, .headerCovTableEntryMed, .coverPerMed, .coverNumMed {
+	background-color: #add9ad;
+}
+
+.coverLegendCovHi, .headerCovTableEntryHi, .coverPerHi, .coverNumHi {
+	background-color: #59b359;
+}
+
+.coverBarOutline {
+	border-style: solid;
+	border-width: 1px;
+	border-color: black;
+	padding: 0px;
+}
+
+.coverFnLo, .coverFnHi {
+	text-align: right;
+}
+
+.lineNum {
+	background-color: #d9d9d9;
+}
+
+.coverLegendCov, .lineCov, .branchCov {
+	background-image: url('data:image/png;base64,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');
+	background-repeat: repeat-y;
+	background-position: left top;
+	background-color: #c6ffb8;
+}
+
+.coverLegendNoCov, .lineNoCov, .branchNoCov, .branchNoExec {
+	background-image: url('data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAgAAAABCAIAAABsYngUAAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAB3RJTUUH5QMUCiMidNgp2gAAABl0RVh0Q29tbWVudABDcmVhdGVkIHdpdGggR0lNUFeBDhcAAAAPSURBVAjXY/wZIcWADQAAIa4BbZaExr0AAAAASUVORK5CYII=');
+	background-repeat: repeat-y;
+	background-position: left top;
+	background-color: #ffcfbb;
+}
+
+.coverLegendCov, .coverLegendNoCov {
+	padding: 0em 1em 0em 1em;
+}
+
+.headerItem, .headerValue, .headerValueLeg {
+	white-space: nowrap;
+}
+
+.headerItem {
+	text-align: right;
+	font-weight: bold;
+}
+
+.ruler {
+	background-color: #d9d9d9;
+}
+
+.detail {
+	font-size: 80%;
+}
+
+.versionInfo {
+	font-size: 80%;
+	text-align: right;
+}
+

dev/Dockerfile

@@ -0,0 +1,33 @@
+FROM rust:bullseye
+
+# Dependencies
+RUN apt-get update -y \
+    && apt-get install -y \
+    llvm-11 psmisc postgresql-contrib postgresql-client \
+    ruby ruby-dev libpq-dev python3 python3-pip lcov curl sudo iproute2 \
+    strace ngrep iproute2 dnsutils lsof net-tools telnet
+
+# Rust
+RUN cargo install cargo-binutils rustfilt
+RUN rustup component add llvm-tools-preview
+
+# Ruby
+RUN sudo gem install bundler
+
+# Toxyproxy
+RUN wget -O toxiproxy-2.4.0.deb https://github.com/Shopify/toxiproxy/releases/download/v2.4.0/toxiproxy_2.4.0_linux_$(dpkg --print-architecture).deb && \
+    sudo dpkg -i toxiproxy-2.4.0.deb
+
+# Config
+ENV APP_ROOT=/app
+ARG APP_USER=pgcat
+COPY dev_bashrc /etc/bash.bashrc
+
+RUN useradd -m -o -u 999 ${APP_USER} || exit 0 && mkdir ${APP_ROOT} && chown ${APP_USER} ${APP_ROOT}
+RUN adduser ${APP_USER} sudo \
+    && echo "${APP_USER} ALL=NOPASSWD: ALL" > /etc/sudoers.d/${APP_USER} \
+    && chmod ugo+s /usr/sbin/usermod /usr/sbin/groupmod
+ENV HOME=${APP_ROOT}
+WORKDIR ${APP_ROOT}
+
+ENTRYPOINT ["/bin/bash"]

dev/dev_bashrc

@@ -0,0 +1,120 @@
+# ~/.bashrc: executed by bash(1) for non-login shells.
+# see /usr/share/doc/bash/examples/startup-files (in the package bash-doc)
+# for examples
+
+# FIX USER NEEDED SO WE CAN SHARE UID BETWEEN HOST AND DEV ENV
+usermod -o -u $(id -u) pgcat 
+groupmod -o -g $(id -g) pgcat 
+
+# We fix the setuid in those commands as we now have sudo
+sudo chmod ugo-s /usr/sbin/usermod /usr/sbin/groupmod 
+
+# Environment customization
+export DEV_ROOT="${APP_ROOT}/dev"
+export HISTFILE="${DEV_ROOT}/.bash_history"
+export CARGO_TARGET_DIR="${DEV_ROOT}/cache/target"
+export CARGO_HOME="${DEV_ROOT}/cache/target/.cargo"
+export BUNDLE_PATH="${DEV_ROOT}/cache/bundle"
+
+# Regular bashrc
+# If not running interactively, don't do anything
+case $- in
+  *i*) ;;
+  *) return;;
+esac
+
+# don't put duplicate lines or lines starting with space in the history.
+# See bash(1) for more options
+HISTCONTROL=ignoreboth
+
+# append to the history file, don't overwrite it
+shopt -s histappend
+
+# for setting history length see HISTSIZE and HISTFILESIZE in bash(1)
+HISTSIZE=1000
+HISTFILESIZE=2000
+
+# check the window size after each command and, if necessary,
+# update the values of LINES and COLUMNS.
+shopt -s checkwinsize
+
+# If set, the pattern "**" used in a pathname expansion context will
+# match all files and zero or more directories and subdirectories.
+#shopt -s globstar
+
+# make less more friendly for non-text input files, see lesspipe(1)
+[ -x /usr/bin/lesspipe ] && eval "$(SHELL=/bin/sh lesspipe)"
+
+# set variable identifying the chroot you work in (used in the prompt below)
+if [ -z "${debian_chroot:-}" ] && [ -r /etc/debian_chroot ]; then
+  debian_chroot=$(cat /etc/debian_chroot)
+fi
+
+# set a fancy prompt (non-color, unless we know we "want" color)
+case "$TERM" in
+  xterm-color|*-256color) color_prompt=yes;;
+esac
+
+# uncomment for a colored prompt, if the terminal has the capability; turned
+# off by default to not distract the user: the focus in a terminal window
+# should be on the output of commands, not on the prompt
+#force_color_prompt=yes
+
+if [ -n "$force_color_prompt" ]; then
+  if [ -x /usr/bin/tput ] && tput setaf 1 >&/dev/null; then
+    # We have color support; assume it's compliant with Ecma-48
+    # (ISO/IEC-6429). (Lack of such support is extremely rare, and such
+    # a case would tend to support setf rather than setaf.)
+    color_prompt=yes
+  else
+    color_prompt=
+  fi
+fi
+
+PS1='\[\e]0;pgcat@dev-container\h: \w\a\]${debian_chroot:+($debian_chroot)}\[\033[01;32m\]pgcat\[\033[00m\]@\[\033[01;32m\]dev-container\[\033[00m\]:\[\033[01;34m\]\w\[\033[00m\]\[\033[01;31m\]$(git branch &>/dev/null; if [ $? -eq 0 ]; then echo " ($(git branch | grep ^* |sed s/\*\ //))"; fi)\[\033[00m\]\$ '
+
+unset color_prompt force_color_prompt
+
+# enable color support of ls and also add handy aliases
+if [ -x /usr/bin/dircolors ]; then
+  test -r ~/.dircolors && eval "$(dircolors -b ~/.dircolors)" || eval "$(dircolors -b)"
+  alias ls='ls --color=auto'
+  #alias dir='dir --color=auto'
+  #alias vdir='vdir --color=auto'
+
+  alias grep='grep --color=auto'
+  alias fgrep='fgrep --color=auto'
+  alias egrep='egrep --color=auto'
+fi
+
+# colored GCC warnings and errors
+#export GCC_COLORS='error=01;31:warning=01;35:note=01;36:caret=01;32:locus=01:quote=01'
+
+# some more ls aliases
+alias ll='ls -alF'
+alias la='ls -A'
+alias l='ls -CF'
+
+# Add an "alert" alias for long running commands.  Use like so:
+#   sleep 10; alert
+alias alert='notify-send --urgency=low -i "$([ $? = 0 ] && echo terminal || echo error)" "$(history|tail -n1|sed -e '\''s/^\s*[0-9]\+\s*//;s/[;&|]\s*alert$//'\'')"'
+
+# Alias definitions.
+# You may want to put all your additions into a separate file like
+# ~/.bash_aliases, instead of adding them here directly.
+# See /usr/share/doc/bash-doc/examples in the bash-doc package.
+
+if [ -f ~/.bash_aliases ]; then
+  . ~/.bash_aliases
+fi
+
+# enable programmable completion features (you don't need to enable
+# this, if it's already enabled in /etc/bash.bashrc and /etc/profile
+# sources /etc/bash.bashrc).
+if ! shopt -oq posix; then
+  if [ -f /usr/share/bash-completion/bash_completion ]; then
+    . /usr/share/bash-completion/bash_completion
+  elif [ -f /etc/bash_completion ]; then
+    . /etc/bash_completion
+  fi
+fi

dev/docker-compose.yaml

@@ -0,0 +1,94 @@
+version: "3"
+
+x-common-definition-pg:
+  &common-definition-pg
+  image: postgres:14
+  network_mode: "service:main"
+  healthcheck:
+    test: [ "CMD-SHELL", "pg_isready -U postgres -d postgres" ]
+    interval: 5s
+    timeout: 5s
+    retries: 5
+  volumes:
+    - type: bind
+      source: ../tests/sharding/query_routing_setup.sql
+      target: /docker-entrypoint-initdb.d/query_routing_setup.sql
+    - type: bind
+      source: ../tests/sharding/partition_hash_test_setup.sql
+      target: /docker-entrypoint-initdb.d/partition_hash_test_setup.sql
+
+x-common-env-pg:
+  &common-env-pg
+  POSTGRES_USER: postgres
+  POSTGRES_DB: postgres
+  POSTGRES_PASSWORD: postgres
+
+services:
+  main:
+    image: kubernetes/pause
+    ports:
+      - 6432
+
+  pg1:
+    <<: *common-definition-pg
+    environment:
+      <<: *common-env-pg
+      POSTGRES_INITDB_ARGS: --auth-local=md5 --auth-host=md5 --auth=md5
+      PGPORT: 5432
+    command: ["postgres", "-p", "5432", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-c", "pg_stat_statements.max=100000"]
+
+  pg2:
+    <<: *common-definition-pg
+    environment:
+      <<: *common-env-pg
+      POSTGRES_INITDB_ARGS: --auth-local=scram-sha-256 --auth-host=scram-sha-256 --auth=scram-sha-256
+      PGPORT: 7432
+    command: ["postgres", "-p", "7432", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-c", "pg_stat_statements.max=100000"]
+  pg3:
+    <<: *common-definition-pg
+    environment:
+      <<: *common-env-pg
+      POSTGRES_INITDB_ARGS: --auth-local=scram-sha-256 --auth-host=scram-sha-256 --auth=scram-sha-256
+      PGPORT: 8432
+    command: ["postgres", "-p", "8432", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-c", "pg_stat_statements.max=100000"]
+  pg4:
+    <<: *common-definition-pg
+    environment:
+      <<: *common-env-pg
+      POSTGRES_INITDB_ARGS: --auth-local=scram-sha-256 --auth-host=scram-sha-256 --auth=scram-sha-256
+      PGPORT: 9432
+    command: ["postgres", "-p", "9432", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-c", "pg_stat_statements.max=100000"]
+  pg5:
+    <<: *common-definition-pg
+    environment:
+      <<: *common-env-pg
+      POSTGRES_INITDB_ARGS: --auth-local=md5 --auth-host=md5 --auth=md5
+      PGPORT: 10432
+    command: ["postgres", "-p", "5432", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-c", "pg_stat_statements.max=100000"]
+
+  toxiproxy:
+    build: .
+    network_mode: "service:main"
+    container_name: toxiproxy
+    environment:
+      LOG_LEVEL: info
+    entrypoint: toxiproxy-server
+    depends_on:
+      - pg1
+      - pg2
+      - pg3
+      - pg4
+      - pg5
+
+  pgcat-shell:
+    stdin_open: true
+    user: "${HOST_UID}:${HOST_GID}"
+    build: .
+    network_mode: "service:main"
+    depends_on:
+      - toxiproxy
+    volumes:
+      - ../:/app/
+    entrypoint:
+      - /bin/bash
+      - -i

dev/script/console

@@ -0,0 +1,12 @@
+#!/bin/bash
+
+DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" >/dev/null 2>&1 && pwd )"
+export HOST_UID="$(id -u)"
+export HOST_GID="$(id -g)"
+
+if [[ "${1}" == "down" ]]; then
+	docker-compose -f "${DIR}/../docker-compose.yaml" down
+	exit 0
+else
+	docker-compose -f "${DIR}/../docker-compose.yaml" run --rm pgcat-shell
+fi

examples/docker/pgcat.toml

@@ -32,8 +32,11 @@ shutdown_timeout = 60000
 # For how long to ban a server if it fails a health check (seconds).
 ban_time = 60 # seconds
 
-# Reload config automatically if it changes.
-autoreload = false
+# If we should log client connections
+log_client_connections = false
+
+# If we should log client disconnections
+log_client_disconnections = false
 
 # TLS
 # tls_certificate = "server.cert"
@@ -48,7 +51,7 @@ admin_password = "postgres"
 # configs are structured as pool.<pool_name>
 # the pool_name is what clients use as database name when connecting
 # For the example below a client can connect using "postgres://sharding_user:sharding_user@pgcat_host:pgcat_port/sharded"
-[pools.sharded]
+[pools.postgres]
 # Pool mode (see PgBouncer docs for more).
 # session: one server connection per connected client
 # transaction: one server connection per client transaction
@@ -70,7 +73,7 @@ query_parser_enabled = true
 
 # If the query parser is enabled and this setting is enabled, the primary will be part of the pool of databases used for
 # load balancing of read queries. Otherwise, the primary will only be used for write
-# queries. The primary can always be explicitely selected with our custom protocol.
+# queries. The primary can always be explicitly selected with our custom protocol.
 primary_reads_enabled = true
 
 # So what if you wanted to implement a different hashing function,
@@ -84,7 +87,7 @@ primary_reads_enabled = true
 sharding_function = "pg_bigint_hash"
 
 # Credentials for users that may connect to this cluster
-[pools.sharded.users.0]
+[pools.postgres.users.0]
 username = "postgres"
 password = "postgres"
 # Maximum number of server connections that can be established for this user
@@ -95,14 +98,8 @@ pool_size = 9
 # Maximum query duration. Dangerous, but protects against DBs that died in a non-obvious way.
 statement_timeout = 0
 
-[pools.sharded.users.1]
-username = "postgres"
-password = "postgres"
-pool_size = 21
-statement_timeout = 15000
-
 # Shard 0
-[pools.sharded.shards.0]
+[pools.postgres.shards.0]
 # [ host, port, role ]
 servers = [
     [ "postgres", 5432, "primary" ],
@@ -111,37 +108,16 @@ servers = [
 # Database name (e.g. "postgres")
 database = "postgres"
 
-[pools.sharded.shards.1]
+[pools.postgres.shards.1]
 servers = [
     [ "postgres", 5432, "primary" ],
     [ "postgres", 5432, "replica" ],
 ]
 database = "postgres"
 
-[pools.sharded.shards.2]
+[pools.postgres.shards.2]
 servers = [
     [ "postgres", 5432, "primary" ],
     [ "postgres", 5432, "replica" ],
 ]
 database = "postgres"
-
-
-[pools.simple_db]
-pool_mode = "session"
-default_role = "primary"
-query_parser_enabled = true
-primary_reads_enabled = true
-sharding_function = "pg_bigint_hash"
-
-[pools.simple_db.users.0]
-username = "postgres"
-password = "postgres"
-pool_size = 5
-statement_timeout = 0
-
-[pools.simple_db.shards.0]
-servers = [
-    [ "postgres", 5432, "primary" ],
-    [ "postgres", 5432, "replica" ]
-]
-database = "postgres"

pgcat.toml

@@ -18,51 +18,87 @@ enable_prometheus_exporter = true
 prometheus_exporter_port = 9930
 
 # How long to wait before aborting a server connection (ms).
-connect_timeout = 5000
+connect_timeout = 5000 # milliseconds
+
+# How long an idle connection with a server is left open (ms).
+idle_timeout = 30000 # milliseconds
+
+# Max connection lifetime before it's closed, even if actively used.
+server_lifetime = 86400000 # 24 hours
+
+# How long a client is allowed to be idle while in a transaction (ms).
+idle_client_in_transaction_timeout = 0 # milliseconds
 
 # How much time to give the health check query to return with a result (ms).
-healthcheck_timeout = 1000
+healthcheck_timeout = 1000 # milliseconds
 
 # How long to keep connection available for immediate re-use, without running a healthcheck query on it
-healthcheck_delay = 30000
+healthcheck_delay = 30000 # milliseconds
 
 # How much time to give clients during shutdown before forcibly killing client connections (ms).
-shutdown_timeout = 60000
+shutdown_timeout = 60000 # milliseconds
 
-# For how long to ban a server if it fails a health check (seconds).
+# How long to ban a server if it fails a health check (seconds).
 ban_time = 60 # seconds
 
-# Reload config automatically if it changes.
-autoreload = false
+# If we should log client connections
+log_client_connections = false
 
-# TLS
-# tls_certificate = "server.cert"
-# tls_private_key = "server.key"
+# If we should log client disconnections
+log_client_disconnections = false
 
-# Credentials to access the virtual administrative database (pgbouncer or pgcat)
+# When set to true, PgCat reloads configs if it detects a change in the config file.
+autoreload = 15000
+
+# Number of worker threads the Runtime will use (4 by default).
+worker_threads = 5
+
+# Number of seconds of connection idleness to wait before sending a keepalive packet to the server.
+tcp_keepalives_idle = 5
+# Number of unacknowledged keepalive packets allowed before giving up and closing the connection.
+tcp_keepalives_count = 5
+# Number of seconds between keepalive packets.
+tcp_keepalives_interval = 5
+
+# Path to TLS Certificate file to use for TLS connections
+# tls_certificate = ".circleci/server.cert"
+# Path to TLS private key file to use for TLS connections
+# tls_private_key = ".circleci/server.key"
+
+# Enable/disable server TLS
+server_tls = false
+
+# Verify server certificate is completely authentic.
+verify_server_certificate = false
+
+# User name to access the virtual administrative database (pgbouncer or pgcat)
 # Connecting to that database allows running commands like `SHOW POOLS`, `SHOW DATABASES`, etc..
 admin_username = "admin_user"
+# Password to access the virtual administrative database
 admin_password = "admin_pass"
 
-# pool
-# configs are structured as pool.<pool_name>
-# the pool_name is what clients use as database name when connecting
-# For the example below a client can connect using "postgres://sharding_user:sharding_user@pgcat_host:pgcat_port/sharded_db"
+# pool configs are structured as pool.<pool_name>
+# the pool_name is what clients use as database name when connecting.
+# For a pool named `sharded_db`, clients access that pool using connection string like
+# `postgres://sharding_user:sharding_user@pgcat_host:pgcat_port/sharded_db`
 [pools.sharded_db]
 # Pool mode (see PgBouncer docs for more).
-# session: one server connection per connected client
-# transaction: one server connection per client transaction
+# `session` one server connection per connected client
+# `transaction` one server connection per client transaction
 pool_mode = "transaction"
 
-# If the client doesn't specify, route traffic to
-# this role by default.
-#
-# any: round-robin between primary and replicas,
-# replica: round-robin between replicas only without touching the primary,
-# primary: all queries go to the primary unless otherwise specified.
+# Load balancing mode
+# `random` selects the server at random
+# `loc` selects the server with the least outstanding busy conncetions
+load_balancing_mode = "random"
+
+# If the client doesn't specify, PgCat routes traffic to this role by default.
+# `any` round-robin between primary and replicas,
+# `replica` round-robin between replicas only without touching the primary,
+# `primary` all queries go to the primary unless otherwise specified.
 default_role = "any"
 
-# Query parser. If enabled, we'll attempt to parse
+# If Query Parser is enabled, we'll attempt to parse
 # every incoming query to determine if it's a read or a write.
 # If it's a read query, we'll direct it to a replica. Otherwise, if it's a write,
 # we'll direct it to the primary.
@@ -73,26 +109,70 @@ query_parser_enabled = true
 # queries. The primary can always be explicitly selected with our custom protocol.
 primary_reads_enabled = true
 
+# Allow sharding commands to be passed as statement comments instead of
+# separate commands. If these are unset this functionality is disabled.
+# sharding_key_regex = '/\* sharding_key: (\d+) \*/'
+# shard_id_regex = '/\* shard_id: (\d+) \*/'
+# regex_search_limit = 1000 # only look at the first 1000 characters of SQL statements
+
 # So what if you wanted to implement a different hashing function,
 # or you've already built one and you want this pooler to use it?
-#
 # Current options:
-#
-# pg_bigint_hash: PARTITION BY HASH (Postgres hashing function)
-# sha1: A hashing function based on SHA1
-#
+# `pg_bigint_hash`: PARTITION BY HASH (Postgres hashing function)
+# `sha1`: A hashing function based on SHA1
 sharding_function = "pg_bigint_hash"
 
-# Credentials for users that may connect to this cluster
+# Query to be sent to servers to obtain the hash used for md5 authentication. The connection will be
+# established using the database configured in the pool. This parameter is inherited by every pool
+# and can be redefined in pool configuration.
+# auth_query = "SELECT $1"
+
+# User to be used for connecting to servers to obtain the hash used for md5 authentication by sending the query
+# specified in `auth_query_user`. The connection will be established using the database configured in the pool.
+# This parameter is inherited by every pool and can be redefined in pool configuration.
+# auth_query_user = "sharding_user"
+
+# Password to be used for connecting to servers to obtain the hash used for md5 authentication by sending the query
+# specified in `auth_query_user`. The connection will be established using the database configured in the pool.
+# This parameter is inherited by every pool and can be redefined in pool configuration.
+# auth_query_password = "sharding_user"
+
+# Automatically parse this from queries and route queries to the right shard!
+# automatic_sharding_key = "data.id"
+
+# Idle timeout can be overwritten in the pool
+idle_timeout = 40000
+
+# Connect timeout can be overwritten in the pool
+connect_timeout = 3000
+
+# User configs are structured as pool.<pool_name>.users.<user_index>
+# This section holds the credentials for users that may connect to this cluster
 [pools.sharded_db.users.0]
+# PostgreSQL username used to authenticate the user and connect to the server
+# if `server_username` is not set.
 username = "sharding_user"
+
+# PostgreSQL password used to authenticate the user and connect to the server
+# if `server_password` is not set.
 password = "sharding_user"
+
+pool_mode = "session"
+
+# PostgreSQL username used to connect to the server.
+# server_username = "another_user"
+
+# PostgreSQL password used to connect to the server.
+# server_password = "another_password"
+
 # Maximum number of server connections that can be established for this user
 # The maximum number of connection from a single Pgcat process to any database in the cluster
 # is the sum of pool_size across all users.
 pool_size = 9
 
+
 # Maximum query duration. Dangerous, but protects against DBs that died in a non-obvious way.
+# 0 means it is disabled.
 statement_timeout = 0
 
 [pools.sharded_db.users.1]
@@ -101,28 +181,26 @@ password = "other_user"
 pool_size = 21
 statement_timeout = 15000
 
-# Shard 0
+# Shard configs are structured as pool.<pool_name>.shards.<shard_id>
+# Each shard config contains a list of servers that make up the shard
+# and the database name to use.
 [pools.sharded_db.shards.0]
-# [ host, port, role ]
-servers = [
-    [ "127.0.0.1", 5432, "primary" ],
-    [ "localhost", 5432, "replica" ]
-]
+# Array of servers in the shard, each server entry is an array of `[host, port, role]`
+servers = [["127.0.0.1", 5432, "primary"], ["localhost", 5432, "replica"]]
+
+# Array of mirrors for the shard, each mirror entry is an array of `[host, port, index of server in servers array]`
+# Traffic hitting the server identified by the index will be sent to the mirror.
+# mirrors = [["1.2.3.4", 5432, 0], ["1.2.3.4", 5432, 1]]
+
 # Database name (e.g. "postgres")
 database = "shard0"
 
 [pools.sharded_db.shards.1]
-servers = [
-    [ "127.0.0.1", 5432, "primary" ],
-    [ "localhost", 5432, "replica" ],
-]
+servers = [["127.0.0.1", 5432, "primary"], ["localhost", 5432, "replica"]]
 database = "shard1"
 
 [pools.sharded_db.shards.2]
-servers = [
-    [ "127.0.0.1", 5432, "primary" ],
-    [ "localhost", 5432, "replica" ],
-]
+servers = [["127.0.0.1", 5432, "primary" ], ["localhost", 5432, "replica" ]]
 database = "shard2"
 
 
@@ -137,6 +215,8 @@ sharding_function = "pg_bigint_hash"
 username = "simple_user"
 password = "simple_user"
 pool_size = 5
+min_pool_size = 3
+server_lifetime = 60000
 statement_timeout = 0
 
 [pools.simple_db.shards.0]

src/admin.rs

@@ -1,16 +1,19 @@
-/// Admin database.
+use crate::pool::BanReason;
 use bytes::{Buf, BufMut, BytesMut};
-use log::{info, trace};
+use log::{error, info, trace};
+use nix::sys::signal::{self, Signal};
+use nix::unistd::Pid;
 use std::collections::HashMap;
+/// Admin database.
+use std::sync::atomic::Ordering;
+use std::time::{SystemTime, UNIX_EPOCH};
 use tokio::time::Instant;
 
 use crate::config::{get_config, reload_config, VERSION};
 use crate::errors::Error;
 use crate::messages::*;
-use crate::pool::get_all_pools;
-use crate::stats::{
-    get_address_stats, get_client_stats, get_pool_stats, get_server_stats, ClientState, ServerState,
-};
+use crate::pool::{get_all_pools, get_pool};
+use crate::stats::{get_client_stats, get_pool_stats, get_server_stats, ClientState, ServerState};
 use crate::ClientServerMap;
 
 pub fn generate_server_info_for_admin() -> BytesMut {
@@ -22,7 +25,7 @@ pub fn generate_server_info_for_admin() -> BytesMut {
     server_info.put(server_parameter_message("server_version", VERSION));
     server_info.put(server_parameter_message("DateStyle", "ISO, MDY"));
 
-    return server_info;
+    server_info
 }
 
 /// Handle admin client.
@@ -37,19 +40,28 @@ where
     let code = query.get_u8() as char;
 
     if code != 'Q' {
-        return Err(Error::ProtocolSyncError);
+        return Err(Error::ProtocolSyncError(format!(
+            "Invalid code, expected 'Q' but got '{}'",
+            code
+        )));
     }
 
     let len = query.get_i32() as usize;
-    let query = String::from_utf8_lossy(&query[..len - 5])
-        .to_string()
-        .to_ascii_uppercase();
+    let query = String::from_utf8_lossy(&query[..len - 5]).to_string();
 
     trace!("Admin query: {}", query);
 
     let query_parts: Vec<&str> = query.trim_end_matches(';').split_whitespace().collect();
 
-    match query_parts[0] {
+    match query_parts[0].to_ascii_uppercase().as_str() {
+        "BAN" => {
+            trace!("BAN");
+            ban(stream, query_parts).await
+        }
+        "UNBAN" => {
+            trace!("UNBAN");
+            unban(stream, query_parts).await
+        }
         "RELOAD" => {
             trace!("RELOAD");
             reload(stream, client_server_map).await
@@ -58,7 +70,23 @@ where
             trace!("SET");
             ignore_set(stream).await
         }
-        "SHOW" => match query_parts[1] {
+        "PAUSE" => {
+            trace!("PAUSE");
+            pause(stream, query_parts[1]).await
+        }
+        "RESUME" => {
+            trace!("RESUME");
+            resume(stream, query_parts[1]).await
+        }
+        "SHUTDOWN" => {
+            trace!("SHUTDOWN");
+            shutdown(stream).await
+        }
+        "SHOW" => match query_parts[1].to_ascii_uppercase().as_str() {
+            "BANS" => {
+                trace!("SHOW BANS");
+                show_bans(stream).await
+            }
             "CONFIG" => {
                 trace!("SHOW CONFIG");
                 show_config(stream).await
@@ -91,6 +119,10 @@ where
                 trace!("SHOW VERSION");
                 show_version(stream).await
             }
+            "USERS" => {
+                trace!("SHOW USERS");
+                show_users(stream).await
+            }
             _ => error_response(stream, "Unsupported SHOW query against the admin database").await,
         },
         _ => error_response(stream, "Unsupported query against the admin database").await,
@@ -125,7 +157,14 @@ where
         "free_clients".to_string(),
         client_stats
             .keys()
-            .filter(|client_id| client_stats.get(client_id).unwrap().state == ClientState::Idle)
+            .filter(|client_id| {
+                client_stats
+                    .get(client_id)
+                    .unwrap()
+                    .state
+                    .load(Ordering::Relaxed)
+                    == ClientState::Idle
+            })
             .count()
             .to_string(),
     ]));
@@ -133,7 +172,14 @@ where
         "used_clients".to_string(),
         client_stats
             .keys()
-            .filter(|client_id| client_stats.get(client_id).unwrap().state == ClientState::Active)
+            .filter(|client_id| {
+                client_stats
+                    .get(client_id)
+                    .unwrap()
+                    .state
+                    .load(Ordering::Relaxed)
+                    == ClientState::Active
+            })
             .count()
             .to_string(),
     ]));
@@ -145,7 +191,14 @@ where
         "free_servers".to_string(),
         server_stats
             .keys()
-            .filter(|server_id| server_stats.get(server_id).unwrap().state == ServerState::Idle)
+            .filter(|server_id| {
+                server_stats
+                    .get(server_id)
+                    .unwrap()
+                    .state
+                    .load(Ordering::Relaxed)
+                    == ServerState::Idle
+            })
             .count()
             .to_string(),
     ]));
@@ -153,7 +206,14 @@ where
         "used_servers".to_string(),
         server_stats
             .keys()
-            .filter(|server_id| server_stats.get(server_id).unwrap().state == ServerState::Active)
+            .filter(|server_id| {
+                server_stats
+                    .get(server_id)
+                    .unwrap()
+                    .state
+                    .load(Ordering::Relaxed)
+                    == ServerState::Active
+            })
             .count()
             .to_string(),
     ]));
@@ -168,7 +228,7 @@ where
     res.put_i32(5);
     res.put_u8(b'I');
 
-    write_all_half(stream, res).await
+    write_all_half(stream, &res).await
 }
 
 /// Show PgCat version.
@@ -179,14 +239,14 @@ where
     let mut res = BytesMut::new();
 
     res.put(row_description(&vec![("version", DataType::Text)]));
-    res.put(data_row(&vec![format!("PgCat {}", VERSION).to_string()]));
+    res.put(data_row(&vec![format!("PgCat {}", VERSION)]));
     res.put(command_complete("SHOW"));
 
     res.put_u8(b'Z');
     res.put_i32(5);
     res.put_u8(b'I');
 
-    write_all_half(stream, res).await
+    write_all_half(stream, &res).await
 }
 
 /// Show utilization of connection pools for each shard and replicas.
@@ -215,28 +275,15 @@ where
 
     let mut res = BytesMut::new();
     res.put(row_description(&columns));
-    for (user_pool, pool) in get_all_pools() {
-        let def = HashMap::default();
-        let pool_stats = all_pool_stats
-            .get(&(user_pool.db.clone(), user_pool.user.clone()))
-            .unwrap_or(&def);
 
-        let pool_config = &pool.settings;
+    for ((_user_pool, _pool), pool_stats) in all_pool_stats {
         let mut row = vec![
-            user_pool.db.clone(),
-            user_pool.user.clone(),
-            pool_config.pool_mode.to_string(),
+            pool_stats.database(),
+            pool_stats.user(),
+            pool_stats.pool_mode().to_string(),
         ];
-        for column in &columns[3..columns.len()] {
-            let value = match column.0 {
-                "maxwait" => (pool_stats.get("maxwait_us").unwrap_or(&0) / 1_000_000).to_string(),
-                "maxwait_us" => {
-                    (pool_stats.get("maxwait_us").unwrap_or(&0) % 1_000_000).to_string()
-                }
-                _other_values => pool_stats.get(column.0).unwrap_or(&0).to_string(),
-            };
-            row.push(value);
-        }
+        pool_stats.populate_row(&mut row);
+        pool_stats.clear_maxwait();
         res.put(data_row(&row));
     }
 
@@ -247,7 +294,7 @@ where
     res.put_i32(5);
     res.put_u8(b'I');
 
-    write_all_half(stream, res).await
+    write_all_half(stream, &res).await
 }
 
 /// Show shards and replicas.
@@ -283,7 +330,8 @@ where
             for server in 0..pool.servers(shard) {
                 let address = pool.address(shard, server);
                 let pool_state = pool.pool_state(shard, server);
-                let banned = pool.is_banned(address, Some(address.role));
+                let banned = pool.is_banned(address);
+                let paused = pool.paused();
 
                 res.put(data_row(&vec![
                     address.name(),                         // name
@@ -297,7 +345,11 @@ where
                     pool_config.pool_mode.to_string(),      // pool_mode
                     pool_config.user.pool_size.to_string(), // max_connections
                     pool_state.connections.to_string(),     // current_connections
-                    "0".to_string(),                        // paused
+                    match paused {
+                        // paused
+                        true => "1".to_string(),
+                        false => "0".to_string(),
+                    },
                     match banned {
                         // disabled
                         true => "1".to_string(),
@@ -314,7 +366,7 @@ where
     res.put_i32(5);
     res.put_u8(b'I');
 
-    write_all_half(stream, res).await
+    write_all_half(stream, &res).await
 }
 
 /// Ignore any SET commands the client sends.
@@ -326,6 +378,163 @@ where
     custom_protocol_response_ok(stream, "SET").await
 }
 
+/// Bans a host from being used
+async fn ban<T>(stream: &mut T, tokens: Vec<&str>) -> Result<(), Error>
+where
+    T: tokio::io::AsyncWrite + std::marker::Unpin,
+{
+    let host = match tokens.get(1) {
+        Some(host) => host,
+        None => return error_response(stream, "usage: BAN hostname duration_seconds").await,
+    };
+
+    let duration_seconds = match tokens.get(2) {
+        Some(duration_seconds) => match duration_seconds.parse::<i64>() {
+            Ok(duration_seconds) => duration_seconds,
+            Err(_) => {
+                return error_response(stream, "duration_seconds must be an integer").await;
+            }
+        },
+        None => return error_response(stream, "usage: BAN hostname duration_seconds").await,
+    };
+
+    if duration_seconds <= 0 {
+        return error_response(stream, "duration_seconds must be >= 0").await;
+    }
+
+    let columns = vec![
+        ("db", DataType::Text),
+        ("user", DataType::Text),
+        ("role", DataType::Text),
+        ("host", DataType::Text),
+    ];
+    let mut res = BytesMut::new();
+    res.put(row_description(&columns));
+
+    for (id, pool) in get_all_pools().iter() {
+        for address in pool.get_addresses_from_host(host) {
+            if !pool.is_banned(&address) {
+                pool.ban(&address, BanReason::AdminBan(duration_seconds), None);
+                res.put(data_row(&vec![
+                    id.db.clone(),
+                    id.user.clone(),
+                    address.role.to_string(),
+                    address.host,
+                ]));
+            }
+        }
+    }
+
+    res.put(command_complete("BAN"));
+
+    // ReadyForQuery
+    res.put_u8(b'Z');
+    res.put_i32(5);
+    res.put_u8(b'I');
+
+    write_all_half(stream, &res).await
+}
+
+/// Clear a host for use
+async fn unban<T>(stream: &mut T, tokens: Vec<&str>) -> Result<(), Error>
+where
+    T: tokio::io::AsyncWrite + std::marker::Unpin,
+{
+    let host = match tokens.get(1) {
+        Some(host) => host,
+        None => return error_response(stream, "UNBAN command requires a hostname to unban").await,
+    };
+
+    let columns = vec![
+        ("db", DataType::Text),
+        ("user", DataType::Text),
+        ("role", DataType::Text),
+        ("host", DataType::Text),
+    ];
+    let mut res = BytesMut::new();
+    res.put(row_description(&columns));
+
+    for (id, pool) in get_all_pools().iter() {
+        for address in pool.get_addresses_from_host(host) {
+            if pool.is_banned(&address) {
+                pool.unban(&address);
+                res.put(data_row(&vec![
+                    id.db.clone(),
+                    id.user.clone(),
+                    address.role.to_string(),
+                    address.host,
+                ]));
+            }
+        }
+    }
+
+    res.put(command_complete("UNBAN"));
+
+    // ReadyForQuery
+    res.put_u8(b'Z');
+    res.put_i32(5);
+    res.put_u8(b'I');
+
+    write_all_half(stream, &res).await
+}
+
+/// Shows all the bans
+async fn show_bans<T>(stream: &mut T) -> Result<(), Error>
+where
+    T: tokio::io::AsyncWrite + std::marker::Unpin,
+{
+    let columns = vec![
+        ("db", DataType::Text),
+        ("user", DataType::Text),
+        ("role", DataType::Text),
+        ("host", DataType::Text),
+        ("reason", DataType::Text),
+        ("ban_time", DataType::Text),
+        ("ban_duration_seconds", DataType::Text),
+        ("ban_remaining_seconds", DataType::Text),
+    ];
+    let mut res = BytesMut::new();
+    res.put(row_description(&columns));
+
+    // The block should be pretty quick so we cache the time outside
+    let now = SystemTime::now()
+        .duration_since(UNIX_EPOCH)
+        .expect("Time went backwards")
+        .as_secs() as i64;
+
+    for (id, pool) in get_all_pools().iter() {
+        for (address, (ban_reason, ban_time)) in pool.get_bans().iter() {
+            let ban_duration = match ban_reason {
+                BanReason::AdminBan(duration) => *duration,
+                _ => pool.settings.ban_time,
+            };
+            let remaining = ban_duration - (now - ban_time.timestamp());
+            if remaining <= 0 {
+                continue;
+            }
+            res.put(data_row(&vec![
+                id.db.clone(),
+                id.user.clone(),
+                address.role.to_string(),
+                address.host.clone(),
+                format!("{:?}", ban_reason),
+                ban_time.to_string(),
+                ban_duration.to_string(),
+                remaining.to_string(),
+            ]));
+        }
+    }
+
+    res.put(command_complete("SHOW BANS"));
+
+    // ReadyForQuery
+    res.put_u8(b'Z');
+    res.put_i32(5);
+    res.put_u8(b'I');
+
+    write_all_half(stream, &res).await
+}
+
 /// Reload the configuration file without restarting the process.
 async fn reload<T>(stream: &mut T, client_server_map: ClientServerMap) -> Result<(), Error>
 where
@@ -346,7 +555,7 @@ where
     res.put_i32(5);
     res.put_u8(b'I');
 
-    write_all_half(stream, res).await
+    write_all_half(stream, &res).await
 }
 
 /// Shows current configuration.
@@ -392,7 +601,7 @@ where
     res.put_i32(5);
     res.put_u8(b'I');
 
-    write_all_half(stream, res).await
+    write_all_half(stream, &res).await
 }
 
 /// Show shard and replicas statistics.
@@ -422,7 +631,6 @@ where
         ("avg_wait_time", DataType::Numeric),
     ];
 
-    let all_stats = get_address_stats();
     let mut res = BytesMut::new();
     res.put(row_description(&columns));
 
@@ -430,15 +638,10 @@ where
         for shard in 0..pool.shards() {
             for server in 0..pool.servers(shard) {
                 let address = pool.address(shard, server);
-                let stats = match all_stats.get(&address.id) {
-                    Some(stats) => stats.clone(),
-                    None => HashMap::new(),
-                };
 
                 let mut row = vec![address.name(), user_pool.db.clone(), user_pool.user.clone()];
-                for column in &columns[3..] {
-                    row.push(stats.get(column.0).unwrap_or(&0).to_string());
-                }
+                let stats = address.stats.clone();
+                stats.populate_row(&mut row);
 
                 res.put(data_row(&row));
             }
@@ -452,7 +655,7 @@ where
     res.put_i32(5);
     res.put_u8(b'I');
 
-    write_all_half(stream, res).await
+    write_all_half(stream, &res).await
 }
 
 /// Show currently connected clients
@@ -478,16 +681,16 @@ where
 
     for (_, client) in new_map {
         let row = vec![
-            format!("{:#010X}", client.client_id),
-            client.pool_name,
-            client.username,
-            client.application_name.clone(),
-            client.state.to_string(),
-            client.transaction_count.to_string(),
-            client.query_count.to_string(),
-            client.error_count.to_string(),
+            format!("{:#010X}", client.client_id()),
+            client.pool_name(),
+            client.username(),
+            client.application_name(),
+            client.state.load(Ordering::Relaxed).to_string(),
+            client.transaction_count.load(Ordering::Relaxed).to_string(),
+            client.query_count.load(Ordering::Relaxed).to_string(),
+            client.error_count.load(Ordering::Relaxed).to_string(),
             Instant::now()
-                .duration_since(client.connect_time)
+                .duration_since(client.connect_time())
                 .as_secs()
                 .to_string(),
         ];
@@ -502,7 +705,7 @@ where
     res.put_i32(5);
     res.put_u8(b'I');
 
-    write_all_half(stream, res).await
+    write_all_half(stream, &res).await
 }
 
 /// Show currently connected servers
@@ -529,19 +732,20 @@ where
     res.put(row_description(&columns));
 
     for (_, server) in new_map {
+        let application_name = server.application_name.read();
         let row = vec![
-            format!("{:#010X}", server.server_id),
-            server.pool_name,
-            server.username,
-            server.address_name,
-            server.application_name,
-            server.state.to_string(),
-            server.transaction_count.to_string(),
-            server.query_count.to_string(),
-            server.bytes_sent.to_string(),
-            server.bytes_received.to_string(),
+            format!("{:#010X}", server.server_id()),
+            server.pool_name(),
+            server.username(),
+            server.address_name(),
+            application_name.clone(),
+            server.state.load(Ordering::Relaxed).to_string(),
+            server.transaction_count.load(Ordering::Relaxed).to_string(),
+            server.query_count.load(Ordering::Relaxed).to_string(),
+            server.bytes_sent.load(Ordering::Relaxed).to_string(),
+            server.bytes_received.load(Ordering::Relaxed).to_string(),
             Instant::now()
-                .duration_since(server.connect_time)
+                .duration_since(server.connect_time())
                 .as_secs()
                 .to_string(),
         ];
@@ -556,5 +760,156 @@ where
     res.put_i32(5);
     res.put_u8(b'I');
 
-    write_all_half(stream, res).await
+    write_all_half(stream, &res).await
+}
+
+/// Pause a pool. It won't pass any more queries to the backends.
+async fn pause<T>(stream: &mut T, query: &str) -> Result<(), Error>
+where
+    T: tokio::io::AsyncWrite + std::marker::Unpin,
+{
+    let parts: Vec<&str> = query.split(",").map(|part| part.trim()).collect();
+
+    if parts.len() != 2 {
+        error_response(
+            stream,
+            "PAUSE requires a database and a user, e.g. PAUSE my_db,my_user",
+        )
+        .await
+    } else {
+        let database = parts[0];
+        let user = parts[1];
+
+        match get_pool(database, user) {
+            Some(pool) => {
+                pool.pause();
+
+                let mut res = BytesMut::new();
+
+                res.put(command_complete(&format!("PAUSE {},{}", database, user)));
+
+                // ReadyForQuery
+                res.put_u8(b'Z');
+                res.put_i32(5);
+                res.put_u8(b'I');
+
+                write_all_half(stream, &res).await
+            }
+
+            None => {
+                error_response(
+                    stream,
+                    &format!(
+                        "No pool configured for database: {}, user: {}",
+                        database, user
+                    ),
+                )
+                .await
+            }
+        }
+    }
+}
+
+/// Resume a pool. Queries are allowed again.
+async fn resume<T>(stream: &mut T, query: &str) -> Result<(), Error>
+where
+    T: tokio::io::AsyncWrite + std::marker::Unpin,
+{
+    let parts: Vec<&str> = query.split(",").map(|part| part.trim()).collect();
+
+    if parts.len() != 2 {
+        error_response(
+            stream,
+            "RESUME requires a database and a user, e.g. RESUME my_db,my_user",
+        )
+        .await
+    } else {
+        let database = parts[0];
+        let user = parts[1];
+
+        match get_pool(database, user) {
+            Some(pool) => {
+                pool.resume();
+
+                let mut res = BytesMut::new();
+
+                res.put(command_complete(&format!("RESUME {},{}", database, user)));
+
+                // ReadyForQuery
+                res.put_u8(b'Z');
+                res.put_i32(5);
+                res.put_u8(b'I');
+
+                write_all_half(stream, &res).await
+            }
+
+            None => {
+                error_response(
+                    stream,
+                    &format!(
+                        "No pool configured for database: {}, user: {}",
+                        database, user
+                    ),
+                )
+                .await
+            }
+        }
+    }
+}
+
+/// Send response packets for shutdown.
+async fn shutdown<T>(stream: &mut T) -> Result<(), Error>
+where
+    T: tokio::io::AsyncWrite + std::marker::Unpin,
+{
+    let mut res = BytesMut::new();
+
+    res.put(row_description(&vec![("success", DataType::Text)]));
+
+    let mut shutdown_success = "t";
+
+    let pid = std::process::id();
+    if signal::kill(Pid::from_raw(pid.try_into().unwrap()), Signal::SIGINT).is_err() {
+        error!("Unable to send SIGINT to PID: {}", pid);
+        shutdown_success = "f";
+    }
+
+    res.put(data_row(&vec![shutdown_success.to_string()]));
+
+    res.put(command_complete("SHUTDOWN"));
+
+    res.put_u8(b'Z');
+    res.put_i32(5);
+    res.put_u8(b'I');
+
+    write_all_half(stream, &res).await
+}
+
+/// Show Users.
+async fn show_users<T>(stream: &mut T) -> Result<(), Error>
+where
+    T: tokio::io::AsyncWrite + std::marker::Unpin,
+{
+    let mut res = BytesMut::new();
+
+    res.put(row_description(&vec![
+        ("name", DataType::Text),
+        ("pool_mode", DataType::Text),
+    ]));
+
+    for (user_pool, pool) in get_all_pools() {
+        let pool_config = &pool.settings;
+        res.put(data_row(&vec![
+            user_pool.user.clone(),
+            pool_config.pool_mode.to_string(),
+        ]));
+    }
+
+    res.put(command_complete("SHOW"));
+
+    res.put_u8(b'Z');
+    res.put_i32(5);
+    res.put_u8(b'I');
+
+    write_all_half(stream, &res).await
 }

src/auth_passthrough.rs

@@ -0,0 +1,134 @@
+use crate::errors::Error;
+use crate::pool::ConnectionPool;
+use crate::server::Server;
+use log::debug;
+
+#[derive(Clone, Debug)]
+pub struct AuthPassthrough {
+    password: String,
+    query: String,
+    user: String,
+}
+
+impl AuthPassthrough {
+    /// Initializes an AuthPassthrough.
+    pub fn new(query: &str, user: &str, password: &str) -> Self {
+        AuthPassthrough {
+            password: password.to_string(),
+            query: query.to_string(),
+            user: user.to_string(),
+        }
+    }
+
+    /// Returns an AuthPassthrough given the pool configuration.
+    /// If any of required values is not set, None is returned.
+    pub fn from_pool_config(pool_config: &crate::config::Pool) -> Option<Self> {
+        if pool_config.is_auth_query_configured() {
+            return Some(AuthPassthrough::new(
+                pool_config.auth_query.as_ref().unwrap(),
+                pool_config.auth_query_user.as_ref().unwrap(),
+                pool_config.auth_query_password.as_ref().unwrap(),
+            ));
+        }
+
+        None
+    }
+
+    /// Returns an AuthPassthrough given the pool settings.
+    /// If any of required values is not set, None is returned.
+    pub fn from_pool_settings(pool_settings: &crate::pool::PoolSettings) -> Option<Self> {
+        let pool_config = crate::config::Pool {
+            auth_query: pool_settings.auth_query.clone(),
+            auth_query_password: pool_settings.auth_query_password.clone(),
+            auth_query_user: pool_settings.auth_query_user.clone(),
+            ..Default::default()
+        };
+
+        AuthPassthrough::from_pool_config(&pool_config)
+    }
+
+    /// Connects to server and executes auth_query for the specified address.
+    /// If the response is a row with two columns containing the username set in the address.
+    /// and its MD5 hash, the MD5 hash returned.
+    ///
+    /// Note that the query is executed, changing $1 with the name of the user
+    /// this is so we only hold in memory (and transfer) the least amount of 'sensitive' data.
+    /// Also, it is compatible with pgbouncer.
+    ///
+    /// # Arguments
+    ///
+    /// * `address` - An Address of the server we want to connect to. The username for the hash will be obtained from this value.
+    ///
+    /// # Examples
+    ///
+    /// ```
+    /// use pgcat::auth_passthrough::AuthPassthrough;
+    /// use pgcat::config::Address;
+    /// let auth_passthrough = AuthPassthrough::new("SELECT * FROM public.user_lookup('$1');", "postgres", "postgres");
+    /// auth_passthrough.fetch_hash(&Address::default());
+    /// ```
+    ///
+    pub async fn fetch_hash(&self, address: &crate::config::Address) -> Result<String, Error> {
+        let auth_user = crate::config::User {
+            username: self.user.clone(),
+            password: Some(self.password.clone()),
+            server_username: None,
+            server_password: None,
+            pool_size: 1,
+            statement_timeout: 0,
+            pool_mode: None,
+            server_lifetime: None,
+            min_pool_size: None,
+        };
+
+        let user = &address.username;
+
+        debug!("Connecting to server to obtain auth hashes");
+
+        let auth_query = self.query.replace("$1", user);
+
+        match Server::exec_simple_query(address, &auth_user, &auth_query).await {
+            Ok(password_data) => {
+                if password_data.len() == 2 && password_data.first().unwrap() == user {
+                    if let Some(stripped_hash) = password_data
+                        .last()
+                        .unwrap()
+                        .to_string()
+                        .strip_prefix("md5") {
+                            Ok(stripped_hash.to_string())
+                        }
+                    else {
+                        Err(Error::AuthPassthroughError(
+                            "Obtained hash from auth_query does not seem to be in md5 format.".to_string(),
+                        ))
+                    }
+                } else {
+                    Err(Error::AuthPassthroughError(
+                        "Data obtained from query does not follow the scheme 'user','hash'."
+                            .to_string(),
+                    ))
+                 }
+            }
+            Err(err) => {
+                Err(Error::AuthPassthroughError(
+                    format!("Error trying to obtain password from auth_query, ignoring hash for user '{}'. Error: {:?}",
+                        user, err))
+                )
+            }
+        }
+    }
+}
+
+pub async fn refetch_auth_hash(pool: &ConnectionPool) -> Result<String, Error> {
+    let address = pool.address(0, 0);
+    if let Some(apt) = AuthPassthrough::from_pool_settings(&pool.settings) {
+        let hash = apt.fetch_hash(address).await?;
+
+        return Ok(hash);
+    }
+
+    Err(Error::ClientError(format!(
+        "Could not obtain hash for {{ username: {:?}, database: {:?} }}. Auth passthrough not enabled.",
+        address.username, address.database
+    )))
+}

src/config.rs

@@ -2,18 +2,20 @@
 use arc_swap::ArcSwap;
 use log::{error, info};
 use once_cell::sync::Lazy;
+use regex::Regex;
 use serde_derive::{Deserialize, Serialize};
+use std::collections::hash_map::DefaultHasher;
 use std::collections::{BTreeMap, HashMap, HashSet};
-use std::hash::Hash;
+use std::hash::{Hash, Hasher};
 use std::path::Path;
 use std::sync::Arc;
 use tokio::fs::File;
 use tokio::io::AsyncReadExt;
-use toml;
 
 use crate::errors::Error;
 use crate::pool::{ClientServerMap, ConnectionPool};
 use crate::sharding::ShardingFunction;
+use crate::stats::AddressStats;
 use crate::tls::{load_certs, load_keys};
 
 pub const VERSION: &str = env!("CARGO_PKG_VERSION");
@@ -28,6 +30,8 @@ pub enum Role {
     Primary,
     #[serde(alias = "replica", alias = "Replica")]
     Replica,
+    #[serde(alias = "mirror", alias = "Mirror")]
+    Mirror,
 }
 
 impl ToString for Role {
@@ -35,6 +39,7 @@ impl ToString for Role {
         match *self {
             Role::Primary => "primary".to_string(),
             Role::Replica => "replica".to_string(),
+            Role::Mirror => "mirror".to_string(),
         }
     }
 }
@@ -58,7 +63,7 @@ impl PartialEq<Role> for Option<Role> {
 }
 
 /// Address identifying a PostgreSQL server uniquely.
-#[derive(Clone, PartialEq, Hash, std::cmp::Eq, Debug)]
+#[derive(Clone, Debug)]
 pub struct Address {
     /// Unique ID per addressable Postgres server.
     pub id: usize,
@@ -89,6 +94,12 @@ pub struct Address {
 
     /// The name of this pool (i.e. database name visible to the client).
     pub pool_name: String,
+
+    /// List of addresses to receive mirrored traffic.
+    pub mirrors: Vec<Address>,
+
+    /// Address stats
+    pub stats: Arc<AddressStats>,
 }
 
 impl Default for Address {
@@ -104,20 +115,60 @@ impl Default for Address {
             role: Role::Replica,
             username: String::from("username"),
             pool_name: String::from("pool_name"),
+            mirrors: Vec::new(),
+            stats: Arc::new(AddressStats::default()),
         }
     }
 }
 
+// We need to implement PartialEq by ourselves so we skip stats in the comparison
+impl PartialEq for Address {
+    fn eq(&self, other: &Self) -> bool {
+        self.id == other.id
+            && self.host == other.host
+            && self.port == other.port
+            && self.shard == other.shard
+            && self.address_index == other.address_index
+            && self.replica_number == other.replica_number
+            && self.database == other.database
+            && self.role == other.role
+            && self.username == other.username
+            && self.pool_name == other.pool_name
+            && self.mirrors == other.mirrors
+    }
+}
+impl Eq for Address {}
+
+// We need to implement Hash by ourselves so we skip stats in the comparison
+impl Hash for Address {
+    fn hash<H: Hasher>(&self, state: &mut H) {
+        self.id.hash(state);
+        self.host.hash(state);
+        self.port.hash(state);
+        self.shard.hash(state);
+        self.address_index.hash(state);
+        self.replica_number.hash(state);
+        self.database.hash(state);
+        self.role.hash(state);
+        self.username.hash(state);
+        self.pool_name.hash(state);
+        self.mirrors.hash(state);
+    }
+}
+
 impl Address {
     /// Address name (aka database) used in `SHOW STATS`, `SHOW DATABASES`, and `SHOW POOLS`.
     pub fn name(&self) -> String {
         match self.role {
             Role::Primary => format!("{}_shard_{}_primary", self.pool_name, self.shard),
-
             Role::Replica => format!(
                 "{}_shard_{}_replica_{}",
                 self.pool_name, self.shard, self.replica_number
             ),
+            Role::Mirror => format!(
+                "{}_shard_{}_mirror_{}",
+                self.pool_name, self.shard, self.replica_number
+            ),
         }
     }
 }
@@ -126,8 +177,13 @@ impl Address {
 #[derive(Clone, PartialEq, Hash, Eq, Serialize, Deserialize, Debug)]
 pub struct User {
     pub username: String,
-    pub password: String,
+    pub password: Option<String>,
+    pub server_username: Option<String>,
+    pub server_password: Option<String>,
     pub pool_size: u32,
+    pub min_pool_size: Option<u32>,
+    pub pool_mode: Option<PoolMode>,
+    pub server_lifetime: Option<u64>,
     #[serde(default)] // 0
     pub statement_timeout: u64,
 }
@@ -136,13 +192,38 @@ impl Default for User {
     fn default() -> User {
         User {
             username: String::from("postgres"),
-            password: String::new(),
+            password: None,
+            server_username: None,
+            server_password: None,
             pool_size: 15,
+            min_pool_size: None,
             statement_timeout: 0,
+            pool_mode: None,
+            server_lifetime: None,
         }
     }
 }
 
+impl User {
+    fn validate(&self) -> Result<(), Error> {
+        match self.min_pool_size {
+            Some(min_pool_size) => {
+                if min_pool_size > self.pool_size {
+                    error!(
+                        "min_pool_size of {} cannot be larger than pool_size of {}",
+                        min_pool_size, self.pool_size
+                    );
+                    return Err(Error::BadConfig);
+                }
+            }
+
+            None => (),
+        };
+
+        Ok(())
+    }
+}
+
 /// General configuration.
 #[derive(Serialize, Deserialize, Debug, Clone, PartialEq)]
 pub struct General {
@@ -150,7 +231,7 @@ pub struct General {
     pub host: String,
 
     #[serde(default = "General::default_port")]
-    pub port: i16,
+    pub port: u16,
 
     pub enable_prometheus_exporter: Option<bool>,
     pub prometheus_exporter_port: i16,
@@ -158,6 +239,22 @@ pub struct General {
     #[serde(default = "General::default_connect_timeout")]
     pub connect_timeout: u64,
 
+    #[serde(default = "General::default_idle_timeout")]
+    pub idle_timeout: u64,
+
+    #[serde(default = "General::default_tcp_keepalives_idle")]
+    pub tcp_keepalives_idle: u64,
+    #[serde(default = "General::default_tcp_keepalives_count")]
+    pub tcp_keepalives_count: u32,
+    #[serde(default = "General::default_tcp_keepalives_interval")]
+    pub tcp_keepalives_interval: u64,
+
+    #[serde(default)] // False
+    pub log_client_connections: bool,
+
+    #[serde(default)] // False
+    pub log_client_disconnections: bool,
+
     #[serde(default = "General::default_shutdown_timeout")]
     pub shutdown_timeout: u64,
 
@@ -170,13 +267,33 @@ pub struct General {
     #[serde(default = "General::default_ban_time")]
     pub ban_time: i64,
 
-    #[serde(default)] // False
-    pub autoreload: bool,
+    #[serde(default = "General::default_idle_client_in_transaction_timeout")]
+    pub idle_client_in_transaction_timeout: u64,
+
+    #[serde(default = "General::default_server_lifetime")]
+    pub server_lifetime: u64,
+
+    #[serde(default = "General::default_worker_threads")]
+    pub worker_threads: usize,
+
+    #[serde(default)] // None
+    pub autoreload: Option<u64>,
 
     pub tls_certificate: Option<String>,
     pub tls_private_key: Option<String>,
+
+    #[serde(default)] // false
+    pub server_tls: bool,
+
+    #[serde(default)] // false
+    pub verify_server_certificate: bool,
+
     pub admin_username: String,
     pub admin_password: String,
+
+    pub auth_query: Option<String>,
+    pub auth_query_user: Option<String>,
+    pub auth_query_password: Option<String>,
 }
 
 impl General {
@@ -184,14 +301,37 @@ impl General {
         "0.0.0.0".into()
     }
 
-    pub fn default_port() -> i16 {
+    pub fn default_port() -> u16 {
         5432
     }
 
+    pub fn default_server_lifetime() -> u64 {
+        1000 * 60 * 60 * 24 // 24 hours
+    }
+
     pub fn default_connect_timeout() -> u64 {
         1000
     }
 
+    // These keepalive defaults should detect a dead connection within 30 seconds.
+    // Tokio defaults to disabling keepalives which keeps dead connections around indefinitely.
+    // This can lead to permanent server pool exhaustion
+    pub fn default_tcp_keepalives_idle() -> u64 {
+        5 // 5 seconds
+    }
+
+    pub fn default_tcp_keepalives_count() -> u32 {
+        5 // 5 time
+    }
+
+    pub fn default_tcp_keepalives_interval() -> u64 {
+        5 // 5 seconds
+    }
+
+    pub fn default_idle_timeout() -> u64 {
+        60000 // 10 minutes
+    }
+
     pub fn default_shutdown_timeout() -> u64 {
         60000
     }
@@ -207,6 +347,14 @@ impl General {
     pub fn default_ban_time() -> i64 {
         60
     }
+
+    pub fn default_worker_threads() -> usize {
+        4
+    }
+
+    pub fn default_idle_client_in_transaction_timeout() -> u64 {
+        0
+    }
 }
 
 impl Default for General {
@@ -217,15 +365,29 @@ impl Default for General {
             enable_prometheus_exporter: Some(false),
             prometheus_exporter_port: 9930,
             connect_timeout: General::default_connect_timeout(),
+            idle_timeout: General::default_idle_timeout(),
             shutdown_timeout: Self::default_shutdown_timeout(),
             healthcheck_timeout: Self::default_healthcheck_timeout(),
             healthcheck_delay: Self::default_healthcheck_delay(),
             ban_time: Self::default_ban_time(),
-            autoreload: false,
+            worker_threads: Self::default_worker_threads(),
+            idle_client_in_transaction_timeout: Self::default_idle_client_in_transaction_timeout(),
+            tcp_keepalives_idle: Self::default_tcp_keepalives_idle(),
+            tcp_keepalives_count: Self::default_tcp_keepalives_count(),
+            tcp_keepalives_interval: Self::default_tcp_keepalives_interval(),
+            log_client_connections: false,
+            log_client_disconnections: false,
+            autoreload: None,
             tls_certificate: None,
             tls_private_key: None,
+            server_tls: false,
+            verify_server_certificate: false,
             admin_username: String::from("admin"),
             admin_password: String::from("admin"),
+            auth_query: None,
+            auth_query_user: None,
+            auth_query_password: None,
+            server_lifetime: 1000 * 3600 * 24, // 24 hours,
         }
     }
 }
@@ -251,11 +413,33 @@ impl ToString for PoolMode {
     }
 }
 
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Eq, Copy, Hash)]
+pub enum LoadBalancingMode {
+    #[serde(alias = "random", alias = "Random")]
+    Random,
+
+    #[serde(alias = "loc", alias = "LOC", alias = "least_outstanding_connections")]
+    LeastOutstandingConnections,
+}
+impl ToString for LoadBalancingMode {
+    fn to_string(&self) -> String {
+        match *self {
+            LoadBalancingMode::Random => "random".to_string(),
+            LoadBalancingMode::LeastOutstandingConnections => {
+                "least_outstanding_connections".to_string()
+            }
+        }
+    }
+}
+
 #[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Eq, Hash)]
 pub struct Pool {
     #[serde(default = "Pool::default_pool_mode")]
     pub pool_mode: PoolMode,
 
+    #[serde(default = "Pool::default_load_balancing_mode")]
+    pub load_balancing_mode: LoadBalancingMode,
+
     pub default_role: String,
 
     #[serde(default)] // False
@@ -266,17 +450,56 @@ pub struct Pool {
 
     pub connect_timeout: Option<u64>,
 
+    pub idle_timeout: Option<u64>,
+
+    pub server_lifetime: Option<u64>,
+
     pub sharding_function: ShardingFunction,
+
+    #[serde(default = "Pool::default_automatic_sharding_key")]
+    pub automatic_sharding_key: Option<String>,
+
+    pub sharding_key_regex: Option<String>,
+    pub shard_id_regex: Option<String>,
+    pub regex_search_limit: Option<usize>,
+
+    pub auth_query: Option<String>,
+    pub auth_query_user: Option<String>,
+    pub auth_query_password: Option<String>,
+
     pub shards: BTreeMap<String, Shard>,
     pub users: BTreeMap<String, User>,
+    // Note, don't put simple fields below these configs. There's a compatibility issue with TOML that makes it
+    // incompatible to have simple fields in TOML after complex objects. See
+    // https://users.rust-lang.org/t/why-toml-to-string-get-error-valueaftertable/85903
 }
 
 impl Pool {
+    pub fn hash_value(&self) -> u64 {
+        let mut s = DefaultHasher::new();
+        self.hash(&mut s);
+        s.finish()
+    }
+
+    pub fn is_auth_query_configured(&self) -> bool {
+        self.auth_query_password.is_some()
+            && self.auth_query_user.is_some()
+            && self.auth_query_password.is_some()
+    }
+
     pub fn default_pool_mode() -> PoolMode {
         PoolMode::Transaction
     }
 
-    pub fn validate(&self) -> Result<(), Error> {
+    pub fn default_load_balancing_mode() -> LoadBalancingMode {
+        LoadBalancingMode::Random
+    }
+
+    pub fn default_automatic_sharding_key() -> Option<String> {
+        None
+    }
+
+    pub fn validate(&mut self) -> Result<(), Error> {
         match self.default_role.as_ref() {
             "any" => (),
             "primary" => (),
@@ -304,6 +527,41 @@ impl Pool {
             shard.validate()?;
         }
 
+        for (option, name) in [
+            (&self.shard_id_regex, "shard_id_regex"),
+            (&self.sharding_key_regex, "sharding_key_regex"),
+        ] {
+            if let Some(regex) = option {
+                if let Err(parse_err) = Regex::new(regex.as_str()) {
+                    error!("{} is not a valid Regex: {}", name, parse_err);
+                    return Err(Error::BadConfig);
+                }
+            }
+        }
+
+        self.automatic_sharding_key = match &self.automatic_sharding_key {
+            Some(key) => {
+                // No quotes in the key so we don't have to compare quoted
+                // to unquoted idents.
+                let key = key.replace("\"", "");
+
+                if key.split(".").count() != 2 {
+                    error!(
+                        "automatic_sharding_key '{}' must be fully qualified, e.g. t.{}`",
+                        key, key
+                    );
+                    return Err(Error::BadConfig);
+                }
+
+                Some(key)
+            }
+            None => None,
+        };
+
+        for (_, user) in &self.users {
+            user.validate()?;
+        }
+
         Ok(())
     }
 }
@@ -312,13 +570,23 @@ impl Default for Pool {
     fn default() -> Pool {
         Pool {
             pool_mode: Self::default_pool_mode(),
+            load_balancing_mode: Self::default_load_balancing_mode(),
             shards: BTreeMap::from([(String::from("1"), Shard::default())]),
             users: BTreeMap::default(),
             default_role: String::from("any"),
             query_parser_enabled: false,
             primary_reads_enabled: false,
             sharding_function: ShardingFunction::PgBigintHash,
+            automatic_sharding_key: None,
             connect_timeout: None,
+            idle_timeout: None,
+            sharding_key_regex: None,
+            shard_id_regex: None,
+            regex_search_limit: Some(1000),
+            auth_query: None,
+            auth_query_user: None,
+            auth_query_password: None,
+            server_lifetime: None,
         }
     }
 }
@@ -330,10 +598,18 @@ pub struct ServerConfig {
     pub role: Role,
 }
 
+#[derive(Clone, PartialEq, Serialize, Deserialize, Debug, Hash, Eq)]
+pub struct MirrorServerConfig {
+    pub host: String,
+    pub port: u16,
+    pub mirroring_target_index: usize,
+}
+
 /// Shard configuration.
 #[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Hash, Eq)]
 pub struct Shard {
     pub database: String,
+    pub mirrors: Option<Vec<MirrorServerConfig>>,
     pub servers: Vec<ServerConfig>,
 }
 
@@ -344,7 +620,7 @@ impl Shard {
         let mut dup_check = HashSet::new();
         let mut primary_count = 0;
 
-        if self.servers.len() == 0 {
+        if self.servers.is_empty() {
             error!("Shard {} has no servers configured", self.database);
             return Err(Error::BadConfig);
         }
@@ -353,15 +629,14 @@ impl Shard {
             dup_check.insert(server);
 
             // Check that we define only zero or one primary.
-            match server.role {
-                Role::Primary => primary_count += 1,
-                _ => (),
-            };
+            if server.role == Role::Primary {
+                primary_count += 1
+            }
         }
 
         if primary_count > 1 {
             error!(
-                "Shard {} has more than on primary configured",
+                "Shard {} has more than one primary configured",
                 self.database
             );
             return Err(Error::BadConfig);
@@ -384,6 +659,7 @@ impl Default for Shard {
                 port: 5432,
                 role: Role::Primary,
             }],
+            mirrors: None,
             database: String::from("postgres"),
         }
     }
@@ -411,9 +687,31 @@ pub struct Config {
 }
 
 impl Config {
+    pub fn is_auth_query_configured(&self) -> bool {
+        self.pools
+            .iter()
+            .any(|(_name, pool)| pool.is_auth_query_configured())
+    }
+
     pub fn default_path() -> String {
         String::from("pgcat.toml")
     }
+
+    pub fn fill_up_auth_query_config(&mut self) {
+        for (_name, pool) in self.pools.iter_mut() {
+            if pool.auth_query.is_none() {
+                pool.auth_query = self.general.auth_query.clone();
+            }
+
+            if pool.auth_query_user.is_none() {
+                pool.auth_query_user = self.general.auth_query_user.clone();
+            }
+
+            if pool.auth_query_password.is_none() {
+                pool.auth_query_password = self.general.auth_query_password.clone();
+            }
+        }
+    }
 }
 
 impl Default for Config {
@@ -437,6 +735,10 @@ impl From<&Config> for std::collections::HashMap<String, String> {
                         format!("pools.{}.pool_mode", pool_name),
                         pool.pool_mode.to_string(),
                     ),
+                    (
+                        format!("pools.{}.load_balancing_mode", pool_name),
+                        pool.load_balancing_mode.to_string(),
+                    ),
                     (
                         format!("pools.{}.primary_reads_enabled", pool_name),
                         pool.primary_reads_enabled.to_string(),
@@ -481,6 +783,10 @@ impl From<&Config> for std::collections::HashMap<String, String> {
                 "connect_timeout".to_string(),
                 config.general.connect_timeout.to_string(),
             ),
+            (
+                "idle_timeout".to_string(),
+                config.general.idle_timeout.to_string(),
+            ),
             (
                 "healthcheck_timeout".to_string(),
                 config.general.healthcheck_timeout.to_string(),
@@ -494,6 +800,13 @@ impl From<&Config> for std::collections::HashMap<String, String> {
                 config.general.healthcheck_delay.to_string(),
             ),
             ("ban_time".to_string(), config.general.ban_time.to_string()),
+            (
+                "idle_client_in_transaction_timeout".to_string(),
+                config
+                    .general
+                    .idle_client_in_transaction_timeout
+                    .to_string(),
+            ),
         ];
 
         r.append(&mut static_settings);
@@ -505,13 +818,31 @@ impl Config {
     /// Print current configuration.
     pub fn show(&self) {
         info!("Ban time: {}s", self.general.ban_time);
+        info!(
+            "Idle client in transaction timeout: {}ms",
+            self.general.idle_client_in_transaction_timeout
+        );
+        info!("Worker threads: {}", self.general.worker_threads);
         info!(
             "Healthcheck timeout: {}ms",
             self.general.healthcheck_timeout
         );
         info!("Connection timeout: {}ms", self.general.connect_timeout);
+        info!("Idle timeout: {}ms", self.general.idle_timeout);
+        info!(
+            "Log client connections: {}",
+            self.general.log_client_connections
+        );
+        info!(
+            "Log client disconnections: {}",
+            self.general.log_client_disconnections
+        );
         info!("Shutdown timeout: {}ms", self.general.shutdown_timeout);
         info!("Healthcheck delay: {}ms", self.general.healthcheck_delay);
+        info!(
+            "Default max server lifetime: {}ms",
+            self.general.server_lifetime
+        );
         match self.general.tls_certificate.clone() {
             Some(tls_certificate) => {
                 info!("TLS certificate: {}", tls_certificate);
@@ -530,6 +861,11 @@ impl Config {
                 info!("TLS support is disabled");
             }
         };
+        info!("Server TLS enabled: {}", self.general.server_tls);
+        info!(
+            "Server TLS certificate verification: {}",
+            self.general.verify_server_certificate
+        );
 
         for (pool_name, pool_config) in &self.pools {
             // TODO: Make this output prettier (maybe a table?)
@@ -544,8 +880,13 @@ impl Config {
                     .to_string()
             );
             info!(
-                "[pool: {}] Pool mode: {:?}",
-                pool_name, pool_config.pool_mode
+                "[pool: {}] Default pool mode: {}",
+                pool_name,
+                pool_config.pool_mode.to_string()
+            );
+            info!(
+                "[pool: {}] Load Balancing mode: {:?}",
+                pool_name, pool_config.load_balancing_mode
             );
             let connect_timeout = match pool_config.connect_timeout {
                 Some(connect_timeout) => connect_timeout,
@@ -555,6 +896,11 @@ impl Config {
                 "[pool: {}] Connection timeout: {}ms",
                 pool_name, connect_timeout
             );
+            let idle_timeout = match pool_config.idle_timeout {
+                Some(idle_timeout) => idle_timeout,
+                None => self.general.idle_timeout,
+            };
+            info!("[pool: {}] Idle timeout: {}ms", pool_name, idle_timeout);
             info!(
                 "[pool: {}] Sharding function: {}",
                 pool_name,
@@ -578,40 +924,115 @@ impl Config {
                 pool_name,
                 pool_config.users.len()
             );
+            info!(
+                "[pool: {}] Max server lifetime: {}",
+                pool_name,
+                match pool_config.server_lifetime {
+                    Some(server_lifetime) => format!("{}ms", server_lifetime),
+                    None => "default".to_string(),
+                }
+            );
 
             for user in &pool_config.users {
                 info!(
                     "[pool: {}][user: {}] Pool size: {}",
                     pool_name, user.1.username, user.1.pool_size,
                 );
+                info!(
+                    "[pool: {}][user: {}] Minimum pool size: {}",
+                    pool_name,
+                    user.1.username,
+                    user.1.min_pool_size.unwrap_or(0)
+                );
                 info!(
                     "[pool: {}][user: {}] Statement timeout: {}",
                     pool_name, user.1.username, user.1.statement_timeout
-                )
+                );
+                info!(
+                    "[pool: {}][user: {}] Pool mode: {}",
+                    pool_name,
+                    user.1.username,
+                    match user.1.pool_mode {
+                        Some(pool_mode) => pool_mode.to_string(),
+                        None => pool_config.pool_mode.to_string(),
+                    }
+                );
+                info!(
+                    "[pool: {}][user: {}] Max server lifetime: {}",
+                    pool_name,
+                    user.1.username,
+                    match user.1.server_lifetime {
+                        Some(server_lifetime) => format!("{}ms", server_lifetime),
+                        None => "default".to_string(),
+                    }
+                );
             }
         }
     }
 
     pub fn validate(&mut self) -> Result<(), Error> {
+        // Validation for auth_query feature
+        if self.general.auth_query.is_some()
+            && (self.general.auth_query_user.is_none()
+                || self.general.auth_query_password.is_none())
+        {
+            error!(
+                "If auth_query is specified, \
+                you need to provide a value \
+                for `auth_query_user`, \
+                `auth_query_password`"
+            );
+
+            return Err(Error::BadConfig);
+        }
+
+        for (name, pool) in self.pools.iter() {
+            if pool.auth_query.is_some()
+                && (pool.auth_query_user.is_none() || pool.auth_query_password.is_none())
+            {
+                error!(
+                    "Error in pool {{ {} }}. \
+                    If auth_query is specified, you need \
+                    to provide a value for `auth_query_user`, \
+                    `auth_query_password`",
+                    name
+                );
+
+                return Err(Error::BadConfig);
+            }
+
+            for (_name, user_data) in pool.users.iter() {
+                if (pool.auth_query.is_none()
+                    || pool.auth_query_password.is_none()
+                    || pool.auth_query_user.is_none())
+                    && user_data.password.is_none()
+                {
+                    error!(
+                        "Error in pool {{ {} }}. \
+                        You have to specify a user password \
+                        for every pool if auth_query is not specified",
+                        name
+                    );
+
+                    return Err(Error::BadConfig);
+                }
+            }
+        }
+
         // Validate TLS!
         match self.general.tls_certificate.clone() {
             Some(tls_certificate) => {
-                match load_certs(&Path::new(&tls_certificate)) {
+                match load_certs(Path::new(&tls_certificate)) {
                     Ok(_) => {
                         // Cert is okay, but what about the private key?
                         match self.general.tls_private_key.clone() {
-                            Some(tls_private_key) => {
-                                match load_keys(&Path::new(&tls_private_key)) {
-                                    Ok(_) => (),
-                                    Err(err) => {
-                                        error!(
-                                            "tls_private_key is incorrectly configured: {:?}",
-                                            err
-                                        );
-                                        return Err(Error::BadConfig);
-                                    }
+                            Some(tls_private_key) => match load_keys(Path::new(&tls_private_key)) {
+                                Ok(_) => (),
+                                Err(err) => {
+                                    error!("tls_private_key is incorrectly configured: {:?}", err);
+                                    return Err(Error::BadConfig);
                                 }
-                            }
+                            },
 
                             None => {
                                 error!("tls_certificate is set, but the tls_private_key is not");
@@ -629,7 +1050,7 @@ impl Config {
             None => (),
         };
 
-        for (_, pool) in &mut self.pools {
+        for pool in self.pools.values_mut() {
             pool.validate()?;
         }
 
@@ -644,6 +1065,12 @@ pub fn get_config() -> Config {
     (*(*CONFIG.load())).clone()
 }
 
+pub fn get_idle_client_in_transaction_timeout() -> u64 {
+    (*(*CONFIG.load()))
+        .general
+        .idle_client_in_transaction_timeout
+}
+
 /// Parse the configuration file located at the path.
 pub async fn parse(path: &str) -> Result<(), Error> {
     let mut contents = String::new();
@@ -671,6 +1098,7 @@ pub async fn parse(path: &str) -> Result<(), Error> {
         }
     };
 
+    config.fill_up_auth_query_config();
     config.validate()?;
 
     config.path = path.to_string();
@@ -714,8 +1142,11 @@ mod test {
         assert_eq!(get_config().path, "pgcat.toml".to_string());
 
         assert_eq!(get_config().general.ban_time, 60);
+        assert_eq!(get_config().general.idle_client_in_transaction_timeout, 0);
+        assert_eq!(get_config().general.idle_timeout, 30000);
         assert_eq!(get_config().pools.len(), 2);
         assert_eq!(get_config().pools["sharded_db"].shards.len(), 3);
+        assert_eq!(get_config().pools["sharded_db"].idle_timeout, Some(40000));
         assert_eq!(get_config().pools["simple_db"].shards.len(), 1);
         assert_eq!(get_config().pools["sharded_db"].users.len(), 2);
         assert_eq!(get_config().pools["simple_db"].users.len(), 1);
@@ -737,7 +1168,10 @@ mod test {
             "sharding_user"
         );
         assert_eq!(
-            get_config().pools["sharded_db"].users["1"].password,
+            get_config().pools["sharded_db"].users["1"]
+                .password
+                .as_ref()
+                .unwrap(),
             "other_user"
         );
         assert_eq!(get_config().pools["sharded_db"].users["1"].pool_size, 21);
@@ -762,10 +1196,16 @@ mod test {
             "simple_user"
         );
         assert_eq!(
-            get_config().pools["simple_db"].users["0"].password,
+            get_config().pools["simple_db"].users["0"]
+                .password
+                .as_ref()
+                .unwrap(),
             "simple_user"
         );
         assert_eq!(get_config().pools["simple_db"].users["0"].pool_size, 5);
+        assert_eq!(get_config().general.auth_query, None);
+        assert_eq!(get_config().general.auth_query_user, None);
+        assert_eq!(get_config().general.auth_query_password, None);
     }
 
     #[tokio::test]

src/errors.rs

@@ -1,16 +1,120 @@
-/// Errors.
+//! Errors.
 
 /// Various errors.
 #[derive(Debug, PartialEq)]
 pub enum Error {
-    SocketError,
+    SocketError(String),
+    ClientSocketError(String, ClientIdentifier),
+    ClientGeneralError(String, ClientIdentifier),
+    ClientAuthImpossible(String),
+    ClientAuthPassthroughError(String, ClientIdentifier),
     ClientBadStartup,
-    ProtocolSyncError,
+    ProtocolSyncError(String),
+    BadQuery(String),
     ServerError,
+    ServerStartupError(String, ServerIdentifier),
+    ServerAuthError(String, ServerIdentifier),
     BadConfig,
     AllServersDown,
-    ClientError,
+    ClientError(String),
     TlsError,
     StatementTimeout,
     ShuttingDown,
+    ParseBytesError(String),
+    AuthError(String),
+    AuthPassthroughError(String),
+}
+
+#[derive(Clone, PartialEq, Debug)]
+pub struct ClientIdentifier {
+    pub application_name: String,
+    pub username: String,
+    pub pool_name: String,
+}
+
+impl ClientIdentifier {
+    pub fn new(application_name: &str, username: &str, pool_name: &str) -> ClientIdentifier {
+        ClientIdentifier {
+            application_name: application_name.into(),
+            username: username.into(),
+            pool_name: pool_name.into(),
+        }
+    }
+}
+
+impl std::fmt::Display for ClientIdentifier {
+    fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
+        write!(
+            f,
+            "{{ application_name: {}, username: {}, pool_name: {} }}",
+            self.application_name, self.username, self.pool_name
+        )
+    }
+}
+
+#[derive(Clone, PartialEq, Debug)]
+pub struct ServerIdentifier {
+    pub username: String,
+    pub database: String,
+}
+
+impl ServerIdentifier {
+    pub fn new(username: &str, database: &str) -> ServerIdentifier {
+        ServerIdentifier {
+            username: username.into(),
+            database: database.into(),
+        }
+    }
+}
+
+impl std::fmt::Display for ServerIdentifier {
+    fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
+        write!(
+            f,
+            "{{ username: {}, database: {} }}",
+            self.username, self.database
+        )
+    }
+}
+
+impl std::fmt::Display for Error {
+    fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
+        match &self {
+            &Error::ClientSocketError(error, client_identifier) => write!(
+                f,
+                "Error reading {} from client {}",
+                error, client_identifier
+            ),
+            &Error::ClientGeneralError(error, client_identifier) => {
+                write!(f, "{} {}", error, client_identifier)
+            }
+            &Error::ClientAuthImpossible(username) => write!(
+                f,
+                "Client auth not possible, \
+                no cleartext password set for username: {} \
+                in config and auth passthrough (query_auth) \
+                is not set up.",
+                username
+            ),
+            &Error::ClientAuthPassthroughError(error, client_identifier) => write!(
+                f,
+                "No cleartext password set, \
+                    and no auth passthrough could not \
+                    obtain the hash from server for {}, \
+                    the error was: {}",
+                client_identifier, error
+            ),
+            &Error::ServerStartupError(error, server_identifier) => write!(
+                f,
+                "Error reading {} on server startup {}",
+                error, server_identifier,
+            ),
+            &Error::ServerAuthError(error, server_identifier) => {
+                write!(f, "{} for {}", error, server_identifier,)
+            }
+
+            // The rest can use Debug.
+            err => write!(f, "{:?}", err),
+        }
+    }
 }

src/lib.rs

@@ -1,7 +1,10 @@
+pub mod auth_passthrough;
 pub mod config;
 pub mod constants;
 pub mod errors;
 pub mod messages;
+pub mod mirrors;
+pub mod multi_logger;
 pub mod pool;
 pub mod scram;
 pub mod server;

src/main.rs

@@ -37,14 +37,22 @@ extern crate tokio;
 extern crate tokio_rustls;
 extern crate toml;
 
-use log::{error, info, warn};
+#[cfg(not(target_env = "msvc"))]
+use jemallocator::Jemalloc;
+
+#[cfg(not(target_env = "msvc"))]
+#[global_allocator]
+static GLOBAL: Jemalloc = Jemalloc;
+
+use log::{debug, error, info, warn};
 use parking_lot::Mutex;
 use pgcat::format_duration;
 use tokio::net::TcpListener;
-use tokio::{
-    signal::unix::{signal as unix_signal, SignalKind},
-    sync::mpsc,
-};
+#[cfg(not(windows))]
+use tokio::signal::unix::{signal as unix_signal, SignalKind};
+#[cfg(windows)]
+use tokio::signal::windows as win_signal;
+use tokio::{runtime::Builder, sync::mpsc};
 
 use std::collections::HashMap;
 use std::net::SocketAddr;
@@ -53,11 +61,14 @@ use std::sync::Arc;
 use tokio::sync::broadcast;
 
 mod admin;
+mod auth_passthrough;
 mod client;
 mod config;
 mod constants;
 mod errors;
 mod messages;
+mod mirrors;
+mod multi_logger;
 mod pool;
 mod prometheus;
 mod query_router;
@@ -68,14 +79,13 @@ mod stats;
 mod tls;
 
 use crate::config::{get_config, reload_config, VERSION};
-use crate::errors::Error;
+use crate::messages::configure_socket;
 use crate::pool::{ClientServerMap, ConnectionPool};
 use crate::prometheus::start_metric_server;
 use crate::stats::{Collector, Reporter, REPORTER};
 
-#[tokio::main(worker_threads = 4)]
-async fn main() {
-    env_logger::builder().format_timestamp_micros().init();
+fn main() -> Result<(), Box<dyn std::error::Error>> {
+    multi_logger::MultiLogger::init().unwrap();
 
     info!("Welcome to PgCat! Meow. (Version {})", VERSION);
 
@@ -92,213 +102,255 @@ async fn main() {
         String::from("pgcat.toml")
     };
 
-    match config::parse(&config_file).await {
-        Ok(_) => (),
-        Err(err) => {
-            error!("Config parse error: {:?}", err);
-            std::process::exit(exitcode::CONFIG);
-        }
-    };
+    // Create a transient runtime for loading the config for the first time.
+    {
+        let runtime = Builder::new_multi_thread().worker_threads(1).build()?;
 
-    let config = get_config();
-
-    if let Some(true) = config.general.enable_prometheus_exporter {
-        let http_addr_str = format!(
-            "{}:{}",
-            config.general.host, config.general.prometheus_exporter_port
-        );
-        let http_addr = match SocketAddr::from_str(&http_addr_str) {
-            Ok(addr) => addr,
-            Err(err) => {
-                error!("Invalid http address: {}", err);
-                std::process::exit(exitcode::CONFIG);
-            }
-        };
-        tokio::task::spawn(async move {
-            start_metric_server(http_addr).await;
+        runtime.block_on(async {
+            match config::parse(&config_file).await {
+                Ok(_) => (),
+                Err(err) => {
+                    error!("Config parse error: {:?}", err);
+                    std::process::exit(exitcode::CONFIG);
+                }
+            };
         });
     }
 
-    let addr = format!("{}:{}", config.general.host, config.general.port);
+    let config = get_config();
 
-    let listener = match TcpListener::bind(&addr).await {
-        Ok(sock) => sock,
-        Err(err) => {
-            error!("Listener socket error: {:?}", err);
-            std::process::exit(exitcode::CONFIG);
+    // Create the runtime now we know required worker_threads.
+    let runtime = Builder::new_multi_thread()
+        .worker_threads(config.general.worker_threads)
+        .enable_all()
+        .build()?;
+
+    runtime.block_on(async move {
+
+        if let Some(true) = config.general.enable_prometheus_exporter {
+            let http_addr_str = format!(
+                "{}:{}",
+                config.general.host, config.general.prometheus_exporter_port
+            );
+
+            let http_addr = match SocketAddr::from_str(&http_addr_str) {
+                Ok(addr) => addr,
+                Err(err) => {
+                    error!("Invalid http address: {}", err);
+                    std::process::exit(exitcode::CONFIG);
+                }
+            };
+
+            tokio::task::spawn(async move {
+                start_metric_server(http_addr).await;
+            });
         }
-    };
 
-    info!("Running on {}", addr);
+        let addr = format!("{}:{}", config.general.host, config.general.port);
 
-    config.show();
+        let listener = match TcpListener::bind(&addr).await {
+            Ok(sock) => sock,
+            Err(err) => {
+                error!("Listener socket error: {:?}", err);
+                std::process::exit(exitcode::CONFIG);
+            }
+        };
 
-    // Tracks which client is connected to which server for query cancellation.
-    let client_server_map: ClientServerMap = Arc::new(Mutex::new(HashMap::new()));
+        info!("Running on {}", addr);
 
-    // Statistics reporting.
-    let (stats_tx, stats_rx) = mpsc::channel(100_000);
-    REPORTER.store(Arc::new(Reporter::new(stats_tx.clone())));
+        config.show();
 
-    // Connection pool that allows to query all shards and replicas.
-    match ConnectionPool::from_config(client_server_map.clone()).await {
-        Ok(_) => (),
-        Err(err) => {
-            error!("Pool error: {:?}", err);
-            std::process::exit(exitcode::CONFIG);
-        }
-    };
+        // Tracks which client is connected to which server for query cancellation.
+        let client_server_map: ClientServerMap = Arc::new(Mutex::new(HashMap::new()));
 
-    tokio::task::spawn(async move {
-        let mut stats_collector = Collector::new(stats_rx, stats_tx.clone());
-        stats_collector.collect().await;
-    });
+        // Statistics reporting.
+        REPORTER.store(Arc::new(Reporter::default()));
 
-    info!("Config autoreloader: {}", config.general.autoreload);
+        // Connection pool that allows to query all shards and replicas.
+        match ConnectionPool::from_config(client_server_map.clone()).await {
+            Ok(_) => (),
+            Err(err) => {
+                error!("Pool error: {:?}", err);
+                std::process::exit(exitcode::CONFIG);
+            }
+        };
 
-    let mut autoreload_interval = tokio::time::interval(tokio::time::Duration::from_millis(15_000));
-    let autoreload_client_server_map = client_server_map.clone();
-    tokio::task::spawn(async move {
-        loop {
-            autoreload_interval.tick().await;
-            if config.general.autoreload {
-                info!("Automatically reloading config");
+        tokio::task::spawn(async move {
+            let mut stats_collector = Collector::default();
+            stats_collector.collect().await;
+        });
 
-                match reload_config(autoreload_client_server_map.clone()).await {
-                    Ok(changed) => {
+        info!("Config autoreloader: {}", match config.general.autoreload {
+            Some(interval) => format!("{} ms", interval),
+            None => "disabled".into(),
+        });
+
+        if let Some(interval) = config.general.autoreload {
+            let mut autoreload_interval = tokio::time::interval(tokio::time::Duration::from_millis(interval));
+            let autoreload_client_server_map = client_server_map.clone();
+
+            tokio::task::spawn(async move {
+                loop {
+                    autoreload_interval.tick().await;
+                    debug!("Automatically reloading config");
+
+                    if let Ok(changed) = reload_config(autoreload_client_server_map.clone()).await {
                         if changed {
                             get_config().show()
                         }
-                    }
-                    Err(_) => (),
-                };
-            }
-        }
-    });
+                    };
+                }
+            });
+        };
 
-    let mut term_signal = unix_signal(SignalKind::terminate()).unwrap();
-    let mut interrupt_signal = unix_signal(SignalKind::interrupt()).unwrap();
-    let mut sighup_signal = unix_signal(SignalKind::hangup()).unwrap();
-    let (shutdown_tx, _) = broadcast::channel::<()>(1);
-    let (drain_tx, mut drain_rx) = mpsc::channel::<i32>(2048);
-    let (exit_tx, mut exit_rx) = mpsc::channel::<()>(1);
 
-    info!("Waiting for clients");
 
-    let mut admin_only = false;
-    let mut total_clients = 0;
+        #[cfg(windows)]
+        let mut term_signal = win_signal::ctrl_close().unwrap();
+        #[cfg(windows)]
+        let mut interrupt_signal = win_signal::ctrl_c().unwrap();
+        #[cfg(windows)]
+        let mut sighup_signal = win_signal::ctrl_shutdown().unwrap();
 
-    loop {
-        tokio::select! {
-            // Reload config:
-            // kill -SIGHUP $(pgrep pgcat)
-            _ = sighup_signal.recv() => {
-                info!("Reloading config");
+        #[cfg(not(windows))]
+        let mut term_signal = unix_signal(SignalKind::terminate()).unwrap();
+        #[cfg(not(windows))]
+        let mut interrupt_signal = unix_signal(SignalKind::interrupt()).unwrap();
+        #[cfg(not(windows))]
+        let mut sighup_signal = unix_signal(SignalKind::hangup()).unwrap();
+        let (shutdown_tx, _) = broadcast::channel::<()>(1);
+        let (drain_tx, mut drain_rx) = mpsc::channel::<i32>(2048);
+        let (exit_tx, mut exit_rx) = mpsc::channel::<()>(1);
+        let mut admin_only = false;
+        let mut total_clients = 0;
 
-                match reload_config(client_server_map.clone()).await {
-                    Ok(_) => (),
-                    Err(_) => (),
-                };
+        info!("Waiting for clients");
 
-                get_config().show();
-            },
+        loop {
+            tokio::select! {
+                // Reload config:
+                // kill -SIGHUP $(pgrep pgcat)
+                _ = sighup_signal.recv() => {
+                    info!("Reloading config");
 
-            // Initiate graceful shutdown sequence on sig int
-            _ = interrupt_signal.recv() => {
-                info!("Got SIGINT, waiting for client connection drain now");
-                admin_only = true;
+                    _ = reload_config(client_server_map.clone()).await;
 
-                // Broadcast that client tasks need to finish
-                let _ = shutdown_tx.send(());
-                let exit_tx = exit_tx.clone();
-                let _ = drain_tx.send(0).await;
+                    get_config().show();
+                },
 
-                tokio::task::spawn(async move {
-                    let mut interval = tokio::time::interval(tokio::time::Duration::from_millis(config.general.shutdown_timeout));
+                // Initiate graceful shutdown sequence on sig int
+                _ = interrupt_signal.recv() => {
+                    info!("Got SIGINT");
 
-                    // First tick fires immediately.
-                    interval.tick().await;
-
-                    // Second one in the interval time.
-                    interval.tick().await;
-
-                    // We're done waiting.
-                    error!("Graceful shutdown timed out. {} active clients being closed", total_clients);
-
-                    let _ = exit_tx.send(()).await;
-                });
-            },
-
-            _ = term_signal.recv() => {
-                info!("Got SIGTERM, closing with {} clients active", total_clients);
-                break;
-            },
-
-            new_client = listener.accept() => {
-                let (socket, addr) = match new_client {
-                    Ok((socket, addr)) => (socket, addr),
-                    Err(err) => {
-                        error!("{:?}", err);
+                    // Don't want this to happen more than once
+                    if admin_only {
                         continue;
                     }
-                };
 
-                let shutdown_rx = shutdown_tx.subscribe();
-                let drain_tx = drain_tx.clone();
-                let client_server_map = client_server_map.clone();
+                    admin_only = true;
 
-                tokio::task::spawn(async move {
-                    let start = chrono::offset::Utc::now().naive_utc();
+                    // Broadcast that client tasks need to finish
+                    let _ = shutdown_tx.send(());
+                    let exit_tx = exit_tx.clone();
+                    let _ = drain_tx.send(0).await;
 
-                    match client::client_entrypoint(
-                        socket,
-                        client_server_map,
-                        shutdown_rx,
-                        drain_tx,
-                        admin_only,
-                    )
-                    .await
-                    {
-                        Ok(()) => {
+                    tokio::task::spawn(async move {
+                        let mut interval = tokio::time::interval(tokio::time::Duration::from_millis(config.general.shutdown_timeout));
 
-                            let duration = chrono::offset::Utc::now().naive_utc() - start;
+                        // First tick fires immediately.
+                        interval.tick().await;
 
-                            info!(
-                                "Client {:?} disconnected, session duration: {}",
-                                addr,
-                                format_duration(&duration)
-                            );
-                        }
+                        // Second one in the interval time.
+                        interval.tick().await;
 
+                        // We're done waiting.
+                        error!("Graceful shutdown timed out. {} active clients being closed", total_clients);
+
+                        let _ = exit_tx.send(()).await;
+                    });
+                },
+
+                _ = term_signal.recv() => {
+                    info!("Got SIGTERM, closing with {} clients active", total_clients);
+                    break;
+                },
+
+                new_client = listener.accept() => {
+                    let (socket, addr) = match new_client {
+                        Ok((socket, addr)) => (socket, addr),
                         Err(err) => {
-                            match err {
-                                // Don't count the clients we rejected.
-                                Error::ShuttingDown => (),
-                                _ => {
-                                    // drain_tx.send(-1).await.unwrap();
+                            error!("{:?}", err);
+                            continue;
+                        }
+                    };
+
+                    let shutdown_rx = shutdown_tx.subscribe();
+                    let drain_tx = drain_tx.clone();
+                    let client_server_map = client_server_map.clone();
+
+                    let tls_certificate = get_config().general.tls_certificate.clone();
+
+                    configure_socket(&socket);
+
+                    tokio::task::spawn(async move {
+                        let start = chrono::offset::Utc::now().naive_utc();
+
+                        match client::client_entrypoint(
+                            socket,
+                            client_server_map,
+                            shutdown_rx,
+                            drain_tx,
+                            admin_only,
+                            tls_certificate,
+                            config.general.log_client_connections,
+                        )
+                        .await
+                        {
+                            Ok(()) => {
+                                let duration = chrono::offset::Utc::now().naive_utc() - start;
+
+                                if get_config().general.log_client_disconnections {
+                                    info!(
+                                        "Client {:?} disconnected, session duration: {}",
+                                        addr,
+                                        format_duration(&duration)
+                                    );
+                                } else {
+                                    debug!(
+                                        "Client {:?} disconnected, session duration: {}",
+                                        addr,
+                                        format_duration(&duration)
+                                    );
                                 }
                             }
 
-                            warn!("Client disconnected with error {:?}", err);
-                        }
-                    };
-                });
-            }
+                            Err(err) => {
+                                match err {
+                                    errors::Error::ClientBadStartup => debug!("Client disconnected with error {:?}", err),
+                                    _ => warn!("Client disconnected with error {:?}", err),
+                                }
 
-            _ = exit_rx.recv() => {
-                break;
-            }
+                            }
+                        };
+                    });
+                }
 
-            client_ping = drain_rx.recv() => {
-                let client_ping = client_ping.unwrap();
-                total_clients += client_ping;
+                _ = exit_rx.recv() => {
+                    break;
+                }
 
-                if total_clients == 0 && admin_only {
-                    let _ = exit_tx.send(()).await;
+                client_ping = drain_rx.recv() => {
+                    let client_ping = client_ping.unwrap();
+                    total_clients += client_ping;
+
+                    if total_clients == 0 && admin_only {
+                        let _ = exit_tx.send(()).await;
+                    }
                 }
             }
         }
-    }
 
     info!("Shutting down...");
+    });
+    Ok(())
 }

src/messages.rs

@@ -1,13 +1,18 @@
 /// Helper functions to send one-off protocol messages
 /// and handle TcpStream (TCP socket).
 use bytes::{Buf, BufMut, BytesMut};
+use log::error;
 use md5::{Digest, Md5};
+use socket2::{SockRef, TcpKeepalive};
 use tokio::io::{AsyncReadExt, AsyncWriteExt};
 use tokio::net::TcpStream;
 
+use crate::config::get_config;
 use crate::errors::Error;
 use std::collections::HashMap;
+use std::io::{BufRead, Cursor};
 use std::mem;
+use std::time::Duration;
 
 /// Postgres data type mappings
 /// used in RowDescription ('T') message.
@@ -38,7 +43,7 @@ where
     auth_ok.put_i32(8);
     auth_ok.put_i32(0);
 
-    Ok(write_all(stream, auth_ok).await?)
+    write_all(stream, auth_ok).await
 }
 
 /// Generate md5 password challenge.
@@ -79,7 +84,7 @@ where
     key_data.put_i32(backend_id);
     key_data.put_i32(secret_key);
 
-    Ok(write_all(stream, key_data).await?)
+    write_all(stream, key_data).await
 }
 
 /// Construct a `Q`: Query message.
@@ -88,7 +93,7 @@ pub fn simple_query(query: &str) -> BytesMut {
     let query = format!("{}\0", query);
 
     res.put_i32(query.len() as i32 + 4);
-    res.put_slice(&query.as_bytes());
+    res.put_slice(query.as_bytes());
 
     res
 }
@@ -106,24 +111,27 @@ where
     bytes.put_i32(5);
     bytes.put_u8(b'I'); // Idle
 
-    Ok(write_all(stream, bytes).await?)
+    write_all(stream, bytes).await
 }
 
 /// Send the startup packet the server. We're pretending we're a Pg client.
 /// This tells the server which user we are and what database we want.
-pub async fn startup(stream: &mut TcpStream, user: &str, database: &str) -> Result<(), Error> {
+pub async fn startup<S>(stream: &mut S, user: &str, database: &str) -> Result<(), Error>
+where
+    S: tokio::io::AsyncWrite + std::marker::Unpin,
+{
     let mut bytes = BytesMut::with_capacity(25);
 
     bytes.put_i32(196608); // Protocol number
 
     // User
     bytes.put(&b"user\0"[..]);
-    bytes.put_slice(&user.as_bytes());
+    bytes.put_slice(user.as_bytes());
     bytes.put_u8(0);
 
     // Database
     bytes.put(&b"database\0"[..]);
-    bytes.put_slice(&database.as_bytes());
+    bytes.put_slice(database.as_bytes());
     bytes.put_u8(0);
     bytes.put_u8(0); // Null terminator
 
@@ -136,7 +144,27 @@ pub async fn startup(stream: &mut TcpStream, user: &str, database: &str) -> Resu
 
     match stream.write_all(&startup).await {
         Ok(_) => Ok(()),
-        Err(_) => return Err(Error::SocketError),
+        Err(err) => {
+            return Err(Error::SocketError(format!(
+                "Error writing startup to server socket - Error: {:?}",
+                err
+            )))
+        }
+    }
+}
+
+pub async fn ssl_request(stream: &mut TcpStream) -> Result<(), Error> {
+    let mut bytes = BytesMut::with_capacity(12);
+
+    bytes.put_i32(8);
+    bytes.put_i32(80877103);
+
+    match stream.write_all(&bytes).await {
+        Ok(_) => Ok(()),
+        Err(err) => Err(Error::SocketError(format!(
+            "Error writing SSLRequest to server socket - Error: {:?}",
+            err
+        ))),
     }
 }
 
@@ -155,7 +183,7 @@ pub fn parse_params(mut bytes: BytesMut) -> Result<HashMap<String, String>, Erro
             c = bytes.get_u8();
         }
 
-        if tmp.len() > 0 {
+        if !tmp.is_empty() {
             buf.push(tmp.clone());
             tmp.clear();
         }
@@ -203,7 +231,13 @@ pub fn md5_hash_password(user: &str, password: &str, salt: &[u8]) -> Vec<u8> {
     let output = md5.finalize_reset();
 
     // Second pass
-    md5.update(format!("{:x}", output));
+    md5_hash_second_pass(&(format!("{:x}", output)), salt)
+}
+
+pub fn md5_hash_second_pass(hash: &str, salt: &[u8]) -> Vec<u8> {
+    let mut md5 = Md5::new();
+    // Second pass
+    md5.update(hash);
     md5.update(salt);
 
     let mut password = format!("md5{:x}", md5.finalize())
@@ -234,7 +268,21 @@ where
     message.put_i32(password.len() as i32 + 4);
     message.put_slice(&password[..]);
 
-    Ok(write_all(stream, message).await?)
+    write_all(stream, message).await
+}
+
+pub async fn md5_password_with_hash<S>(stream: &mut S, hash: &str, salt: &[u8]) -> Result<(), Error>
+where
+    S: tokio::io::AsyncWrite + std::marker::Unpin,
+{
+    let password = md5_hash_second_pass(hash, salt);
+    let mut message = BytesMut::with_capacity(password.len() as usize + 5);
+
+    message.put_u8(b'p');
+    message.put_i32(password.len() as i32 + 4);
+    message.put_slice(&password[..]);
+
+    write_all(stream, message).await
 }
 
 /// Implements a response to our custom `SET SHARDING KEY`
@@ -254,7 +302,7 @@ where
     res.put_i32(len);
     res.put_slice(&set_complete[..]);
 
-    write_all_half(stream, res).await?;
+    write_all_half(stream, &res).await?;
     ready_for_query(stream).await
 }
 
@@ -292,7 +340,7 @@ where
 
     // The short error message.
     error.put_u8(b'M');
-    error.put_slice(&format!("{}\0", message).as_bytes());
+    error.put_slice(format!("{}\0", message).as_bytes());
 
     // No more fields follow.
     error.put_u8(0);
@@ -304,7 +352,7 @@ where
     res.put_i32(error.len() as i32 + 4);
     res.put(error);
 
-    Ok(write_all_half(stream, res).await?)
+    write_all_half(stream, &res).await
 }
 
 pub async fn wrong_password<S>(stream: &mut S, user: &str) -> Result<(), Error>
@@ -327,7 +375,7 @@ where
 
     // The short error message.
     error.put_u8(b'M');
-    error.put_slice(&format!("password authentication failed for user \"{}\"\0", user).as_bytes());
+    error.put_slice(format!("password authentication failed for user \"{}\"\0", user).as_bytes());
 
     // No more fields follow.
     error.put_u8(0);
@@ -366,7 +414,7 @@ where
     // CommandComplete
     res.put(command_complete("SELECT 1"));
 
-    write_all_half(stream, res).await?;
+    write_all_half(stream, &res).await?;
     ready_for_query(stream).await
 }
 
@@ -374,12 +422,12 @@ pub fn row_description(columns: &Vec<(&str, DataType)>) -> BytesMut {
     let mut res = BytesMut::new();
     let mut row_desc = BytesMut::new();
 
-    // how many colums we are storing
+    // how many columns we are storing
     row_desc.put_i16(columns.len() as i16);
 
     for (name, data_type) in columns {
         // Column name
-        row_desc.put_slice(&format!("{}\0", name).as_bytes());
+        row_desc.put_slice(format!("{}\0", name).as_bytes());
 
         // Doesn't belong to any table
         row_desc.put_i32(0);
@@ -423,7 +471,7 @@ pub fn data_row(row: &Vec<String>) -> BytesMut {
     for column in row {
         let column = column.as_bytes();
         data_row.put_i32(column.len() as i32);
-        data_row.put_slice(&column);
+        data_row.put_slice(column);
     }
 
     res.put_u8(b'D');
@@ -450,18 +498,51 @@ where
 {
     match stream.write_all(&buf).await {
         Ok(_) => Ok(()),
-        Err(_) => return Err(Error::SocketError),
+        Err(err) => {
+            return Err(Error::SocketError(format!(
+                "Error writing to socket - Error: {:?}",
+                err
+            )))
+        }
     }
 }
 
 /// Write all the data in the buffer to the TcpStream, write owned half (see mpsc).
-pub async fn write_all_half<S>(stream: &mut S, buf: BytesMut) -> Result<(), Error>
+pub async fn write_all_half<S>(stream: &mut S, buf: &BytesMut) -> Result<(), Error>
 where
     S: tokio::io::AsyncWrite + std::marker::Unpin,
 {
-    match stream.write_all(&buf).await {
+    match stream.write_all(buf).await {
         Ok(_) => Ok(()),
-        Err(_) => return Err(Error::SocketError),
+        Err(err) => {
+            return Err(Error::SocketError(format!(
+                "Error writing to socket - Error: {:?}",
+                err
+            )))
+        }
+    }
+}
+
+pub async fn write_all_flush<S>(stream: &mut S, buf: &[u8]) -> Result<(), Error>
+where
+    S: tokio::io::AsyncWrite + std::marker::Unpin,
+{
+    match stream.write_all(buf).await {
+        Ok(_) => match stream.flush().await {
+            Ok(_) => Ok(()),
+            Err(err) => {
+                return Err(Error::SocketError(format!(
+                    "Error flushing socket - Error: {:?}",
+                    err
+                )))
+            }
+        },
+        Err(err) => {
+            return Err(Error::SocketError(format!(
+                "Error writing to socket - Error: {:?}",
+                err
+            )))
+        }
     }
 }
 
@@ -472,26 +553,51 @@ where
 {
     let code = match stream.read_u8().await {
         Ok(code) => code,
-        Err(_) => return Err(Error::SocketError),
+        Err(err) => {
+            return Err(Error::SocketError(format!(
+                "Error reading message code from socket - Error {:?}",
+                err
+            )))
+        }
     };
 
     let len = match stream.read_i32().await {
         Ok(len) => len,
-        Err(_) => return Err(Error::SocketError),
-    };
-
-    let mut buf = vec![0u8; len as usize - 4];
-
-    match stream.read_exact(&mut buf).await {
-        Ok(_) => (),
-        Err(_) => return Err(Error::SocketError),
+        Err(err) => {
+            return Err(Error::SocketError(format!(
+                "Error reading message len from socket - Code: {:?}, Error: {:?}",
+                code, err
+            )))
+        }
     };
 
     let mut bytes = BytesMut::with_capacity(len as usize + 1);
 
     bytes.put_u8(code);
     bytes.put_i32(len);
-    bytes.put_slice(&buf);
+
+    bytes.resize(bytes.len() + len as usize - mem::size_of::<i32>(), b'0');
+
+    let slice_start = mem::size_of::<u8>() + mem::size_of::<i32>();
+    let slice_end = slice_start + len as usize - mem::size_of::<i32>();
+
+    // Avoids a panic
+    if slice_end < slice_start {
+        return Err(Error::SocketError(format!(
+            "Error reading message from socket - Code: {:?} - Length {:?}, Error: {:?}",
+            code, len, "Unexpected length value for message"
+        )));
+    }
+
+    match stream.read_exact(&mut bytes[slice_start..slice_end]).await {
+        Ok(_) => (),
+        Err(err) => {
+            return Err(Error::SocketError(format!(
+                "Error reading message from socket - Code: {:?}, Error: {:?}",
+                code, err
+            )))
+        }
+    };
 
     Ok(bytes)
 }
@@ -510,5 +616,41 @@ pub fn server_parameter_message(key: &str, value: &str) -> BytesMut {
     server_info.put_slice(value.as_bytes());
     server_info.put_bytes(0, 1);
 
-    return server_info;
+    server_info
+}
+
+pub fn configure_socket(stream: &TcpStream) {
+    let sock_ref = SockRef::from(stream);
+    let conf = get_config();
+
+    match sock_ref.set_keepalive(true) {
+        Ok(_) => {
+            match sock_ref.set_tcp_keepalive(
+                &TcpKeepalive::new()
+                    .with_interval(Duration::from_secs(conf.general.tcp_keepalives_interval))
+                    .with_retries(conf.general.tcp_keepalives_count)
+                    .with_time(Duration::from_secs(conf.general.tcp_keepalives_idle)),
+            ) {
+                Ok(_) => (),
+                Err(err) => error!("Could not configure socket: {}", err),
+            }
+        }
+        Err(err) => error!("Could not configure socket: {}", err),
+    }
+}
+
+pub trait BytesMutReader {
+    fn read_string(&mut self) -> Result<String, Error>;
+}
+
+impl BytesMutReader for Cursor<&BytesMut> {
+    /// Should only be used when reading strings from the message protocol.
+    /// Can be used to read multiple strings from the same message which are separated by the null byte
+    fn read_string(&mut self) -> Result<String, Error> {
+        let mut buf = vec![];
+        match self.read_until(b'\0', &mut buf) {
+            Ok(_) => Ok(String::from_utf8_lossy(&buf[..buf.len() - 1]).to_string()),
+            Err(err) => return Err(Error::ParseBytesError(err.to_string())),
+        }
+    }
 }

src/mirrors.rs

@@ -0,0 +1,187 @@
+use std::sync::Arc;
+
+/// A mirrored PostgreSQL client.
+/// Packets arrive to us through a channel from the main client and we send them to the server.
+use bb8::Pool;
+use bytes::{Bytes, BytesMut};
+use parking_lot::RwLock;
+
+use crate::config::{get_config, Address, Role, User};
+use crate::pool::{ClientServerMap, PoolIdentifier, ServerPool};
+use crate::stats::PoolStats;
+use log::{error, info, trace, warn};
+use tokio::sync::mpsc::{channel, Receiver, Sender};
+
+pub struct MirroredClient {
+    address: Address,
+    user: User,
+    database: String,
+    bytes_rx: Receiver<Bytes>,
+    disconnect_rx: Receiver<()>,
+}
+
+impl MirroredClient {
+    async fn create_pool(&self) -> Pool<ServerPool> {
+        let config = get_config();
+        let default = std::time::Duration::from_millis(10_000).as_millis() as u64;
+        let (connection_timeout, idle_timeout, cfg) =
+            match config.pools.get(&self.address.pool_name) {
+                Some(cfg) => (
+                    cfg.connect_timeout.unwrap_or(default),
+                    cfg.idle_timeout.unwrap_or(default),
+                    cfg.clone(),
+                ),
+                None => (default, default, crate::config::Pool::default()),
+            };
+
+        let identifier = PoolIdentifier::new(&self.database, &self.user.username);
+
+        let manager = ServerPool::new(
+            self.address.clone(),
+            self.user.clone(),
+            self.database.as_str(),
+            ClientServerMap::default(),
+            Arc::new(PoolStats::new(identifier, cfg.clone())),
+            Arc::new(RwLock::new(None)),
+        );
+
+        Pool::builder()
+            .max_size(1)
+            .connection_timeout(std::time::Duration::from_millis(connection_timeout))
+            .idle_timeout(Some(std::time::Duration::from_millis(idle_timeout)))
+            .test_on_check_out(false)
+            .build(manager)
+            .await
+            .unwrap()
+    }
+
+    pub fn start(mut self) {
+        tokio::spawn(async move {
+            let pool = self.create_pool().await;
+            let address = self.address.clone();
+            loop {
+                let mut server = match pool.get().await {
+                    Ok(server) => server,
+                    Err(err) => {
+                        error!(
+                            "Failed to get connection from pool, Discarding message {:?}, {:?}",
+                            err,
+                            address.clone()
+                        );
+                        continue;
+                    }
+                };
+
+                tokio::select! {
+                    // Exit channel events
+                    _ = self.disconnect_rx.recv() => {
+                        info!("Got mirror exit signal, exiting {:?}", address.clone());
+                        break;
+                    }
+
+                    // Incoming data from server (we read to clear the socket buffer and discard the data)
+                    recv_result = server.recv() => {
+                        match recv_result {
+                            Ok(message) => trace!("Received from mirror: {} {:?}", String::from_utf8_lossy(&message[..]), address.clone()),
+                            Err(err) => {
+                                server.mark_bad();
+                                error!("Failed to receive from mirror {:?} {:?}", err, address.clone());
+                            }
+                        }
+                    }
+
+                    // Messages to send to the server
+                    message = self.bytes_rx.recv() => {
+                        match message {
+                            Some(bytes) => {
+                                match server.send(&BytesMut::from(&bytes[..])).await {
+                                    Ok(_) => trace!("Sent to mirror: {} {:?}", String::from_utf8_lossy(&bytes[..]), address.clone()),
+                                    Err(err) => {
+                                        server.mark_bad();
+                                        error!("Failed to send to mirror, Discarding message {:?}, {:?}", err, address.clone())
+                                    }
+                                }
+                            }
+                            None => {
+                                info!("Mirror channel closed, exiting {:?}", address.clone());
+                                break;
+                            },
+                        }
+                    }
+                }
+            }
+        });
+    }
+}
+pub struct MirroringManager {
+    pub byte_senders: Vec<Sender<Bytes>>,
+    pub disconnect_senders: Vec<Sender<()>>,
+}
+impl MirroringManager {
+    pub fn from_addresses(
+        user: User,
+        database: String,
+        addresses: Vec<Address>,
+    ) -> MirroringManager {
+        let mut byte_senders: Vec<Sender<Bytes>> = vec![];
+        let mut exit_senders: Vec<Sender<()>> = vec![];
+
+        addresses.iter().for_each(|mirror| {
+            let (bytes_tx, bytes_rx) = channel::<Bytes>(10);
+            let (exit_tx, exit_rx) = channel::<()>(1);
+            let mut addr = mirror.clone();
+            addr.role = Role::Mirror;
+            let client = MirroredClient {
+                user: user.clone(),
+                database: database.to_owned(),
+                address: addr,
+                bytes_rx,
+                disconnect_rx: exit_rx,
+            };
+            exit_senders.push(exit_tx.clone());
+            byte_senders.push(bytes_tx.clone());
+            client.start();
+        });
+
+        Self {
+            byte_senders: byte_senders,
+            disconnect_senders: exit_senders,
+        }
+    }
+
+    pub fn send(self: &mut Self, bytes: &BytesMut) {
+        // We want to avoid performing an allocation if we won't be able to send the message
+        // There is a possibility of a race here where we check the capacity and then the channel is
+        // closed or the capacity is reduced to 0, but mirroring is best effort anyway
+        if self
+            .byte_senders
+            .iter()
+            .all(|sender| sender.capacity() == 0 || sender.is_closed())
+        {
+            return;
+        }
+        let immutable_bytes = bytes.clone().freeze();
+        self.byte_senders.iter_mut().for_each(|sender| {
+            match sender.try_send(immutable_bytes.clone()) {
+                Ok(_) => {}
+                Err(err) => {
+                    warn!("Failed to send bytes to a mirror channel {}", err);
+                }
+            }
+        });
+    }
+
+    pub fn disconnect(self: &mut Self) {
+        self.disconnect_senders
+            .iter_mut()
+            .for_each(|sender| match sender.try_send(()) {
+                Ok(_) => {}
+                Err(err) => {
+                    warn!(
+                        "Failed to send disconnect signal to a mirror channel {}",
+                        err
+                    );
+                }
+            });
+    }
+}

src/multi_logger.rs

@@ -0,0 +1,80 @@
+use log::{Level, Log, Metadata, Record, SetLoggerError};
+
+// This is a special kind of logger that allows sending logs to different
+// targets depending on the log level.
+//
+// By default, if nothing is set, it acts as a regular env_log logger,
+// it sends everything to standard error.
+//
+// If the Env variable `STDOUT_LOG` is defined, it will be used for
+// configuring the standard out logger.
+//
+// The behavior is:
+//   - If it is an error, the message is written to standard error.
+//   - If it is not, and it matches the log level of the standard output logger (`STDOUT_LOG` env var), it will be send to standard output.
+//   - If the above is not true, it is sent to the stderr logger that will log it or not depending on the value
+//     of the RUST_LOG env var.
+//
+// So to summarize, if no `STDOUT_LOG` env var is present, the logger is the default logger. If `STDOUT_LOG` is set, everything
+// but errors, that matches the log level set in the `STDOUT_LOG` env var is sent to stdout. You can have also some esoteric configuration
+// where you set `RUST_LOG=debug` and `STDOUT_LOG=info`, in here, errors will go to stderr, warns and infos to stdout and debugs to stderr.
+//
+pub struct MultiLogger {
+    stderr_logger: env_logger::Logger,
+    stdout_logger: env_logger::Logger,
+}
+
+impl MultiLogger {
+    fn new() -> Self {
+        let stderr_logger = env_logger::builder().format_timestamp_micros().build();
+        let stdout_logger = env_logger::Builder::from_env("STDOUT_LOG")
+            .format_timestamp_micros()
+            .target(env_logger::Target::Stdout)
+            .build();
+
+        Self {
+            stderr_logger,
+            stdout_logger,
+        }
+    }
+
+    pub fn init() -> Result<(), SetLoggerError> {
+        let logger = Self::new();
+
+        log::set_max_level(logger.stderr_logger.filter());
+        log::set_boxed_logger(Box::new(logger))
+    }
+}
+
+impl Log for MultiLogger {
+    fn enabled(&self, metadata: &Metadata) -> bool {
+        self.stderr_logger.enabled(metadata) && self.stdout_logger.enabled(metadata)
+    }
+
+    fn log(&self, record: &Record) {
+        if record.level() == Level::Error {
+            self.stderr_logger.log(record);
+        } else {
+            if self.stdout_logger.matches(record) {
+                self.stdout_logger.log(record);
+            } else {
+                self.stderr_logger.log(record);
+            }
+        }
+    }
+
+    fn flush(&self) {
+        self.stderr_logger.flush();
+        self.stdout_logger.flush();
+    }
+}
+
+#[cfg(test)]
+mod test {
+    use super::*;
+
+    #[test]
+    fn test_init() {
+        MultiLogger::init().unwrap();
+    }
+}

src/prometheus.rs

@@ -5,10 +5,12 @@ use phf::phf_map;
 use std::collections::HashMap;
 use std::fmt;
 use std::net::SocketAddr;
+use std::sync::atomic::Ordering;
+use std::sync::Arc;
 
 use crate::config::Address;
 use crate::pool::get_all_pools;
-use crate::stats::get_address_stats;
+use crate::stats::{get_pool_stats, get_server_stats, ServerStats};
 
 struct MetricHelpType {
     help: &'static str,
@@ -19,109 +21,141 @@ struct MetricHelpType {
 // counters only increase
 // gauges can arbitrarily increase or decrease
 static METRIC_HELP_AND_TYPES_LOOKUP: phf::Map<&'static str, MetricHelpType> = phf_map! {
-    "total_query_count" => MetricHelpType {
+    "stats_total_query_count" => MetricHelpType {
         help: "Number of queries sent by all clients",
         ty: "counter",
     },
-    "total_query_time" => MetricHelpType {
+    "stats_total_query_time" => MetricHelpType {
         help: "Total amount of time for queries to execute",
         ty: "counter",
     },
-    "total_received" => MetricHelpType {
+    "stats_total_received" => MetricHelpType {
         help: "Number of bytes received from the server",
         ty: "counter",
     },
-    "total_sent" => MetricHelpType {
+    "stats_total_sent" => MetricHelpType {
         help: "Number of bytes sent to the server",
         ty: "counter",
     },
-    "total_xact_count" => MetricHelpType {
+    "stats_total_xact_count" => MetricHelpType {
         help: "Total number of transactions started by the client",
         ty: "counter",
     },
-    "total_xact_time" => MetricHelpType {
+    "stats_total_xact_time" => MetricHelpType {
         help: "Total amount of time for all transactions to execute",
         ty: "counter",
     },
-    "total_wait_time" => MetricHelpType {
+    "stats_total_wait_time" => MetricHelpType {
         help: "Total time client waited for a server connection",
         ty: "counter",
     },
-    "avg_query_count" => MetricHelpType {
+    "stats_avg_query_count" => MetricHelpType {
         help: "Average of total_query_count every 15 seconds",
         ty: "gauge",
     },
-    "avg_query_time" => MetricHelpType {
+    "stats_avg_query_time" => MetricHelpType {
         help: "Average time taken for queries to execute every 15 seconds",
         ty: "gauge",
     },
-    "avg_recv" => MetricHelpType {
+    "stats_avg_recv" => MetricHelpType {
         help: "Average of total_received bytes every 15 seconds",
         ty: "gauge",
     },
-    "avg_sent" => MetricHelpType {
+    "stats_avg_sent" => MetricHelpType {
         help: "Average of total_sent bytes every 15 seconds",
         ty: "gauge",
     },
-    "avg_xact_count" => MetricHelpType {
+    "stats_avg_errors" => MetricHelpType {
+        help: "Average number of errors every 15 seconds",
+        ty: "gauge",
+    },
+    "stats_avg_xact_count" => MetricHelpType {
         help: "Average of total_xact_count every 15 seconds",
         ty: "gauge",
     },
-    "avg_xact_time" => MetricHelpType {
+    "stats_avg_xact_time" => MetricHelpType {
         help: "Average of total_xact_time every 15 seconds",
         ty: "gauge",
     },
-    "avg_wait_time" => MetricHelpType {
+    "stats_avg_wait_time" => MetricHelpType {
         help: "Average of total_wait_time every 15 seconds",
         ty: "gauge",
     },
-    "maxwait_us" => MetricHelpType {
+    "pools_maxwait_us" => MetricHelpType {
         help: "The time a client waited for a server connection in microseconds",
         ty: "gauge",
     },
-    "maxwait" => MetricHelpType {
+    "pools_maxwait" => MetricHelpType {
         help: "The time a client waited for a server connection in seconds",
         ty: "gauge",
     },
-    "cl_waiting" => MetricHelpType {
+    "pools_cl_waiting" => MetricHelpType {
         help: "How many clients are waiting for a connection from the pool",
         ty: "gauge",
     },
-    "cl_active" => MetricHelpType {
+    "pools_cl_active" => MetricHelpType {
         help: "How many clients are actively communicating with a server",
         ty: "gauge",
     },
-    "cl_idle" => MetricHelpType {
+    "pools_cl_idle" => MetricHelpType {
         help: "How many clients are idle",
         ty: "gauge",
     },
-    "sv_idle" => MetricHelpType {
+    "pools_sv_idle" => MetricHelpType {
         help: "How many server connections are idle",
         ty: "gauge",
     },
-    "sv_active" => MetricHelpType {
+    "pools_sv_active" => MetricHelpType {
         help: "How many server connections are actively communicating with a client",
         ty: "gauge",
     },
-    "sv_login" => MetricHelpType {
+    "pools_sv_login" => MetricHelpType {
         help: "How many server connections are currently being created",
         ty: "gauge",
     },
-    "sv_tested" => MetricHelpType {
+    "pools_sv_tested" => MetricHelpType {
         help: "How many server connections are currently waiting on a health check to succeed",
         ty: "gauge",
     },
+    "servers_bytes_received" => MetricHelpType {
+        help: "Volume in bytes of network traffic received by server",
+        ty: "gauge",
+    },
+    "servers_bytes_sent" => MetricHelpType {
+        help: "Volume in bytes of network traffic sent by server",
+        ty: "gauge",
+    },
+    "servers_transaction_count" => MetricHelpType {
+        help: "Number of transactions executed by server",
+        ty: "gauge",
+    },
+    "servers_query_count" => MetricHelpType {
+        help: "Number of queries executed by server",
+        ty: "gauge",
+    },
+    "servers_error_count" => MetricHelpType {
+        help: "Number of errors",
+        ty: "gauge",
+    },
+    "databases_pool_size" => MetricHelpType {
+        help: "Maximum number of server connections",
+        ty: "gauge",
+    },
+    "databases_current_connections" => MetricHelpType {
+        help: "Current number of connections for this database",
+        ty: "gauge",
+    },
 };
 
-struct PrometheusMetric {
+struct PrometheusMetric<Value: fmt::Display> {
     name: String,
     help: String,
     ty: String,
     labels: HashMap<&'static str, String>,
-    value: i64,
+    value: Value,
 }
 
-impl fmt::Display for PrometheusMetric {
+impl<Value: fmt::Display> fmt::Display for PrometheusMetric<Value> {
     fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
         let formatted_labels = self
             .labels
@@ -141,50 +175,81 @@ impl fmt::Display for PrometheusMetric {
     }
 }
 
-impl PrometheusMetric {
-    fn new(address: &Address, name: &str, value: i64) -> Option<PrometheusMetric> {
+impl<Value: fmt::Display> PrometheusMetric<Value> {
+    fn from_name<V: fmt::Display>(
+        name: &str,
+        value: V,
+        labels: HashMap<&'static str, String>,
+    ) -> Option<PrometheusMetric<V>> {
+        METRIC_HELP_AND_TYPES_LOOKUP
+            .get(name)
+            .map(|metric| PrometheusMetric::<V> {
+                name: name.to_owned(),
+                help: metric.help.to_owned(),
+                ty: metric.ty.to_owned(),
+                value,
+                labels,
+            })
+    }
+
+    fn from_database_info(
+        address: &Address,
+        name: &str,
+        value: u32,
+    ) -> Option<PrometheusMetric<u32>> {
         let mut labels = HashMap::new();
         labels.insert("host", address.host.clone());
         labels.insert("shard", address.shard.to_string());
         labels.insert("role", address.role.to_string());
+        labels.insert("pool", address.pool_name.clone());
         labels.insert("database", address.database.to_string());
 
-        METRIC_HELP_AND_TYPES_LOOKUP
-            .get(name)
-            .map(|metric| PrometheusMetric {
-                name: name.to_owned(),
-                help: metric.help.to_owned(),
-                ty: metric.ty.to_owned(),
-                labels,
-                value,
-            })
+        Self::from_name(&format!("databases_{}", name), value, labels)
+    }
+
+    fn from_server_info(
+        address: &Address,
+        name: &str,
+        value: u64,
+    ) -> Option<PrometheusMetric<u64>> {
+        let mut labels = HashMap::new();
+        labels.insert("host", address.host.clone());
+        labels.insert("shard", address.shard.to_string());
+        labels.insert("role", address.role.to_string());
+        labels.insert("pool", address.pool_name.clone());
+        labels.insert("database", address.database.to_string());
+
+        Self::from_name(&format!("servers_{}", name), value, labels)
+    }
+
+    fn from_address(address: &Address, name: &str, value: u64) -> Option<PrometheusMetric<u64>> {
+        let mut labels = HashMap::new();
+        labels.insert("host", address.host.clone());
+        labels.insert("shard", address.shard.to_string());
+        labels.insert("pool", address.pool_name.clone());
+        labels.insert("role", address.role.to_string());
+        labels.insert("database", address.database.to_string());
+
+        Self::from_name(&format!("stats_{}", name), value, labels)
+    }
+
+    fn from_pool(pool: &(String, String), name: &str, value: u64) -> Option<PrometheusMetric<u64>> {
+        let mut labels = HashMap::new();
+        labels.insert("pool", pool.0.clone());
+        labels.insert("user", pool.1.clone());
+
+        Self::from_name(&format!("pools_{}", name), value, labels)
     }
 }
 
 async fn prometheus_stats(request: Request<Body>) -> Result<Response<Body>, hyper::http::Error> {
     match (request.method(), request.uri().path()) {
         (&Method::GET, "/metrics") => {
-            let stats: HashMap<usize, HashMap<String, i64>> = get_address_stats();
-
             let mut lines = Vec::new();
-            for (_, pool) in get_all_pools() {
-                for shard in 0..pool.shards() {
-                    for server in 0..pool.servers(shard) {
-                        let address = pool.address(shard, server);
-                        if let Some(address_stats) = stats.get(&address.id) {
-                            for (key, value) in address_stats.iter() {
-                                if let Some(prometheus_metric) =
-                                    PrometheusMetric::new(address, key, *value)
-                                {
-                                    lines.push(prometheus_metric.to_string());
-                                } else {
-                                    warn!("Metric {} not implemented for {}", key, address.name());
-                                }
-                            }
-                        }
-                    }
-                }
-            }
+            push_address_stats(&mut lines);
+            push_pool_stats(&mut lines);
+            push_server_stats(&mut lines);
+            push_database_stats(&mut lines);
 
             Response::builder()
                 .header("content-type", "text/plain; version=0.0.4")
@@ -196,10 +261,124 @@ async fn prometheus_stats(request: Request<Body>) -> Result<Response<Body>, hype
     }
 }
 
+// Adds metrics shown in a SHOW STATS admin command.
+fn push_address_stats(lines: &mut Vec<String>) {
+    for (_, pool) in get_all_pools() {
+        for shard in 0..pool.shards() {
+            for server in 0..pool.servers(shard) {
+                let address = pool.address(shard, server);
+                let stats = &*address.stats;
+                for (key, value) in stats.clone() {
+                    if let Some(prometheus_metric) =
+                        PrometheusMetric::<u64>::from_address(address, &key, value)
+                    {
+                        lines.push(prometheus_metric.to_string());
+                    } else {
+                        warn!("Metric {} not implemented for {}", key, address.name());
+                    }
+                }
+            }
+        }
+    }
+}
+
+// Adds relevant metrics shown in a SHOW POOLS admin command.
+fn push_pool_stats(lines: &mut Vec<String>) {
+    let pool_stats = get_pool_stats();
+    for (pool, stats) in pool_stats.iter() {
+        let stats = &**stats;
+        for (name, value) in stats.clone() {
+            if let Some(prometheus_metric) = PrometheusMetric::<u64>::from_pool(pool, &name, value)
+            {
+                lines.push(prometheus_metric.to_string());
+            } else {
+                warn!(
+                    "Metric {} not implemented for ({},{})",
+                    name, pool.0, pool.1
+                );
+            }
+        }
+    }
+}
+
+// Adds relevant metrics shown in a SHOW DATABASES admin command.
+fn push_database_stats(lines: &mut Vec<String>) {
+    for (_, pool) in get_all_pools() {
+        let pool_config = pool.settings.clone();
+        for shard in 0..pool.shards() {
+            for server in 0..pool.servers(shard) {
+                let address = pool.address(shard, server);
+                let pool_state = pool.pool_state(shard, server);
+
+                let metrics = vec![
+                    ("pool_size", pool_config.user.pool_size),
+                    ("current_connections", pool_state.connections),
+                ];
+                for (key, value) in metrics {
+                    if let Some(prometheus_metric) =
+                        PrometheusMetric::<u32>::from_database_info(address, key, value)
+                    {
+                        lines.push(prometheus_metric.to_string());
+                    } else {
+                        warn!("Metric {} not implemented for {}", key, address.name());
+                    }
+                }
+            }
+        }
+    }
+}
+
+// Adds relevant metrics shown in a SHOW SERVERS admin command.
+fn push_server_stats(lines: &mut Vec<String>) {
+    let server_stats = get_server_stats();
+    let mut server_stats_by_addresses = HashMap::<String, Arc<ServerStats>>::new();
+    for (_, stats) in server_stats {
+        server_stats_by_addresses.insert(stats.address_name(), stats);
+    }
+
+    for (_, pool) in get_all_pools() {
+        for shard in 0..pool.shards() {
+            for server in 0..pool.servers(shard) {
+                let address = pool.address(shard, server);
+                if let Some(server_info) = server_stats_by_addresses.get(&address.name()) {
+                    let metrics = [
+                        (
+                            "bytes_received",
+                            server_info.bytes_received.load(Ordering::Relaxed),
+                        ),
+                        ("bytes_sent", server_info.bytes_sent.load(Ordering::Relaxed)),
+                        (
+                            "transaction_count",
+                            server_info.transaction_count.load(Ordering::Relaxed),
+                        ),
+                        (
+                            "query_count",
+                            server_info.query_count.load(Ordering::Relaxed),
+                        ),
+                        (
+                            "error_count",
+                            server_info.error_count.load(Ordering::Relaxed),
+                        ),
+                    ];
+                    for (key, value) in metrics {
+                        if let Some(prometheus_metric) =
+                            PrometheusMetric::<u64>::from_server_info(address, key, value)
+                        {
+                            lines.push(prometheus_metric.to_string());
+                        } else {
+                            warn!("Metric {} not implemented for {}", key, address.name());
+                        }
+                    }
+                }
+            }
+        }
+    }
+}
+
 pub async fn start_metric_server(http_addr: SocketAddr) {
     let http_service_factory =
         make_service_fn(|_conn| async { Ok::<_, hyper::Error>(service_fn(prometheus_stats)) });
-    let server = Server::bind(&http_addr.into()).serve(http_service_factory);
+    let server = Server::bind(&http_addr).serve(http_service_factory);
     info!(
         "Exposing prometheus metrics on http://{}/metrics.",
         http_addr

src/scram.rs

@@ -2,6 +2,7 @@
 // https://github.com/sfackler/rust-postgres/
 // SASL implementation.
 
+use base64::{engine::general_purpose, Engine as _};
 use bytes::BytesMut;
 use hmac::{Hmac, Mac};
 use rand::{self, Rng};
@@ -57,7 +58,7 @@ impl ScramSha256 {
 
     /// Used for testing.
     pub fn from_nonce(password: &str, nonce: &str) -> ScramSha256 {
-        let message = BytesMut::from(&format!("{}n=,r={}", "n,,", nonce).as_bytes()[..]);
+        let message = BytesMut::from(format!("{}n=,r={}", "n,,", nonce).as_bytes());
 
         ScramSha256 {
             password: password.to_string(),
@@ -78,16 +79,16 @@ impl ScramSha256 {
         let server_message = Message::parse(message)?;
 
         if !server_message.nonce.starts_with(&self.nonce) {
-            return Err(Error::ProtocolSyncError);
+            return Err(Error::ProtocolSyncError(format!("SCRAM")));
         }
 
-        let salt = match base64::decode(&server_message.salt) {
+        let salt = match general_purpose::STANDARD.decode(&server_message.salt) {
             Ok(salt) => salt,
-            Err(_) => return Err(Error::ProtocolSyncError),
+            Err(_) => return Err(Error::ProtocolSyncError(format!("SCRAM"))),
         };
 
         let salted_password = Self::hi(
-            &normalize(&self.password.as_bytes()[..]),
+            &normalize(self.password.as_bytes()),
             &salt,
             server_message.iterations,
         );
@@ -111,7 +112,7 @@ impl ScramSha256 {
         let mut cbind_input = vec![];
         cbind_input.extend("n,,".as_bytes());
 
-        let cbind_input = base64::encode(&cbind_input);
+        let cbind_input = general_purpose::STANDARD.encode(&cbind_input);
 
         self.message.clear();
 
@@ -149,7 +150,11 @@ impl ScramSha256 {
             *proof ^= signature;
         }
 
-        match write!(&mut self.message, ",p={}", base64::encode(&*client_proof)) {
+        match write!(
+            &mut self.message,
+            ",p={}",
+            general_purpose::STANDARD.encode(&*client_proof)
+        ) {
             Ok(_) => (),
             Err(_) => return Err(Error::ServerError),
         };
@@ -161,9 +166,9 @@ impl ScramSha256 {
     pub fn finish(&mut self, message: &BytesMut) -> Result<(), Error> {
         let final_message = FinalMessage::parse(message)?;
 
-        let verifier = match base64::decode(&final_message.value) {
+        let verifier = match general_purpose::STANDARD.decode(&final_message.value) {
             Ok(verifier) => verifier,
-            Err(_) => return Err(Error::ProtocolSyncError),
+            Err(_) => return Err(Error::ProtocolSyncError(format!("SCRAM"))),
         };
 
         let mut hmac = match Hmac::<Sha256>::new_from_slice(&self.salted_password) {
@@ -181,7 +186,7 @@ impl ScramSha256 {
 
         match hmac.verify_slice(&verifier) {
             Ok(_) => Ok(()),
-            Err(_) => return Err(Error::ServerError),
+            Err(_) => Err(Error::ServerError),
         }
     }
 
@@ -220,19 +225,19 @@ impl Message {
     /// Parse the server SASL challenge.
     fn parse(message: &BytesMut) -> Result<Message, Error> {
         let parts = String::from_utf8_lossy(&message[..])
-            .split(",")
+            .split(',')
             .map(|s| s.to_string())
             .collect::<Vec<String>>();
 
         if parts.len() != 3 {
-            return Err(Error::ProtocolSyncError);
+            return Err(Error::ProtocolSyncError(format!("SCRAM")));
         }
 
         let nonce = str::replace(&parts[0], "r=", "");
         let salt = str::replace(&parts[1], "s=", "");
         let iterations = match str::replace(&parts[2], "i=", "").parse::<u32>() {
             Ok(iterations) => iterations,
-            Err(_) => return Err(Error::ProtocolSyncError),
+            Err(_) => return Err(Error::ProtocolSyncError(format!("SCRAM"))),
         };
 
         Ok(Message {
@@ -252,7 +257,7 @@ impl FinalMessage {
     /// Parse the server final validation message.
     pub fn parse(message: &BytesMut) -> Result<FinalMessage, Error> {
         if !message.starts_with(b"v=") || message.len() < 4 {
-            return Err(Error::ProtocolSyncError);
+            return Err(Error::ProtocolSyncError(format!("SCRAM")));
         }
 
         Ok(FinalMessage {
@@ -268,7 +273,7 @@ mod test {
     #[test]
     fn parse_server_first_message() {
         let message = BytesMut::from(
-            &"r=fyko+d2lbbFgONRv9qkxdawL3rfcNHYJY1ZVvWVs7j,s=QSXCR+Q6sek8bf92,i=4096".as_bytes()[..],
+            "r=fyko+d2lbbFgONRv9qkxdawL3rfcNHYJY1ZVvWVs7j,s=QSXCR+Q6sek8bf92,i=4096".as_bytes(),
         );
         let message = Message::parse(&message).unwrap();
         assert_eq!(message.nonce, "fyko+d2lbbFgONRv9qkxdawL3rfcNHYJY1ZVvWVs7j");
@@ -279,7 +284,7 @@ mod test {
     #[test]
     fn parse_server_last_message() {
         let f = FinalMessage::parse(&BytesMut::from(
-            &"v=U+ppxD5XUKtradnv8e2MkeupiA8FU87Sg8CXzXHDAzw".as_bytes()[..],
+            "v=U+ppxD5XUKtradnv8e2MkeupiA8FU87Sg8CXzXHDAzw".as_bytes(),
         ))
         .unwrap();
         assert_eq!(
@@ -309,12 +314,12 @@ mod test {
         assert_eq!(std::str::from_utf8(&message).unwrap(), client_first);
 
         let result = scram
-            .update(&BytesMut::from(&server_first.as_bytes()[..]))
+            .update(&BytesMut::from(server_first.as_bytes()))
             .unwrap();
         assert_eq!(std::str::from_utf8(&result).unwrap(), client_final);
 
         scram
-            .finish(&BytesMut::from(&server_final.as_bytes()[..]))
+            .finish(&BytesMut::from(server_final.as_bytes()))
             .unwrap();
     }
 }

src/server.rs

@@ -1,36 +1,114 @@
 /// Implementation of the PostgreSQL server (database) protocol.
 /// Here we are pretending to the a Postgres client.
 use bytes::{Buf, BufMut, BytesMut};
+use fallible_iterator::FallibleIterator;
 use log::{debug, error, info, trace, warn};
+use parking_lot::{Mutex, RwLock};
+use postgres_protocol::message;
+use std::collections::HashMap;
 use std::io::Read;
+use std::sync::Arc;
 use std::time::SystemTime;
-use tokio::io::{AsyncReadExt, BufReader};
-use tokio::net::{
-    tcp::{OwnedReadHalf, OwnedWriteHalf},
-    TcpStream,
-};
+use tokio::io::{AsyncRead, AsyncReadExt, AsyncWrite, BufStream};
+use tokio::net::TcpStream;
+use tokio_rustls::rustls::{OwnedTrustAnchor, RootCertStore};
+use tokio_rustls::{client::TlsStream, TlsConnector};
 
-use crate::config::{Address, User};
+use crate::config::{get_config, Address, User};
 use crate::constants::*;
-use crate::errors::Error;
+use crate::errors::{Error, ServerIdentifier};
 use crate::messages::*;
+use crate::mirrors::MirroringManager;
 use crate::pool::ClientServerMap;
 use crate::scram::ScramSha256;
-use crate::stats::Reporter;
+use crate::stats::ServerStats;
+use std::io::Write;
+
+use pin_project::pin_project;
+
+#[pin_project(project = SteamInnerProj)]
+pub enum StreamInner {
+    Plain {
+        #[pin]
+        stream: TcpStream,
+    },
+    Tls {
+        #[pin]
+        stream: TlsStream<TcpStream>,
+    },
+}
+
+impl AsyncWrite for StreamInner {
+    fn poll_write(
+        self: std::pin::Pin<&mut Self>,
+        cx: &mut std::task::Context<'_>,
+        buf: &[u8],
+    ) -> std::task::Poll<Result<usize, std::io::Error>> {
+        let this = self.project();
+        match this {
+            SteamInnerProj::Tls { stream } => stream.poll_write(cx, buf),
+            SteamInnerProj::Plain { stream } => stream.poll_write(cx, buf),
+        }
+    }
+
+    fn poll_flush(
+        self: std::pin::Pin<&mut Self>,
+        cx: &mut std::task::Context<'_>,
+    ) -> std::task::Poll<Result<(), std::io::Error>> {
+        let this = self.project();
+        match this {
+            SteamInnerProj::Tls { stream } => stream.poll_flush(cx),
+            SteamInnerProj::Plain { stream } => stream.poll_flush(cx),
+        }
+    }
+
+    fn poll_shutdown(
+        self: std::pin::Pin<&mut Self>,
+        cx: &mut std::task::Context<'_>,
+    ) -> std::task::Poll<Result<(), std::io::Error>> {
+        let this = self.project();
+        match this {
+            SteamInnerProj::Tls { stream } => stream.poll_shutdown(cx),
+            SteamInnerProj::Plain { stream } => stream.poll_shutdown(cx),
+        }
+    }
+}
+
+impl AsyncRead for StreamInner {
+    fn poll_read(
+        self: std::pin::Pin<&mut Self>,
+        cx: &mut std::task::Context<'_>,
+        buf: &mut tokio::io::ReadBuf<'_>,
+    ) -> std::task::Poll<std::io::Result<()>> {
+        let this = self.project();
+        match this {
+            SteamInnerProj::Tls { stream } => stream.poll_read(cx, buf),
+            SteamInnerProj::Plain { stream } => stream.poll_read(cx, buf),
+        }
+    }
+}
+
+impl StreamInner {
+    pub fn try_write(&mut self, buf: &[u8]) -> std::io::Result<usize> {
+        match self {
+            StreamInner::Tls { stream } => {
+                let r = stream.get_mut();
+                let mut w = r.1.writer();
+                w.write(buf)
+            }
+            StreamInner::Plain { stream } => stream.try_write(buf),
+        }
+    }
+}
 
 /// Server state.
 pub struct Server {
-    server_id: i32,
-
     /// Server host, e.g. localhost,
     /// port, e.g. 5432, and role, e.g. primary or replica.
     address: Address,
 
-    /// Buffered read socket.
-    read: BufReader<OwnedReadHalf>,
-
-    /// Unbuffered write socket (our client code buffers).
-    write: OwnedWriteHalf,
+    /// Server TCP connection.
+    stream: BufStream<StreamInner>,
 
     /// Our server response buffer. We buffer data before we give it to the client.
     buffer: BytesMut,
@@ -61,57 +139,170 @@ pub struct Server {
     connected_at: chrono::naive::NaiveDateTime,
 
     /// Reports various metrics, e.g. data sent & received.
-    stats: Reporter,
+    stats: Arc<ServerStats>,
 
     /// Application name using the server at the moment.
     application_name: String,
 
     // Last time that a successful server send or response happened
     last_activity: SystemTime,
+
+    mirror_manager: Option<MirroringManager>,
 }
 
 impl Server {
     /// Pretend to be the Postgres client and connect to the server given host, port and credentials.
     /// Perform the authentication and return the server in a ready for query state.
     pub async fn startup(
-        server_id: i32,
         address: &Address,
         user: &User,
         database: &str,
         client_server_map: ClientServerMap,
-        stats: Reporter,
+        stats: Arc<ServerStats>,
+        auth_hash: Arc<RwLock<Option<String>>>,
     ) -> Result<Server, Error> {
         let mut stream =
             match TcpStream::connect(&format!("{}:{}", &address.host, address.port)).await {
                 Ok(stream) => stream,
                 Err(err) => {
                     error!("Could not connect to server: {}", err);
-                    return Err(Error::SocketError);
+                    return Err(Error::SocketError(format!(
+                        "Could not connect to server: {}",
+                        err
+                    )));
                 }
             };
 
+        // TCP timeouts.
+        configure_socket(&stream);
+
+        let config = get_config();
+
+        let mut stream = if config.general.server_tls {
+            // Request a TLS connection
+            ssl_request(&mut stream).await?;
+
+            let response = match stream.read_u8().await {
+                Ok(response) => response as char,
+                Err(err) => {
+                    return Err(Error::SocketError(format!(
+                        "Server socket error: {:?}",
+                        err
+                    )))
+                }
+            };
+
+            match response {
+                // Server supports TLS
+                'S' => {
+                    debug!("Connecting to server using TLS");
+
+                    let mut root_store = RootCertStore::empty();
+                    root_store.add_server_trust_anchors(
+                        webpki_roots::TLS_SERVER_ROOTS.0.iter().map(|ta| {
+                            OwnedTrustAnchor::from_subject_spki_name_constraints(
+                                ta.subject,
+                                ta.spki,
+                                ta.name_constraints,
+                            )
+                        }),
+                    );
+
+                    let mut tls_config = rustls::ClientConfig::builder()
+                        .with_safe_defaults()
+                        .with_root_certificates(root_store)
+                        .with_no_client_auth();
+
+                    // Equivalent to sslmode=prefer which is fine most places.
+                    // If you want verify-full, change `verify_server_certificate` to true.
+                    if !config.general.verify_server_certificate {
+                        let mut dangerous = tls_config.dangerous();
+                        dangerous.set_certificate_verifier(Arc::new(
+                            crate::tls::NoCertificateVerification {},
+                        ));
+                    }
+
+                    let connector = TlsConnector::from(Arc::new(tls_config));
+                    let stream = match connector
+                        .connect(address.host.as_str().try_into().unwrap(), stream)
+                        .await
+                    {
+                        Ok(stream) => stream,
+                        Err(err) => {
+                            return Err(Error::SocketError(format!("Server TLS error: {:?}", err)))
+                        }
+                    };
+
+                    StreamInner::Tls { stream }
+                }
+
+                // Server does not support TLS
+                'N' => StreamInner::Plain { stream },
+
+                // Something else?
+                m => {
+                    return Err(Error::SocketError(format!(
+                        "Unknown message: {}",
+                        m as char
+                    )));
+                }
+            }
+        } else {
+            StreamInner::Plain { stream }
+        };
+
+        // let (read, write) = split(stream);
+        // let (mut read, mut write) = (ReadInner::Plain { stream: read }, WriteInner::Plain { stream: write });
+
         trace!("Sending StartupMessage");
 
         // StartupMessage
-        startup(&mut stream, &user.username, database).await?;
+        let username = match user.server_username {
+            Some(ref server_username) => server_username,
+            None => &user.username,
+        };
+
+        let password = match user.server_password {
+            Some(ref server_password) => Some(server_password),
+            None => match user.password {
+                Some(ref password) => Some(password),
+                None => None,
+            },
+        };
+
+        startup(&mut stream, username, database).await?;
 
         let mut server_info = BytesMut::new();
         let mut process_id: i32 = 0;
         let mut secret_key: i32 = 0;
+        let server_identifier = ServerIdentifier::new(username, &database);
 
         // We'll be handling multiple packets, but they will all be structured the same.
         // We'll loop here until this exchange is complete.
-        let mut scram = ScramSha256::new(&user.password);
+        let mut scram: Option<ScramSha256> = match password {
+            Some(password) => Some(ScramSha256::new(password)),
+            None => None,
+        };
 
         loop {
             let code = match stream.read_u8().await {
                 Ok(code) => code as char,
-                Err(_) => return Err(Error::SocketError),
+                Err(_) => {
+                    return Err(Error::ServerStartupError(
+                        "message code".into(),
+                        server_identifier,
+                    ))
+                }
             };
 
             let len = match stream.read_i32().await {
                 Ok(len) => len,
-                Err(_) => return Err(Error::SocketError),
+                Err(_) => {
+                    return Err(Error::ServerStartupError(
+                        "message len".into(),
+                        server_identifier,
+                    ))
+                }
             };
 
             trace!("Message: {}", code);
@@ -122,7 +313,12 @@ impl Server {
                     // Determine which kind of authentication is required, if any.
                     let auth_code = match stream.read_i32().await {
                         Ok(auth_code) => auth_code,
-                        Err(_) => return Err(Error::SocketError),
+                        Err(_) => {
+                            return Err(Error::ServerStartupError(
+                                "auth code".into(),
+                                server_identifier,
+                            ))
+                        }
                     };
 
                     trace!("Auth: {}", auth_code);
@@ -135,32 +331,79 @@ impl Server {
 
                             match stream.read_exact(&mut salt).await {
                                 Ok(_) => (),
-                                Err(_) => return Err(Error::SocketError),
+                                Err(_) => {
+                                    return Err(Error::ServerStartupError(
+                                        "salt".into(),
+                                        server_identifier,
+                                    ))
+                                }
                             };
 
-                            md5_password(&mut stream, &user.username, &user.password, &salt[..])
-                                .await?;
+                            match password {
+                                // Using plaintext password
+                                Some(password) => {
+                                    md5_password(&mut stream, username, password, &salt[..]).await?
+                                }
+
+                                // Using auth passthrough, in this case we should already have a
+                                // hash obtained when the pool was validated. If we reach this point
+                                // and don't have a hash, we return an error.
+                                None => {
+                                    let option_hash = (*auth_hash.read()).clone();
+                                    match option_hash {
+                                        Some(hash) =>
+                                            md5_password_with_hash(
+                                                &mut stream,
+                                                &hash,
+                                                &salt[..],
+                                            )
+                                            .await?,
+                                        None => return Err(
+                                            Error::ServerAuthError(
+                                                "Auth passthrough (auth_query) failed and no user password is set in cleartext".into(),
+                                                server_identifier
+                                            )
+                                        ),
+                                    }
+                                }
+                            }
                         }
 
                         AUTHENTICATION_SUCCESSFUL => (),
 
                         SASL => {
+                            if scram.is_none() {
+                                return Err(Error::ServerAuthError(
+                                    "SASL auth required and no password specified. \
+                                    Auth passthrough (auth_query) method is currently \
+                                    unsupported for SASL auth"
+                                        .into(),
+                                    server_identifier,
+                                ));
+                            }
+
                             debug!("Starting SASL authentication");
+
                             let sasl_len = (len - 8) as usize;
                             let mut sasl_auth = vec![0u8; sasl_len];
 
                             match stream.read_exact(&mut sasl_auth).await {
                                 Ok(_) => (),
-                                Err(_) => return Err(Error::SocketError),
+                                Err(_) => {
+                                    return Err(Error::ServerStartupError(
+                                        "sasl message".into(),
+                                        server_identifier,
+                                    ))
+                                }
                             };
 
                             let sasl_type = String::from_utf8_lossy(&sasl_auth[..sasl_len - 2]);
 
-                            if sasl_type == SCRAM_SHA_256 {
+                            if sasl_type.contains(SCRAM_SHA_256) {
                                 debug!("Using {}", SCRAM_SHA_256);
 
                                 // Generate client message.
-                                let sasl_response = scram.message();
+                                let sasl_response = scram.as_mut().unwrap().message();
 
                                 // SASLInitialResponse (F)
                                 let mut res = BytesMut::new();
@@ -175,11 +418,11 @@ impl Server {
                                         + sasl_response.len() as i32, // length of SASL response
                                 );
 
-                                res.put_slice(&format!("{}\0", SCRAM_SHA_256).as_bytes()[..]);
+                                res.put_slice(format!("{}\0", SCRAM_SHA_256).as_bytes());
                                 res.put_i32(sasl_response.len() as i32);
                                 res.put(sasl_response);
 
-                                write_all(&mut stream, res).await?;
+                                write_all_flush(&mut stream, &res).await?;
                             } else {
                                 error!("Unsupported SCRAM version: {}", sasl_type);
                                 return Err(Error::ServerError);
@@ -193,11 +436,16 @@ impl Server {
 
                             match stream.read_exact(&mut sasl_data).await {
                                 Ok(_) => (),
-                                Err(_) => return Err(Error::SocketError),
+                                Err(_) => {
+                                    return Err(Error::ServerStartupError(
+                                        "sasl cont message".into(),
+                                        server_identifier,
+                                    ))
+                                }
                             };
 
                             let msg = BytesMut::from(&sasl_data[..]);
-                            let sasl_response = scram.update(&msg)?;
+                            let sasl_response = scram.as_mut().unwrap().update(&msg)?;
 
                             // SASLResponse
                             let mut res = BytesMut::new();
@@ -205,7 +453,7 @@ impl Server {
                             res.put_i32(4 + sasl_response.len() as i32);
                             res.put(sasl_response);
 
-                            write_all(&mut stream, res).await?;
+                            write_all_flush(&mut stream, &res).await?;
                         }
 
                         SASL_FINAL => {
@@ -214,10 +462,19 @@ impl Server {
                             let mut sasl_final = vec![0u8; len as usize - 8];
                             match stream.read_exact(&mut sasl_final).await {
                                 Ok(_) => (),
-                                Err(_) => return Err(Error::SocketError),
+                                Err(_) => {
+                                    return Err(Error::ServerStartupError(
+                                        "sasl final message".into(),
+                                        server_identifier,
+                                    ))
+                                }
                             };
 
-                            match scram.finish(&BytesMut::from(&sasl_final[..])) {
+                            match scram
+                                .as_mut()
+                                .unwrap()
+                                .finish(&BytesMut::from(&sasl_final[..]))
+                            {
                                 Ok(_) => {
                                     debug!("SASL authentication successful");
                                 }
@@ -240,7 +497,12 @@ impl Server {
                 'E' => {
                     let error_code = match stream.read_u8().await {
                         Ok(error_code) => error_code,
-                        Err(_) => return Err(Error::SocketError),
+                        Err(_) => {
+                            return Err(Error::ServerStartupError(
+                                "error code message".into(),
+                                server_identifier,
+                            ))
+                        }
                     };
 
                     trace!("Error: {}", error_code);
@@ -256,7 +518,12 @@ impl Server {
 
                             match stream.read_exact(&mut error).await {
                                 Ok(_) => (),
-                                Err(_) => return Err(Error::SocketError),
+                                Err(_) => {
+                                    return Err(Error::ServerStartupError(
+                                        "error message".into(),
+                                        server_identifier,
+                                    ))
+                                }
                             };
 
                             // TODO: the error message contains multiple fields; we can decode them and
@@ -275,7 +542,12 @@ impl Server {
 
                     match stream.read_exact(&mut param).await {
                         Ok(_) => (),
-                        Err(_) => return Err(Error::SocketError),
+                        Err(_) => {
+                            return Err(Error::ServerStartupError(
+                                "parameter status message".into(),
+                                server_identifier,
+                            ))
+                        }
                     };
 
                     // Save the parameter so we can pass it to the client later.
@@ -292,12 +564,22 @@ impl Server {
                     // See: <https://www.postgresql.org/docs/12/protocol-message-formats.html>.
                     process_id = match stream.read_i32().await {
                         Ok(id) => id,
-                        Err(_) => return Err(Error::SocketError),
+                        Err(_) => {
+                            return Err(Error::ServerStartupError(
+                                "process id message".into(),
+                                server_identifier,
+                            ))
+                        }
                     };
 
                     secret_key = match stream.read_i32().await {
                         Ok(id) => id,
-                        Err(_) => return Err(Error::SocketError),
+                        Err(_) => {
+                            return Err(Error::ServerStartupError(
+                                "secret key message".into(),
+                                server_identifier,
+                            ))
+                        }
                     };
                 }
 
@@ -307,29 +589,38 @@ impl Server {
 
                     match stream.read_exact(&mut idle).await {
                         Ok(_) => (),
-                        Err(_) => return Err(Error::SocketError),
+                        Err(_) => {
+                            return Err(Error::ServerStartupError(
+                                "transaction status message".into(),
+                                server_identifier,
+                            ))
+                        }
                     };
 
-                    let (read, write) = stream.into_split();
-
                     let mut server = Server {
                         address: address.clone(),
-                        read: BufReader::new(read),
-                        write: write,
+                        stream: BufStream::new(stream),
                         buffer: BytesMut::with_capacity(8196),
-                        server_info: server_info,
-                        server_id: server_id,
-                        process_id: process_id,
-                        secret_key: secret_key,
+                        server_info,
+                        process_id,
+                        secret_key,
                         in_transaction: false,
                         data_available: false,
                         bad: false,
                         needs_cleanup: false,
-                        client_server_map: client_server_map,
+                        client_server_map,
                         connected_at: chrono::offset::Utc::now().naive_utc(),
-                        stats: stats,
+                        stats,
                         application_name: String::new(),
                         last_activity: SystemTime::now(),
+                        mirror_manager: match address.mirrors.len() {
+                            0 => None,
+                            _ => Some(MirroringManager::from_addresses(
+                                user.clone(),
+                                database.to_owned(),
+                                address.mirrors.clone(),
+                            )),
+                        },
                     };
 
                     server.set_name("pgcat").await?;
@@ -341,7 +632,10 @@ impl Server {
                 // Means we implemented the protocol wrong or we're not talking to a Postgres server.
                 _ => {
                     error!("Unknown code: {}", code);
-                    return Err(Error::ProtocolSyncError);
+                    return Err(Error::ProtocolSyncError(format!(
+                        "Unknown server code: {}",
+                        code
+                    )));
                 }
             };
         }
@@ -359,9 +653,10 @@ impl Server {
             Ok(stream) => stream,
             Err(err) => {
                 error!("Could not connect to server: {}", err);
-                return Err(Error::SocketError);
+                return Err(Error::SocketError("Error reading cancel message".into()));
             }
         };
+        configure_socket(&stream);
 
         debug!("Sending CancelRequest");
 
@@ -371,14 +666,15 @@ impl Server {
         bytes.put_i32(process_id);
         bytes.put_i32(secret_key);
 
-        Ok(write_all(&mut stream, bytes).await?)
+        write_all_flush(&mut stream, &bytes).await
     }
 
     /// Send messages to the server from the client.
-    pub async fn send(&mut self, messages: BytesMut) -> Result<(), Error> {
-        self.stats.data_sent(messages.len(), self.server_id);
+    pub async fn send(&mut self, messages: &BytesMut) -> Result<(), Error> {
+        self.mirror_send(messages);
+        self.stats().data_sent(messages.len());
 
-        match write_all_half(&mut self.write, messages).await {
+        match write_all_flush(&mut self.stream, &messages).await {
             Ok(_) => {
                 // Successfully sent to server
                 self.last_activity = SystemTime::now();
@@ -397,7 +693,7 @@ impl Server {
     /// in order to receive all data the server has to offer.
     pub async fn recv(&mut self) -> Result<BytesMut, Error> {
         loop {
-            let mut message = match read_message(&mut self.read).await {
+            let mut message = match read_message(&mut self.stream).await {
                 Ok(message) => message,
                 Err(err) => {
                     error!("Terminating server because of: {:?}", err);
@@ -438,7 +734,10 @@ impl Server {
                         // Something totally unexpected, this is not a Postgres server we know.
                         _ => {
                             self.bad = true;
-                            return Err(Error::ProtocolSyncError);
+                            return Err(Error::ProtocolSyncError(format!(
+                                "Unknown transaction state: {}",
+                                transaction_state
+                            )));
                         }
                     };
 
@@ -457,7 +756,17 @@ impl Server {
                             // which can leak between clients. This is a best effort to block bad clients
                             // from poisoning a transaction-mode pool by setting inappropriate session variables
                             match command_tag.as_str() {
-                                "SET\0" | "PREPARE\0" => {
+                                "SET\0" => {
+                                    // We don't detect set statements in transactions
+                                    // No great way to differentiate between set and set local
+                                    // As a result, we will miss cases when set statements are used in transactions
+                                    // This will reduce amount of discard statements sent
+                                    if !self.in_transaction {
+                                        debug!("Server connection marked for clean up");
+                                        self.needs_cleanup = true;
+                                    }
+                                }
+                                "PREPARE\0" => {
                                     debug!("Server connection marked for clean up");
                                     self.needs_cleanup = true;
                                 }
@@ -491,9 +800,13 @@ impl Server {
                     break;
                 }
 
-                // CopyData: we are not buffering this one because there will be many more
-                // and we don't know how big this packet could be, best not to take a risk.
-                'd' => break,
+                // CopyData
+                'd' => {
+                    // Don't flush yet, buffer until we reach limit
+                    if self.buffer.len() >= 8196 {
+                        break;
+                    }
+                }
 
                 // CopyDone
                 // Buffer until ReadyForQuery shows up, so don't exit the loop yet.
@@ -508,7 +821,7 @@ impl Server {
         let bytes = self.buffer.clone();
 
         // Keep track of how much data we got from the server for stats.
-        self.stats.data_received(bytes.len(), self.server_id);
+        self.stats().data_received(bytes.len());
 
         // Clear the buffer for next query.
         self.buffer.clear();
@@ -523,6 +836,7 @@ impl Server {
     /// If the server is still inside a transaction.
     /// If the client disconnects while the server is in a transaction, we will clean it up.
     pub fn in_transaction(&self) -> bool {
+        debug!("Server in transaction: {}", self.in_transaction);
         self.in_transaction
     }
 
@@ -570,7 +884,7 @@ impl Server {
     pub async fn query(&mut self, query: &str) -> Result<(), Error> {
         let query = simple_query(query);
 
-        self.send(query).await?;
+        self.send(&query).await?;
 
         loop {
             let _ = self.recv().await?;
@@ -595,7 +909,7 @@ impl Server {
             self.query("ROLLBACK").await?;
         }
 
-        // Client disconnected but it perfromed session-altering operations such as
+        // Client disconnected but it performed session-altering operations such as
         // SET statement_timeout to 1 or create a prepared statement. We clear that
         // to avoid leaking state between clients. For performance reasons we only
         // send `DISCARD ALL` if we think the session is altered instead of just sending
@@ -606,7 +920,7 @@ impl Server {
             self.needs_cleanup = false;
         }
 
-        return Ok(());
+        Ok(())
     }
 
     /// A shorthand for `SET application_name = $1`.
@@ -621,24 +935,23 @@ impl Server {
                 .query(&format!("SET application_name = '{}'", name))
                 .await?);
             self.needs_cleanup = needs_cleanup_before;
-            return result;
+            result
         } else {
             Ok(())
         }
     }
 
+    /// get Server stats
+    pub fn stats(&self) -> Arc<ServerStats> {
+        self.stats.clone()
+    }
+
     /// Get the servers address.
     #[allow(dead_code)]
     pub fn address(&self) -> Address {
         self.address.clone()
     }
 
-    /// Get the server connection identifier
-    /// Used to uniquely identify connection in statistics
-    pub fn server_id(&self) -> i32 {
-        self.server_id
-    }
-
     // Get server's latest response timestamp
     pub fn last_activity(&self) -> SystemTime {
         self.last_activity
@@ -648,6 +961,119 @@ impl Server {
     pub fn mark_dirty(&mut self) {
         self.needs_cleanup = true;
     }
+
+    pub fn mirror_send(&mut self, bytes: &BytesMut) {
+        match self.mirror_manager.as_mut() {
+            Some(manager) => manager.send(bytes),
+            None => (),
+        }
+    }
+
+    pub fn mirror_disconnect(&mut self) {
+        match self.mirror_manager.as_mut() {
+            Some(manager) => manager.disconnect(),
+            None => (),
+        }
+    }
+
+    // This is so we can execute out of band queries to the server.
+    // The connection will be opened, the query executed and closed.
+    pub async fn exec_simple_query(
+        address: &Address,
+        user: &User,
+        query: &str,
+    ) -> Result<Vec<String>, Error> {
+        let client_server_map: ClientServerMap = Arc::new(Mutex::new(HashMap::new()));
+
+        debug!("Connecting to server to obtain auth hashes.");
+        let mut server = Server::startup(
+            address,
+            user,
+            &address.database,
+            client_server_map,
+            Arc::new(ServerStats::default()),
+            Arc::new(RwLock::new(None)),
+        )
+        .await?;
+        debug!("Connected!, sending query.");
+        server.send(&simple_query(query)).await?;
+        let mut message = server.recv().await?;
+
+        Ok(parse_query_message(&mut message).await?)
+    }
+}
+
+async fn parse_query_message(message: &mut BytesMut) -> Result<Vec<String>, Error> {
+    let mut pair = Vec::<String>::new();
+    match message::backend::Message::parse(message) {
+        Ok(Some(message::backend::Message::RowDescription(_description))) => {}
+        Ok(Some(message::backend::Message::ErrorResponse(err))) => {
+            return Err(Error::ProtocolSyncError(format!(
+                "Protocol error parsing response. Err: {:?}",
+                err.fields()
+                    .iterator()
+                    .fold(String::default(), |acc, element| acc
+                        + element.unwrap().value())
+            )))
+        }
+        Ok(_) => {
+            return Err(Error::ProtocolSyncError(
+                "Protocol error, expected Row Description.".to_string(),
+            ))
+        }
+        Err(err) => {
+            return Err(Error::ProtocolSyncError(format!(
+                "Protocol error parsing response. Err: {:?}",
+                err
+            )))
+        }
+    }
+
+    while !message.is_empty() {
+        match message::backend::Message::parse(message) {
+            Ok(postgres_message) => {
+                match postgres_message {
+                    Some(message::backend::Message::DataRow(data)) => {
+                        let buf = data.buffer();
+                        trace!("Data: {:?}", buf);
+
+                        for item in data.ranges().iterator() {
+                            match item.as_ref() {
+                                Ok(range) => match range {
+                                    Some(range) => {
+                                        pair.push(String::from_utf8_lossy(&buf[range.clone()]).to_string());
+                                    }
+                                    None => return Err(Error::ProtocolSyncError(String::from(
+                                        "Data expected while receiving query auth data, found nothing.",
+                                    ))),
+                                },
+                                Err(err) => {
+                                    return Err(Error::ProtocolSyncError(format!(
+                                        "Data error, err: {:?}",
+                                        err
+                                    )))
+                                }
+                            }
+                        }
+                    }
+                    Some(message::backend::Message::CommandComplete(_)) => {}
+                    Some(message::backend::Message::ReadyForQuery(_)) => {}
+                    _ => {
+                        return Err(Error::ProtocolSyncError(
+                            "Unexpected message while receiving auth query data.".to_string(),
+                        ))
+                    }
+                }
+            }
+            Err(err) => {
+                return Err(Error::ProtocolSyncError(format!(
+                    "Parse error, err: {:?}",
+                    err
+                )))
+            }
+        };
+    }
+    Ok(pair)
 }
 
 impl Drop for Server {
@@ -655,15 +1081,18 @@ impl Drop for Server {
     /// the socket is in non-blocking mode, so it may not be ready
     /// for a write.
     fn drop(&mut self) {
-        self.stats.server_disconnecting(self.server_id);
+        self.mirror_disconnect();
 
-        let mut bytes = BytesMut::with_capacity(4);
+        // Update statistics
+        self.stats.disconnect();
+
+        let mut bytes = BytesMut::with_capacity(5);
         bytes.put_u8(b'X');
         bytes.put_i32(4);
 
-        match self.write.try_write(&bytes) {
-            Ok(_) => (),
-            Err(_) => debug!("Dirty shutdown"),
+        match self.stream.get_mut().try_write(&bytes) {
+            Ok(5) => (),
+            _ => debug!("Dirty shutdown"),
         };
 
         // Should not matter.

src/sharding.rs

@@ -133,7 +133,7 @@ impl Sharder {
     #[inline]
     fn combine(mut a: u64, b: u64) -> u64 {
         a ^= b
-            .wrapping_add(0x49a0f4dd15e5a8e3 as u64)
+            .wrapping_add(0x49a0f4dd15e5a8e3_u64)
             .wrapping_add(a << 54)
             .wrapping_add(a >> 7);
         a
@@ -141,7 +141,7 @@ impl Sharder {
 
     #[inline]
     fn pg_u32_hash(k: u32) -> u64 {
-        let mut a: u32 = 0x9e3779b9 as u32 + std::mem::size_of::<u32>() as u32 + 3923095 as u32;
+        let mut a: u32 = 0x9e3779b9_u32 + std::mem::size_of::<u32>() as u32 + 3923095_u32;
         let mut b = a;
         let c = a;
 

src/stats/address.rs

@@ -0,0 +1,149 @@
+use log::warn;
+use std::sync::atomic::*;
+use std::sync::Arc;
+
+/// Internal address stats
+#[derive(Debug, Clone, Default)]
+pub struct AddressStats {
+    pub total_xact_count: Arc<AtomicU64>,
+    pub total_query_count: Arc<AtomicU64>,
+    pub total_received: Arc<AtomicU64>,
+    pub total_sent: Arc<AtomicU64>,
+    pub total_xact_time: Arc<AtomicU64>,
+    pub total_query_time: Arc<AtomicU64>,
+    pub total_wait_time: Arc<AtomicU64>,
+    pub total_errors: Arc<AtomicU64>,
+    pub avg_query_count: Arc<AtomicU64>,
+    pub avg_query_time: Arc<AtomicU64>,
+    pub avg_recv: Arc<AtomicU64>,
+    pub avg_sent: Arc<AtomicU64>,
+    pub avg_errors: Arc<AtomicU64>,
+    pub avg_xact_time: Arc<AtomicU64>,
+    pub avg_xact_count: Arc<AtomicU64>,
+    pub avg_wait_time: Arc<AtomicU64>,
+}
+
+impl IntoIterator for AddressStats {
+    type Item = (String, u64);
+    type IntoIter = std::vec::IntoIter<Self::Item>;
+
+    fn into_iter(self) -> Self::IntoIter {
+        vec![
+            (
+                "total_xact_count".to_string(),
+                self.total_xact_count.load(Ordering::Relaxed),
+            ),
+            (
+                "total_query_count".to_string(),
+                self.total_query_count.load(Ordering::Relaxed),
+            ),
+            (
+                "total_received".to_string(),
+                self.total_received.load(Ordering::Relaxed),
+            ),
+            (
+                "total_sent".to_string(),
+                self.total_sent.load(Ordering::Relaxed),
+            ),
+            (
+                "total_xact_time".to_string(),
+                self.total_xact_time.load(Ordering::Relaxed),
+            ),
+            (
+                "total_query_time".to_string(),
+                self.total_query_time.load(Ordering::Relaxed),
+            ),
+            (
+                "total_wait_time".to_string(),
+                self.total_wait_time.load(Ordering::Relaxed),
+            ),
+            (
+                "total_errors".to_string(),
+                self.total_errors.load(Ordering::Relaxed),
+            ),
+            (
+                "avg_xact_count".to_string(),
+                self.avg_xact_count.load(Ordering::Relaxed),
+            ),
+            (
+                "avg_query_count".to_string(),
+                self.avg_query_count.load(Ordering::Relaxed),
+            ),
+            (
+                "avg_recv".to_string(),
+                self.avg_recv.load(Ordering::Relaxed),
+            ),
+            (
+                "avg_sent".to_string(),
+                self.avg_sent.load(Ordering::Relaxed),
+            ),
+            (
+                "avg_errors".to_string(),
+                self.avg_errors.load(Ordering::Relaxed),
+            ),
+            (
+                "avg_xact_time".to_string(),
+                self.avg_xact_time.load(Ordering::Relaxed),
+            ),
+            (
+                "avg_query_time".to_string(),
+                self.avg_query_time.load(Ordering::Relaxed),
+            ),
+            (
+                "avg_wait_time".to_string(),
+                self.avg_wait_time.load(Ordering::Relaxed),
+            ),
+        ]
+        .into_iter()
+    }
+}
+
+impl AddressStats {
+    pub fn error(&self) {
+        self.total_errors.fetch_add(1, Ordering::Relaxed);
+    }
+
+    pub fn update_averages(&self) {
+        let (totals, averages) = self.fields_iterators();
+        for data in totals.iter().zip(averages.iter()) {
+            let (total, average) = data;
+            if let Err(err) = average.fetch_update(Ordering::Relaxed, Ordering::Relaxed, |avg| {
+                let total = total.load(Ordering::Relaxed);
+                let avg = (total - avg) / (crate::stats::STAT_PERIOD / 1_000); // Avg / second
+                Some(avg)
+            }) {
+                warn!("Could not update averages for addresses stats, {:?}", err);
+            }
+        }
+    }
+
+    pub fn populate_row(&self, row: &mut Vec<String>) {
+        for (_key, value) in self.clone() {
+            row.push(value.to_string());
+        }
+    }
+
+    fn fields_iterators(&self) -> (Vec<Arc<AtomicU64>>, Vec<Arc<AtomicU64>>) {
+        let mut totals: Vec<Arc<AtomicU64>> = Vec::new();
+        let mut averages: Vec<Arc<AtomicU64>> = Vec::new();
+
+        totals.push(self.total_xact_count.clone());
+        averages.push(self.avg_xact_count.clone());
+        totals.push(self.total_query_count.clone());
+        averages.push(self.avg_query_count.clone());
+        totals.push(self.total_received.clone());
+        averages.push(self.avg_recv.clone());
+        totals.push(self.total_sent.clone());
+        averages.push(self.avg_sent.clone());
+        totals.push(self.total_xact_time.clone());
+        averages.push(self.avg_xact_time.clone());
+        totals.push(self.total_query_time.clone());
+        averages.push(self.avg_query_time.clone());
+        totals.push(self.total_wait_time.clone());
+        averages.push(self.avg_wait_time.clone());
+        totals.push(self.total_errors.clone());
+        averages.push(self.avg_errors.clone());
+
+        (totals, averages)
+    }
+}

src/stats/client.rs

@@ -0,0 +1,182 @@
+use super::PoolStats;
+use super::{get_reporter, Reporter};
+use atomic_enum::atomic_enum;
+use std::sync::atomic::*;
+use std::sync::Arc;
+use tokio::time::Instant;
+/// The various states that a client can be in
+#[atomic_enum]
+#[derive(PartialEq)]
+pub enum ClientState {
+    Idle = 0,
+    Waiting,
+    Active,
+}
+impl std::fmt::Display for ClientState {
+    fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
+        match *self {
+            ClientState::Idle => write!(f, "idle"),
+            ClientState::Waiting => write!(f, "waiting"),
+            ClientState::Active => write!(f, "active"),
+        }
+    }
+}
+
+#[derive(Debug, Clone)]
+/// Information we keep track of which can be queried by SHOW CLIENTS
+pub struct ClientStats {
+    /// A random integer assigned to the client and used by stats to track the client
+    client_id: i32,
+
+    /// Data associated with the client, not writable, only set when we construct the ClientStat
+    application_name: String,
+    username: String,
+    pool_name: String,
+    connect_time: Instant,
+
+    pool_stats: Arc<PoolStats>,
+    reporter: Reporter,
+
+    /// Total time spent waiting for a connection from pool, measures in microseconds
+    pub total_wait_time: Arc<AtomicU64>,
+
+    /// Current state of the client
+    pub state: Arc<AtomicClientState>,
+
+    /// Number of transactions executed by this client
+    pub transaction_count: Arc<AtomicU64>,
+
+    /// Number of queries executed by this client
+    pub query_count: Arc<AtomicU64>,
+
+    /// Number of errors made by this client
+    pub error_count: Arc<AtomicU64>,
+}
+
+impl Default for ClientStats {
+    fn default() -> Self {
+        ClientStats {
+            client_id: 0,
+            connect_time: Instant::now(),
+            application_name: String::new(),
+            username: String::new(),
+            pool_name: String::new(),
+            pool_stats: Arc::new(PoolStats::default()),
+            total_wait_time: Arc::new(AtomicU64::new(0)),
+            state: Arc::new(AtomicClientState::new(ClientState::Idle)),
+            transaction_count: Arc::new(AtomicU64::new(0)),
+            query_count: Arc::new(AtomicU64::new(0)),
+            error_count: Arc::new(AtomicU64::new(0)),
+            reporter: get_reporter(),
+        }
+    }
+}
+
+impl ClientStats {
+    pub fn new(
+        client_id: i32,
+        application_name: &str,
+        username: &str,
+        pool_name: &str,
+        connect_time: Instant,
+        pool_stats: Arc<PoolStats>,
+    ) -> Self {
+        Self {
+            client_id,
+            pool_stats,
+            connect_time,
+            application_name: application_name.to_string(),
+            username: username.to_string(),
+            pool_name: pool_name.to_string(),
+            ..Default::default()
+        }
+    }
+
+    /// Reports a client is disconnecting from the pooler and
+    /// update metrics on the corresponding pool.
+    pub fn disconnect(&self) {
+        self.reporter.client_disconnecting(self.client_id);
+        self.pool_stats
+            .client_disconnect(self.state.load(Ordering::Relaxed))
+    }
+
+    /// Register a client with the stats system. The stats system uses client_id
+    /// to track and aggregate statistics from all source that relate to that client
+    pub fn register(&self, stats: Arc<ClientStats>) {
+        self.reporter.client_register(self.client_id, stats);
+        self.state.store(ClientState::Idle, Ordering::Relaxed);
+        self.pool_stats.cl_idle.fetch_add(1, Ordering::Relaxed);
+    }
+
+    /// Reports a client is done querying the server and is no longer assigned a server connection
+    pub fn idle(&self) {
+        self.pool_stats
+            .client_idle(self.state.load(Ordering::Relaxed));
+        self.state.store(ClientState::Idle, Ordering::Relaxed);
+    }
+
+    /// Reports a client is waiting for a connection
+    pub fn waiting(&self) {
+        self.pool_stats
+            .client_waiting(self.state.load(Ordering::Relaxed));
+        self.state.store(ClientState::Waiting, Ordering::Relaxed);
+    }
+
+    /// Reports a client is done waiting for a connection and is about to query the server.
+    pub fn active(&self) {
+        self.pool_stats
+            .client_active(self.state.load(Ordering::Relaxed));
+        self.state.store(ClientState::Active, Ordering::Relaxed);
+    }
+
+    /// Reports a client has failed to obtain a connection from a connection pool
+    pub fn checkout_error(&self) {
+        self.state.store(ClientState::Idle, Ordering::Relaxed);
+    }
+
+    /// Reports a client has had the server assigned to it be banned
+    pub fn ban_error(&self) {
+        self.state.store(ClientState::Idle, Ordering::Relaxed);
+        self.error_count.fetch_add(1, Ordering::Relaxed);
+    }
+
+    /// Reporters the time spent by a client waiting to get a healthy connection from the pool
+    pub fn checkout_time(&self, microseconds: u64) {
+        self.total_wait_time
+            .fetch_add(microseconds, Ordering::Relaxed);
+    }
+
+    /// Report a query executed by a client against a server
+    pub fn query(&self) {
+        self.query_count.fetch_add(1, Ordering::Relaxed);
+    }
+
+    /// Report a transaction executed by a client a server
+    /// we report each individual queries outside a transaction as a transaction
+    /// We only count the initial BEGIN as a transaction, all queries within do not
+    /// count as transactions
+    pub fn transaction(&self) {
+        self.transaction_count.fetch_add(1, Ordering::Relaxed);
+    }
+
+    // Helper methods for show clients
+    pub fn connect_time(&self) -> Instant {
+        self.connect_time
+    }
+
+    pub fn client_id(&self) -> i32 {
+        self.client_id
+    }
+
+    pub fn application_name(&self) -> String {
+        self.application_name.clone()
+    }
+
+    pub fn username(&self) -> String {
+        self.username.clone()
+    }
+
+    pub fn pool_name(&self) -> String {
+        self.pool_name.clone()
+    }
+}

src/stats/pool.rs

@@ -0,0 +1,274 @@
+use crate::config::Pool;
+use crate::config::PoolMode;
+use crate::pool::PoolIdentifier;
+use std::sync::atomic::*;
+use std::sync::Arc;
+
+use super::get_reporter;
+use super::Reporter;
+use super::{ClientState, ServerState};
+
+#[derive(Debug, Clone, Default)]
+/// A struct that holds information about a Pool .
+pub struct PoolStats {
+    // Pool identifier, cannot be changed after creating the instance
+    identifier: PoolIdentifier,
+
+    // Pool Config, cannot be changed after creating the instance
+    config: Pool,
+
+    // A reference to the global reporter.
+    reporter: Reporter,
+
+    /// Counters (atomics)
+    pub cl_idle: Arc<AtomicU64>,
+    pub cl_active: Arc<AtomicU64>,
+    pub cl_waiting: Arc<AtomicU64>,
+    pub cl_cancel_req: Arc<AtomicU64>,
+    pub sv_active: Arc<AtomicU64>,
+    pub sv_idle: Arc<AtomicU64>,
+    pub sv_used: Arc<AtomicU64>,
+    pub sv_tested: Arc<AtomicU64>,
+    pub sv_login: Arc<AtomicU64>,
+    pub maxwait: Arc<AtomicU64>,
+}
+
+impl IntoIterator for PoolStats {
+    type Item = (String, u64);
+    type IntoIter = std::vec::IntoIter<Self::Item>;
+
+    fn into_iter(self) -> Self::IntoIter {
+        vec![
+            ("cl_idle".to_string(), self.cl_idle.load(Ordering::Relaxed)),
+            (
+                "cl_active".to_string(),
+                self.cl_active.load(Ordering::Relaxed),
+            ),
+            (
+                "cl_waiting".to_string(),
+                self.cl_waiting.load(Ordering::Relaxed),
+            ),
+            (
+                "cl_cancel_req".to_string(),
+                self.cl_cancel_req.load(Ordering::Relaxed),
+            ),
+            (
+                "sv_active".to_string(),
+                self.sv_active.load(Ordering::Relaxed),
+            ),
+            ("sv_idle".to_string(), self.sv_idle.load(Ordering::Relaxed)),
+            ("sv_used".to_string(), self.sv_used.load(Ordering::Relaxed)),
+            (
+                "sv_tested".to_string(),
+                self.sv_tested.load(Ordering::Relaxed),
+            ),
+            (
+                "sv_login".to_string(),
+                self.sv_login.load(Ordering::Relaxed),
+            ),
+            (
+                "maxwait".to_string(),
+                self.maxwait.load(Ordering::Relaxed) / 1_000_000,
+            ),
+            (
+                "maxwait_us".to_string(),
+                self.maxwait.load(Ordering::Relaxed) % 1_000_000,
+            ),
+        ]
+        .into_iter()
+    }
+}
+
+impl PoolStats {
+    pub fn new(identifier: PoolIdentifier, config: Pool) -> Self {
+        Self {
+            identifier,
+            config,
+            reporter: get_reporter(),
+            ..Default::default()
+        }
+    }
+
+    // Getters
+    pub fn register(&self, stats: Arc<PoolStats>) {
+        self.reporter.pool_register(self.identifier.clone(), stats);
+    }
+
+    pub fn database(&self) -> String {
+        self.identifier.db.clone()
+    }
+
+    pub fn user(&self) -> String {
+        self.identifier.user.clone()
+    }
+
+    pub fn pool_mode(&self) -> PoolMode {
+        self.config.pool_mode
+    }
+
+    /// Populates an array of strings with counters (used by admin in show pools)
+    pub fn populate_row(&self, row: &mut Vec<String>) {
+        for (_key, value) in self.clone() {
+            row.push(value.to_string());
+        }
+    }
+
+    /// Deletes the maxwait counter, this is done everytime we obtain metrics
+    pub fn clear_maxwait(&self) {
+        self.maxwait.store(0, Ordering::Relaxed);
+    }
+
+    /// Notified when a server of the pool enters login state.
+    ///
+    /// Arguments:
+    ///
+    /// `from`: The state of the server that notifies.
+    pub fn server_login(&self, from: ServerState) {
+        self.sv_login.fetch_add(1, Ordering::Relaxed);
+        if from != ServerState::Login {
+            self.decrease_from_server_state(from);
+        }
+    }
+
+    /// Notified when a server of the pool become 'active'
+    ///
+    /// Arguments:
+    ///
+    /// `from`: The state of the server that notifies.
+    pub fn server_active(&self, from: ServerState) {
+        self.sv_active.fetch_add(1, Ordering::Relaxed);
+        if from != ServerState::Active {
+            self.decrease_from_server_state(from);
+        }
+    }
+
+    /// Notified when a server of the pool become 'tested'
+    ///
+    /// Arguments:
+    ///
+    /// `from`: The state of the server that notifies.
+    pub fn server_tested(&self, from: ServerState) {
+        self.sv_tested.fetch_add(1, Ordering::Relaxed);
+        if from != ServerState::Tested {
+            self.decrease_from_server_state(from);
+        }
+    }
+
+    /// Notified when a server of the pool become 'idle'
+    ///
+    /// Arguments:
+    ///
+    /// `from`: The state of the server that notifies.
+    pub fn server_idle(&self, from: ServerState) {
+        self.sv_idle.fetch_add(1, Ordering::Relaxed);
+        if from != ServerState::Idle {
+            self.decrease_from_server_state(from);
+        }
+    }
+
+    /// Notified when a client of the pool become 'waiting'
+    ///
+    /// Arguments:
+    ///
+    /// `from`: The state of the client that notifies.
+    pub fn client_waiting(&self, from: ClientState) {
+        if from != ClientState::Waiting {
+            self.cl_waiting.fetch_add(1, Ordering::Relaxed);
+            self.decrease_from_client_state(from);
+        }
+    }
+
+    /// Notified when a client of the pool become 'active'
+    ///
+    /// Arguments:
+    ///
+    /// `from`: The state of the client that notifies.
+    pub fn client_active(&self, from: ClientState) {
+        if from != ClientState::Active {
+            self.cl_active.fetch_add(1, Ordering::Relaxed);
+            self.decrease_from_client_state(from);
+        }
+    }
+
+    /// Notified when a client of the pool become 'idle'
+    ///
+    /// Arguments:
+    ///
+    /// `from`: The state of the client that notifies.
+    pub fn client_idle(&self, from: ClientState) {
+        if from != ClientState::Idle {
+            self.cl_idle.fetch_add(1, Ordering::Relaxed);
+            self.decrease_from_client_state(from);
+        }
+    }
+
+    /// Notified when a client disconnects.
+    ///
+    /// Arguments:
+    ///
+    /// `from`: The state of the client that notifies.
+    pub fn client_disconnect(&self, from: ClientState) {
+        let counter = match from {
+            ClientState::Idle => &self.cl_idle,
+            ClientState::Waiting => &self.cl_waiting,
+            ClientState::Active => &self.cl_active,
+        };
+
+        Self::decrease_counter(counter.clone());
+    }
+
+    /// Notified when a server disconnects.
+    ///
+    /// Arguments:
+    ///
+    /// `from`: The state of the client that notifies.
+    pub fn server_disconnect(&self, from: ServerState) {
+        let counter = match from {
+            ServerState::Active => &self.sv_active,
+            ServerState::Idle => &self.sv_idle,
+            ServerState::Login => &self.sv_login,
+            ServerState::Tested => &self.sv_tested,
+        };
+        Self::decrease_counter(counter.clone());
+    }
+
+    // helpers for counter decrease
+    fn decrease_from_server_state(&self, from: ServerState) {
+        let counter = match from {
+            ServerState::Tested => &self.sv_tested,
+            ServerState::Active => &self.sv_active,
+            ServerState::Idle => &self.sv_idle,
+            ServerState::Login => &self.sv_login,
+        };
+        Self::decrease_counter(counter.clone());
+    }
+
+    fn decrease_from_client_state(&self, from: ClientState) {
+        let counter = match from {
+            ClientState::Active => &self.cl_active,
+            ClientState::Idle => &self.cl_idle,
+            ClientState::Waiting => &self.cl_waiting,
+        };
+        Self::decrease_counter(counter.clone());
+    }
+
+    fn decrease_counter(value: Arc<AtomicU64>) {
+        if value.load(Ordering::Relaxed) > 0 {
+            value.fetch_sub(1, Ordering::Relaxed);
+        }
+    }
+}
+
+#[cfg(test)]
+mod test {
+    use super::*;
+
+    #[test]
+    fn test_decrease() {
+        let stat: PoolStats = PoolStats::default();
+        stat.server_login(ServerState::Login);
+        stat.server_idle(ServerState::Login);
+        assert_eq!(stat.sv_login.load(Ordering::Relaxed), 0);
+        assert_eq!(stat.sv_idle.load(Ordering::Relaxed), 1);
+    }
+}

src/stats/server.rs

@@ -0,0 +1,225 @@
+use super::AddressStats;
+use super::PoolStats;
+use super::{get_reporter, Reporter};
+use crate::config::Address;
+use atomic_enum::atomic_enum;
+use parking_lot::RwLock;
+use std::sync::atomic::*;
+use std::sync::Arc;
+use tokio::time::Instant;
+
+/// The various states that a server can be in
+#[atomic_enum]
+#[derive(PartialEq)]
+pub enum ServerState {
+    Login = 0,
+    Active,
+    Tested,
+    Idle,
+}
+impl std::fmt::Display for ServerState {
+    fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
+        match *self {
+            ServerState::Login => write!(f, "login"),
+            ServerState::Active => write!(f, "active"),
+            ServerState::Tested => write!(f, "tested"),
+            ServerState::Idle => write!(f, "idle"),
+        }
+    }
+}
+
+/// Information we keep track of which can be queried by SHOW SERVERS
+#[derive(Debug, Clone)]
+pub struct ServerStats {
+    /// A random integer assigned to the server and used by stats to track the server
+    server_id: i32,
+
+    /// Context information, only to be read
+    address: Address,
+    connect_time: Instant,
+
+    pool_stats: Arc<PoolStats>,
+    reporter: Reporter,
+
+    /// Data
+    pub application_name: Arc<RwLock<String>>,
+    pub state: Arc<AtomicServerState>,
+    pub bytes_sent: Arc<AtomicU64>,
+    pub bytes_received: Arc<AtomicU64>,
+    pub transaction_count: Arc<AtomicU64>,
+    pub query_count: Arc<AtomicU64>,
+    pub error_count: Arc<AtomicU64>,
+}
+
+impl Default for ServerStats {
+    fn default() -> Self {
+        ServerStats {
+            server_id: 0,
+            application_name: Arc::new(RwLock::new(String::new())),
+            address: Address::default(),
+            pool_stats: Arc::new(PoolStats::default()),
+            connect_time: Instant::now(),
+            state: Arc::new(AtomicServerState::new(ServerState::Login)),
+            bytes_sent: Arc::new(AtomicU64::new(0)),
+            bytes_received: Arc::new(AtomicU64::new(0)),
+            transaction_count: Arc::new(AtomicU64::new(0)),
+            query_count: Arc::new(AtomicU64::new(0)),
+            error_count: Arc::new(AtomicU64::new(0)),
+            reporter: get_reporter(),
+        }
+    }
+}
+
+impl ServerStats {
+    pub fn new(address: Address, pool_stats: Arc<PoolStats>, connect_time: Instant) -> Self {
+        Self {
+            address,
+            pool_stats,
+            connect_time,
+            server_id: rand::random::<i32>(),
+            ..Default::default()
+        }
+    }
+
+    pub fn server_id(&self) -> i32 {
+        self.server_id
+    }
+
+    /// Register a server connection with the stats system. The stats system uses server_id
+    /// to track and aggregate statistics from all source that relate to that server
+    // Delegates to reporter
+    pub fn register(&self, stats: Arc<ServerStats>) {
+        self.reporter.server_register(self.server_id, stats);
+        self.login();
+    }
+
+    /// Reports a server connection is no longer assigned to a client
+    /// and is available for the next client to pick it up
+    pub fn idle(&self) {
+        self.pool_stats
+            .server_idle(self.state.load(Ordering::Relaxed));
+
+        self.state.store(ServerState::Idle, Ordering::Relaxed);
+    }
+
+    /// Reports a server connection is disconnecting from the pooler.
+    /// Also updates metrics on the pool regarding server usage.
+    pub fn disconnect(&self) {
+        self.reporter.server_disconnecting(self.server_id);
+        self.pool_stats
+            .server_disconnect(self.state.load(Ordering::Relaxed))
+    }
+
+    /// Reports a server connection is being tested before being given to a client.
+    pub fn tested(&self) {
+        self.set_undefined_application();
+        self.pool_stats
+            .server_tested(self.state.load(Ordering::Relaxed));
+        self.state.store(ServerState::Tested, Ordering::Relaxed);
+    }
+
+    /// Reports a server connection is attempting to login.
+    pub fn login(&self) {
+        self.pool_stats
+            .server_login(self.state.load(Ordering::Relaxed));
+        self.state.store(ServerState::Login, Ordering::Relaxed);
+        self.set_undefined_application();
+    }
+
+    /// Reports a server connection has been assigned to a client that
+    /// is about to query the server
+    pub fn active(&self, application_name: String) {
+        self.pool_stats
+            .server_active(self.state.load(Ordering::Relaxed));
+        self.state.store(ServerState::Active, Ordering::Relaxed);
+        self.set_application(application_name);
+    }
+
+    pub fn address_stats(&self) -> Arc<AddressStats> {
+        self.address.stats.clone()
+    }
+
+    // Helper methods for show_servers
+    pub fn pool_name(&self) -> String {
+        self.pool_stats.database()
+    }
+
+    pub fn username(&self) -> String {
+        self.pool_stats.user()
+    }
+
+    pub fn address_name(&self) -> String {
+        self.address.name()
+    }
+
+    pub fn connect_time(&self) -> Instant {
+        self.connect_time
+    }
+
+    fn set_application(&self, name: String) {
+        let mut application_name = self.application_name.write();
+        *application_name = name;
+    }
+
+    fn set_undefined_application(&self) {
+        self.set_application(String::from("Undefined"))
+    }
+
+    pub fn checkout_time(&self, microseconds: u64, application_name: String) {
+        // Update server stats and address aggergation stats
+        self.set_application(application_name);
+        self.address
+            .stats
+            .total_wait_time
+            .fetch_add(microseconds, Ordering::Relaxed);
+        self.pool_stats
+            .maxwait
+            .fetch_max(microseconds, Ordering::Relaxed);
+    }
+
+    /// Report a query executed by a client against a server
+    pub fn query(&self, milliseconds: u64, application_name: &str) {
+        self.set_application(application_name.to_string());
+        let address_stats = self.address_stats();
+        address_stats
+            .total_query_count
+            .fetch_add(1, Ordering::Relaxed);
+        address_stats
+            .total_query_time
+            .fetch_add(milliseconds, Ordering::Relaxed);
+    }
+
+    /// Report a transaction executed by a client a server
+    /// we report each individual queries outside a transaction as a transaction
+    /// We only count the initial BEGIN as a transaction, all queries within do not
+    /// count as transactions
+    pub fn transaction(&self, application_name: &str) {
+        self.set_application(application_name.to_string());
+
+        self.transaction_count.fetch_add(1, Ordering::Relaxed);
+        self.address
+            .stats
+            .total_xact_count
+            .fetch_add(1, Ordering::Relaxed);
+    }
+
+    /// Report data sent to a server
+    pub fn data_sent(&self, amount_bytes: usize) {
+        self.bytes_sent
+            .fetch_add(amount_bytes as u64, Ordering::Relaxed);
+        self.address
+            .stats
+            .total_sent
+            .fetch_add(amount_bytes as u64, Ordering::Relaxed);
+    }
+
+    /// Report data received from a server
+    pub fn data_received(&self, amount_bytes: usize) {
+        self.bytes_received
+            .fetch_add(amount_bytes as u64, Ordering::Relaxed);
+        self.address
+            .stats
+            .total_received
+            .fetch_add(amount_bytes as u64, Ordering::Relaxed);
+    }
+}

src/tls.rs

@@ -1,9 +1,15 @@
 // Stream wrapper.
 
-use rustls_pemfile::{certs, rsa_private_keys};
+use rustls_pemfile::{certs, read_one, Item};
+use std::iter;
 use std::path::Path;
 use std::sync::Arc;
-use tokio_rustls::rustls::{self, Certificate, PrivateKey};
+use std::time::SystemTime;
+use tokio_rustls::rustls::{
+    self,
+    client::{ServerCertVerified, ServerCertVerifier},
+    Certificate, PrivateKey, ServerName,
+};
 use tokio_rustls::TlsAcceptor;
 
 use crate::config::get_config;
@@ -17,9 +23,17 @@ pub fn load_certs(path: &Path) -> std::io::Result<Vec<Certificate>> {
 }
 
 pub fn load_keys(path: &Path) -> std::io::Result<Vec<PrivateKey>> {
-    rsa_private_keys(&mut std::io::BufReader::new(std::fs::File::open(path)?))
-        .map_err(|_| std::io::Error::new(std::io::ErrorKind::InvalidInput, "invalid key"))
-        .map(|mut keys| keys.drain(..).map(PrivateKey).collect())
+    let mut rd = std::io::BufReader::new(std::fs::File::open(path)?);
+
+    iter::from_fn(|| read_one(&mut rd).transpose())
+        .filter_map(|item| match item {
+            Err(err) => Some(Err(err)),
+            Ok(Item::RSAKey(key)) => Some(Ok(PrivateKey(key))),
+            Ok(Item::ECKey(key)) => Some(Ok(PrivateKey(key))),
+            Ok(Item::PKCS8Key(key)) => Some(Ok(PrivateKey(key))),
+            _ => None,
+        })
+        .collect()
 }
 
 pub struct Tls {
@@ -30,12 +44,12 @@ impl Tls {
     pub fn new() -> Result<Self, Error> {
         let config = get_config();
 
-        let certs = match load_certs(&Path::new(&config.general.tls_certificate.unwrap())) {
+        let certs = match load_certs(Path::new(&config.general.tls_certificate.unwrap())) {
             Ok(certs) => certs,
             Err(_) => return Err(Error::TlsError),
         };
 
-        let mut keys = match load_keys(&Path::new(&config.general.tls_private_key.unwrap())) {
+        let mut keys = match load_keys(Path::new(&config.general.tls_private_key.unwrap())) {
             Ok(keys) => keys,
             Err(_) => return Err(Error::TlsError),
         };
@@ -55,3 +69,19 @@ impl Tls {
         })
     }
 }
+
+pub struct NoCertificateVerification;
+
+impl ServerCertVerifier for NoCertificateVerification {
+    fn verify_server_cert(
+        &self,
+        _end_entity: &Certificate,
+        _intermediates: &[Certificate],
+        _server_name: &ServerName,
+        _scts: &mut dyn Iterator<Item = &[u8]>,
+        _ocsp_response: &[u8],
+        _now: SystemTime,
+    ) -> Result<ServerCertVerified, rustls::Error> {
+        Ok(ServerCertVerified::assertion())
+    }
+}

tests/docker/Dockerfile

@@ -1,5 +1,8 @@
 FROM rust:bullseye
 
-RUN apt-get update && apt-get install llvm-11 psmisc postgresql-contrib postgresql-client ruby ruby-dev libpq-dev python3 python3-pip lcov sudo curl -y
+RUN apt-get update && apt-get install llvm-11 psmisc postgresql-contrib postgresql-client ruby ruby-dev libpq-dev python3 python3-pip lcov curl sudo iproute2 -y
 RUN cargo install cargo-binutils rustfilt
 RUN rustup component add llvm-tools-preview
+RUN sudo gem install bundler
+RUN wget -O toxiproxy-2.4.0.deb https://github.com/Shopify/toxiproxy/releases/download/v2.4.0/toxiproxy_2.4.0_linux_$(dpkg --print-architecture).deb && \
+	sudo dpkg -i toxiproxy-2.4.0.deb

tests/docker/docker-compose.yml

@@ -7,8 +7,8 @@ services:
       POSTGRES_USER: postgres
       POSTGRES_DB: postgres
       POSTGRES_PASSWORD: postgres
-      POSTGRES_HOST_AUTH_METHOD: scram-sha-256
-    command: ["postgres", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-p", "5432"]
+      POSTGRES_INITDB_ARGS: --auth-local=md5 --auth-host=md5 --auth=md5
+    command: ["postgres", "-p", "5432", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-c", "pg_stat_statements.max=100000"]
   pg2:
     image: postgres:14
     network_mode: "service:main"
@@ -16,8 +16,8 @@ services:
       POSTGRES_USER: postgres
       POSTGRES_DB: postgres
       POSTGRES_PASSWORD: postgres
-      POSTGRES_HOST_AUTH_METHOD: scram-sha-256
-    command: ["postgres", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-p", "7432"]
+      POSTGRES_INITDB_ARGS: --auth-local=scram-sha-256 --auth-host=scram-sha-256 --auth=scram-sha-256
+    command: ["postgres", "-p", "7432", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-c", "pg_stat_statements.max=100000"]
   pg3:
     image: postgres:14
     network_mode: "service:main"
@@ -25,8 +25,8 @@ services:
       POSTGRES_USER: postgres
       POSTGRES_DB: postgres
       POSTGRES_PASSWORD: postgres
-      POSTGRES_HOST_AUTH_METHOD: scram-sha-256
-    command: ["postgres", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-p", "8432"]
+      POSTGRES_INITDB_ARGS: --auth-local=scram-sha-256 --auth-host=scram-sha-256 --auth=scram-sha-256
+    command: ["postgres", "-p", "8432", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-c", "pg_stat_statements.max=100000"]
   pg4:
     image: postgres:14
     network_mode: "service:main"
@@ -34,14 +34,20 @@ services:
       POSTGRES_USER: postgres
       POSTGRES_DB: postgres
       POSTGRES_PASSWORD: postgres
-      POSTGRES_HOST_AUTH_METHOD: scram-sha-256
-    command: ["postgres", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-p", "9432"]
+      POSTGRES_INITDB_ARGS: --auth-local=scram-sha-256 --auth-host=scram-sha-256 --auth=scram-sha-256
+    command: ["postgres", "-p", "9432", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-c", "pg_stat_statements.max=100000"]
+  pg5:
+    image: postgres:14
+    network_mode: "service:main"
+    environment:
+      POSTGRES_USER: postgres
+      POSTGRES_DB: postgres
+      POSTGRES_PASSWORD: postgres
+      POSTGRES_INITDB_ARGS: --auth-local=md5 --auth-host=md5 --auth=md5
+    command: ["postgres", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-p", "10432"]
   main:
     build: .
     command: ["bash", "/app/tests/docker/run.sh"]
-    environment:
-      RUSTFLAGS: "-C instrument-coverage"
-      LLVM_PROFILE_FILE: "pgcat-%m.profraw"
     volumes:
       - ../../:/app/
       - /app/target/

tests/docker/run.sh

@@ -1,21 +1,37 @@
 #!/bin/bash
 
+rm -rf /app/target/ || true
 rm /app/*.profraw || true
 rm /app/pgcat.profdata || true
 rm -rf /app/cov || true
 
-cd /app/
+export LLVM_PROFILE_FILE="/app/pgcat-%m-%p.profraw"
+export RUSTC_BOOTSTRAP=1
+export CARGO_INCREMENTAL=0
+export RUSTFLAGS="-Zprofile -Ccodegen-units=1 -Copt-level=0 -Clink-dead-code -Coverflow-checks=off -Zpanic_abort_tests -Cpanic=abort -Cinstrument-coverage"
+export RUSTDOCFLAGS="-Cpanic=abort"
 
+cd /app/
+cargo clean
 cargo build
 cargo test --tests
 
 bash .circleci/run_tests.sh
 
-rust-profdata merge -sparse pgcat-*.profraw -o pgcat.profdata
+TEST_OBJECTS=$( \
+    for file in $(cargo test --no-run 2>&1 | grep "target/debug/deps/pgcat-[[:alnum:]]\+" -o); \
+    do \
+        printf "%s %s " --object $file; \
+    done \
+)
 
-rust-cov export -ignore-filename-regex="rustc|registry" -Xdemangler=rustfilt -instr-profile=pgcat.profdata --object ./target/debug/pgcat --format lcov > ./lcov.info
+echo "Generating coverage report"
 
-genhtml lcov.info --output-directory cov --prefix $(pwd)
+rust-profdata merge -sparse /app/pgcat-*.profraw -o /app/pgcat.profdata
+
+bash -c "rust-cov export -ignore-filename-regex='rustc|registry' -Xdemangler=rustfilt -instr-profile=/app/pgcat.profdata $TEST_OBJECTS --object ./target/debug/pgcat --format lcov > ./lcov.info"
+
+genhtml lcov.info --title "PgCat Code Coverage" --css-file ./cov-style.css --highlight --no-function-coverage --ignore-errors source --legend  --output-directory cov --prefix $(pwd)
 
 rm /app/*.profraw
 rm /app/pgcat.profdata

tests/python/tests.py

@@ -110,6 +110,37 @@ def test_shutdown_logic():
     cleanup_conn(conn, cur)
     pg_cat_send_signal(signal.SIGTERM)
 
+    # - - - - - - - - - - - - - - - - - -
+    # NO ACTIVE QUERIES ADMIN SHUTDOWN COMMAND
+
+    # Start pgcat
+    pgcat_start()
+
+    # Create client connection and begin transaction
+    conn, cur = connect_db()
+    admin_conn, admin_cur = connect_db(admin=True)
+
+    cur.execute("BEGIN;")
+    cur.execute("SELECT 1;")
+    cur.execute("COMMIT;")
+
+    # Send SHUTDOWN command pgcat while not in transaction
+    admin_cur.execute("SHUTDOWN;")
+    time.sleep(1)
+
+    # Check that any new queries fail after SHUTDOWN command since server should close with no active transactions
+    try:
+        cur.execute("SELECT 1;")
+    except psycopg2.OperationalError as e:
+        pass
+    else:
+        # Fail if query execution succeeded
+        raise Exception("Server not closed after sigint")
+
+    cleanup_conn(conn, cur)
+    cleanup_conn(admin_conn, admin_cur)
+    pg_cat_send_signal(signal.SIGTERM)
+
     # - - - - - - - - - - - - - - - - - -
     # HANDLE TRANSACTION WITH SIGINT
 
@@ -136,6 +167,36 @@ def test_shutdown_logic():
     cleanup_conn(conn, cur)
     pg_cat_send_signal(signal.SIGTERM)
 
+    # - - - - - - - - - - - - - - - - - -
+    # HANDLE TRANSACTION WITH ADMIN SHUTDOWN COMMAND
+
+    # Start pgcat
+    pgcat_start()
+
+    # Create client connection and begin transaction
+    conn, cur = connect_db()
+    admin_conn, admin_cur = connect_db(admin=True)
+
+    cur.execute("BEGIN;")
+    cur.execute("SELECT 1;")
+
+    # Send SHUTDOWN command pgcat while still in transaction
+    admin_cur.execute("SHUTDOWN;")
+    if admin_cur.fetchall()[0][0] != "t":
+        raise Exception("PgCat unable to send signal")
+    time.sleep(1)
+
+    # Check that any new queries succeed after SHUTDOWN command since server should still allow transaction to complete
+    try:
+        cur.execute("SELECT 1;")
+    except psycopg2.OperationalError as e:
+        # Fail if query fails since server closed
+        raise Exception("Server closed while in transaction", e.pgerror)
+
+    cleanup_conn(conn, cur)
+    cleanup_conn(admin_conn, admin_cur)
+    pg_cat_send_signal(signal.SIGTERM)
+
     # - - - - - - - - - - - - - - - - - -
     # NO NEW NON-ADMIN CONNECTIONS DURING SHUTDOWN
     # Start pgcat

tests/ruby/Gemfile.lock

@@ -1,12 +1,12 @@
 GEM
   remote: https://rubygems.org/
   specs:
-    activemodel (7.0.3.1)
-      activesupport (= 7.0.3.1)
-    activerecord (7.0.3.1)
-      activemodel (= 7.0.3.1)
-      activesupport (= 7.0.3.1)
-    activesupport (7.0.3.1)
+    activemodel (7.0.4.1)
+      activesupport (= 7.0.4.1)
+    activerecord (7.0.4.1)
+      activemodel (= 7.0.4.1)
+      activesupport (= 7.0.4.1)
+    activesupport (7.0.4.1)
       concurrent-ruby (~> 1.0, >= 1.0.2)
       i18n (>= 1.6, < 2)
       minitest (>= 5.1)
@@ -14,9 +14,9 @@ GEM
     ast (2.4.2)
     concurrent-ruby (1.1.10)
     diff-lcs (1.5.0)
-    i18n (1.11.0)
+    i18n (1.12.0)
       concurrent-ruby (~> 1.0)
-    minitest (5.16.2)
+    minitest (5.17.0)
     parallel (1.22.1)
     parser (3.1.2.0)
       ast (~> 2.4.1)
@@ -53,7 +53,7 @@ GEM
     toml (0.3.0)
       parslet (>= 1.8.0, < 3.0.0)
     toxiproxy (2.0.1)
-    tzinfo (2.0.4)
+    tzinfo (2.0.5)
       concurrent-ruby (~> 1.0)
     unicode-display_width (2.1.0)
 

tests/ruby/admin_spec.rb

@@ -37,9 +37,9 @@ describe "Admin" do
   describe "SHOW POOLS" do
     context "bad credentials" do
       it "does not change any stats" do
-        bad_passsword_url = URI(pgcat_conn_str)
-        bad_passsword_url.password = "wrong"
-        expect { PG::connect("#{bad_passsword_url.to_s}?application_name=bad_password") }.to raise_error(PG::ConnectionBad)
+        bad_password_url = URI(pgcat_conn_str)
+        bad_password_url.password = "wrong"
+        expect { PG::connect("#{bad_password_url.to_s}?application_name=bad_password") }.to raise_error(PG::ConnectionBad)
 
         sleep(1)
         admin_conn = PG::connect(processes.pgcat.admin_connection_string)
@@ -176,6 +176,47 @@ describe "Admin" do
       end
     end
 
+    context "clients connects and disconnect normally" do
+      let(:processes) { Helpers::Pgcat.single_instance_setup("sharded_db", 2) }
+
+      it 'shows the same number of clients before and after' do
+        clients_before = clients_connected_to_pool(processes: processes)
+        threads = []
+        connections = Array.new(4) { PG::connect("#{pgcat_conn_str}?application_name=one_query") }
+        connections.each do |c|
+          threads << Thread.new { c.async_exec("SELECT 1") }
+        end
+        clients_between = clients_connected_to_pool(processes: processes)
+        expect(clients_before).not_to eq(clients_between)
+        connections.each(&:close)
+        clients_after = clients_connected_to_pool(processes: processes)
+        expect(clients_before).to eq(clients_after)
+      end
+    end
+
+    context "clients connects and disconnect abruptly" do
+      let(:processes) { Helpers::Pgcat.single_instance_setup("sharded_db", 10) }
+
+      it 'shows the same number of clients before and after' do
+        threads = []
+        connections = Array.new(2) { PG::connect("#{pgcat_conn_str}?application_name=one_query") }
+        connections.each do |c|
+          threads << Thread.new { c.async_exec("SELECT 1") }
+        end
+        clients_before = clients_connected_to_pool(processes: processes)
+        random_string = (0...8).map { (65 + rand(26)).chr }.join
+        connection_string = "#{pgcat_conn_str}?application_name=#{random_string}"
+        faulty_client = Process.spawn("psql -Atx #{connection_string} >/dev/null")
+        sleep(1)
+        # psql starts two processes, we only know the pid of the parent, this
+        # ensure both are killed
+        `pkill -9 -f '#{random_string}'`
+        Process.wait(faulty_client)
+        clients_after = clients_connected_to_pool(processes: processes)
+        expect(clients_before).to eq(clients_after)
+      end
+    end
+
     context "clients overwhelm server pools" do
       let(:processes) { Helpers::Pgcat.single_instance_setup("sharded_db", 2) }
 
@@ -199,7 +240,7 @@ describe "Admin" do
 
         sleep(2.5) # Allow time for stats to update
         results = admin_conn.async_exec("SHOW POOLS")[0]
-        %w[cl_active cl_waiting cl_cancel_req sv_active sv_used sv_tested sv_login maxwait].each do |s|
+        %w[cl_active cl_waiting cl_cancel_req sv_active sv_used sv_tested sv_login].each do |s|
           raise StandardError, "Field #{s} was expected to be 0 but found to be #{results[s]}" if results[s] != "0"
         end
         expect(results["cl_idle"]).to eq("4")
@@ -221,7 +262,7 @@ describe "Admin" do
         results = admin_conn.async_exec("SHOW POOLS")[0]
 
         expect(results["maxwait"]).to eq("1")
-        expect(results["maxwait_us"].to_i).to be_within(100_000).of(500_000)
+        expect(results["maxwait_us"].to_i).to be_within(200_000).of(500_000)
 
         sleep(4.5) # Allow time for stats to update
         results = admin_conn.async_exec("SHOW POOLS")[0]
@@ -286,4 +327,84 @@ describe "Admin" do
       connections.map(&:close)
     end
   end
+
+  describe "Manual Banning" do
+    let(:processes) { Helpers::Pgcat.single_shard_setup("sharded_db", 10) }
+    before do
+      new_configs = processes.pgcat.current_config
+      # Prevent immediate unbanning when we ban localhost
+      new_configs["pools"]["sharded_db"]["shards"]["0"]["servers"][0][0] = "127.0.0.1"
+      new_configs["pools"]["sharded_db"]["shards"]["0"]["servers"][1][0] = "127.0.0.1"
+      processes.pgcat.update_config(new_configs)
+      processes.pgcat.reload_config
+    end
+
+    describe "BAN/UNBAN and SHOW BANS" do
+      it "bans/unbans hosts" do
+        admin_conn = PG::connect(processes.pgcat.admin_connection_string)
+
+        # Returns a list of the banned addresses
+        results = admin_conn.async_exec("BAN localhost 10").to_a
+        expect(results.count).to eq(2)
+        expect(results.map{ |r| r["host"] }.uniq).to eq(["localhost"])
+
+        # Subsequent calls should yield no results
+        results = admin_conn.async_exec("BAN localhost 10").to_a
+        expect(results.count).to eq(0)
+
+        results = admin_conn.async_exec("SHOW BANS").to_a
+        expect(results.count).to eq(2)
+        expect(results.map{ |r| r["host"] }.uniq).to eq(["localhost"])
+
+        # Returns a list of the unbanned addresses
+        results = admin_conn.async_exec("UNBAN localhost").to_a
+        expect(results.count).to eq(2)
+        expect(results.map{ |r| r["host"] }.uniq).to eq(["localhost"])
+
+        # Subsequent calls should yield no results
+        results = admin_conn.async_exec("UNBAN localhost").to_a
+        expect(results.count).to eq(0)
+
+        results = admin_conn.async_exec("SHOW BANS").to_a
+        expect(results.count).to eq(0)
+      end
+
+      it "honors ban duration" do
+        admin_conn = PG::connect(processes.pgcat.admin_connection_string)
+
+        # Returns a list of the banned addresses
+        results = admin_conn.async_exec("BAN localhost 1").to_a
+        expect(results.count).to eq(2)
+        expect(results.map{ |r| r["host"] }.uniq).to eq(["localhost"])
+
+        sleep(2)
+
+        # After 2 seconds the ban should be lifted
+        results = admin_conn.async_exec("SHOW BANS").to_a
+        expect(results.count).to eq(0)
+      end
+
+      it "can handle bad input" do
+        admin_conn = PG::connect(processes.pgcat.admin_connection_string)
+
+        expect { admin_conn.async_exec("BAN").to_a }.to raise_error(PG::SystemError)
+        expect { admin_conn.async_exec("BAN a").to_a }.to raise_error(PG::SystemError)
+        expect { admin_conn.async_exec("BAN a a").to_a }.to raise_error(PG::SystemError)
+        expect { admin_conn.async_exec("BAN a -5").to_a }.to raise_error(PG::SystemError)
+        expect { admin_conn.async_exec("BAN a 0").to_a }.to raise_error(PG::SystemError)
+        expect { admin_conn.async_exec("BAN a a a").to_a }.to raise_error(PG::SystemError)
+        expect { admin_conn.async_exec("UNBAN").to_a }.to raise_error(PG::SystemError)
+      end
+    end
+  end
+
+  describe "SHOW users" do
+    it "returns the right users" do
+      admin_conn = PG::connect(processes.pgcat.admin_connection_string)
+      results = admin_conn.async_exec("SHOW USERS")[0]
+      admin_conn.close
+      expect(results["name"]).to eq("sharding_user")
+      expect(results["pool_mode"]).to eq("transaction")
+    end
+  end
 end

tests/ruby/auth_query_spec.rb

@@ -0,0 +1,215 @@
+# frozen_string_literal: true
+
+require_relative 'spec_helper'
+require_relative 'helpers/auth_query_helper'
+
+describe "Auth Query" do
+  let(:configured_instances) {[5432, 10432]}
+  let(:config_user) { { 'username' => 'sharding_user', 'password' => 'sharding_user' } }
+  let(:pg_user) { { 'username' => 'sharding_user', 'password' => 'sharding_user' } }
+  let(:processes) { Helpers::AuthQuery.single_shard_auth_query(pool_name: "sharded_db", pg_user: pg_user, config_user: config_user, extra_conf: config, wait_until_ready: wait_until_ready ) }
+  let(:config) { {} }
+  let(:wait_until_ready) { true }
+
+  after do
+    unless @failing_process
+      processes.all_databases.map(&:reset)
+      processes.pgcat.shutdown
+    end
+    @failing_process = false
+  end
+
+  context "when auth_query is not configured" do
+    context 'and cleartext passwords are set' do
+      it "uses local passwords" do
+        conn = PG.connect(processes.pgcat.connection_string("sharded_db", config_user['username'], config_user['password']))
+
+        expect(conn.async_exec("SELECT 1 + 2")).not_to be_nil
+      end
+    end
+
+    context 'and cleartext passwords are not set' do
+      let(:config_user) { { 'username' => 'sharding_user' } }
+
+      it "does not start because it is not possible to authenticate" do
+        @failing_process = true
+        expect { processes.pgcat }.to raise_error(StandardError, /You have to specify a user password for every pool if auth_query is not specified/)
+      end
+    end
+  end
+
+  context 'when auth_query is configured' do
+    context 'with global configuration' do
+      around(:example) do |example|
+
+        # Set up auth query
+        Helpers::AuthQuery.set_up_auth_query_for_user(
+          user: 'md5_auth_user',
+          password: 'secret'
+        );
+
+        example.run
+
+        # Drop auth query support
+        Helpers::AuthQuery.tear_down_auth_query_for_user(
+          user: 'md5_auth_user',
+          password: 'secret'
+        );
+      end
+
+      context 'with correct global parameters' do
+        let(:config) { { 'general' => { 'auth_query' => "SELECT * FROM public.user_lookup('$1');", 'auth_query_user' => 'md5_auth_user', 'auth_query_password' => 'secret' } } }
+        context 'and with cleartext passwords set' do
+          it 'it uses local passwords' do
+            conn = PG.connect(processes.pgcat.connection_string("sharded_db", pg_user['username'], pg_user['password']))
+            expect(conn.exec("SELECT 1 + 2")).not_to be_nil
+          end
+        end
+
+        context 'and with cleartext passwords not set' do
+          let(:config_user) { { 'username' => 'sharding_user', 'password' => 'sharding_user' } }
+
+          it 'it uses obtained passwords' do
+            connection_string = processes.pgcat.connection_string("sharded_db", pg_user['username'], pg_user['password'])
+            conn = PG.connect(connection_string)
+            expect(conn.async_exec("SELECT 1 + 2")).not_to be_nil
+          end
+
+          it 'allows passwords to be changed without closing existing connections' do
+            pgconn = PG.connect(processes.pgcat.connection_string("sharded_db", pg_user['username']))
+            expect(pgconn.exec("SELECT 1 + 2")).not_to be_nil
+            Helpers::AuthQuery.exec_in_instances(query: "ALTER USER #{pg_user['username']} WITH ENCRYPTED PASSWORD 'secret2';")
+            expect(pgconn.exec("SELECT 1 + 4")).not_to be_nil
+            Helpers::AuthQuery.exec_in_instances(query: "ALTER USER #{pg_user['username']} WITH ENCRYPTED PASSWORD '#{pg_user['password']}';")
+          end
+
+          it 'allows passwords to be changed and that new password is needed when reconnecting' do
+            pgconn = PG.connect(processes.pgcat.connection_string("sharded_db", pg_user['username']))
+            expect(pgconn.exec("SELECT 1 + 2")).not_to be_nil
+            Helpers::AuthQuery.exec_in_instances(query: "ALTER USER #{pg_user['username']} WITH ENCRYPTED PASSWORD 'secret2';")
+            newconn = PG.connect(processes.pgcat.connection_string("sharded_db", pg_user['username'], 'secret2'))
+            expect(newconn.exec("SELECT 1 + 2")).not_to be_nil
+            Helpers::AuthQuery.exec_in_instances(query: "ALTER USER #{pg_user['username']} WITH ENCRYPTED PASSWORD '#{pg_user['password']}';")
+          end
+        end
+      end
+
+      context 'with wrong parameters' do
+        let(:config) { { 'general' => { 'auth_query' => 'SELECT 1', 'auth_query_user' => 'wrong_user', 'auth_query_password' => 'wrong' } } }
+
+        context 'and with clear text passwords set' do
+          it "it uses local passwords" do
+            conn = PG.connect(processes.pgcat.connection_string("sharded_db", pg_user['username'], pg_user['password']))
+
+            expect(conn.async_exec("SELECT 1 + 2")).not_to be_nil
+          end
+        end
+
+        context 'and with cleartext passwords not set' do
+          let(:config_user) { { 'username' => 'sharding_user' } }
+          it "it fails to start as it cannot authenticate against servers" do
+            @failing_process = true
+            expect { PG.connect(processes.pgcat.connection_string("sharded_db", pg_user['username'], pg_user['password'])) }.to raise_error(StandardError, /Error trying to obtain password from auth_query/ )
+          end
+
+          context 'and we fix the issue and reload' do
+            let(:wait_until_ready) { false }
+
+            it 'fails in the beginning but starts working after reloading config' do
+              connection_string = processes.pgcat.connection_string("sharded_db", pg_user['username'], pg_user['password'])
+              while !(processes.pgcat.logs =~ /Waiting for clients/) do
+                sleep 0.5
+              end
+
+              expect { PG.connect(connection_string)}.to raise_error(PG::ConnectionBad)
+              expect(processes.pgcat.logs).to match(/Error trying to obtain password from auth_query/)
+
+              current_config = processes.pgcat.current_config
+              config = { 'general' => { 'auth_query' => "SELECT * FROM public.user_lookup('$1');", 'auth_query_user' => 'md5_auth_user', 'auth_query_password' => 'secret' } }
+              processes.pgcat.update_config(current_config.deep_merge(config))
+              processes.pgcat.reload_config
+
+              conn = nil
+              expect { conn = PG.connect(connection_string)}.not_to raise_error
+              expect(conn.async_exec("SELECT 1 + 2")).not_to be_nil
+            end
+          end
+        end
+      end
+    end
+
+    context 'with per pool configuration' do
+      around(:example) do |example|
+
+        # Set up auth query
+        Helpers::AuthQuery.set_up_auth_query_for_user(
+          user: 'md5_auth_user',
+          password: 'secret'
+        );
+
+        Helpers::AuthQuery.set_up_auth_query_for_user(
+          user: 'md5_auth_user1',
+          password: 'secret',
+          database: 'shard1'
+        );
+
+        example.run
+
+        # Tear down auth query
+        Helpers::AuthQuery.tear_down_auth_query_for_user(
+          user: 'md5_auth_user',
+          password: 'secret'
+        );
+
+        Helpers::AuthQuery.tear_down_auth_query_for_user(
+          user: 'md5_auth_user1',
+          password: 'secret',
+          database: 'shard1'
+        );
+      end
+
+      context 'with correct parameters' do
+        let(:processes) { Helpers::AuthQuery.two_pools_auth_query(pool_names: ["sharded_db0", "sharded_db1"], pg_user: pg_user, config_user: config_user, extra_conf: config ) }
+        let(:config) {
+          { 'pools' =>
+            {
+              'sharded_db0' => {
+                'auth_query' => "SELECT * FROM public.user_lookup('$1');",
+                'auth_query_user' => 'md5_auth_user',
+                'auth_query_password' => 'secret'
+              },
+              'sharded_db1' => {
+                'auth_query' => "SELECT * FROM public.user_lookup('$1');",
+                'auth_query_user' => 'md5_auth_user1',
+                'auth_query_password' => 'secret'
+              },
+            }
+          }
+        } 
+
+        context 'and with cleartext passwords set' do
+          it 'it uses local passwords' do
+            conn = PG.connect(processes.pgcat.connection_string("sharded_db0", pg_user['username'], pg_user['password']))
+            expect(conn.exec("SELECT 1 + 2")).not_to be_nil
+            conn = PG.connect(processes.pgcat.connection_string("sharded_db1", pg_user['username'], pg_user['password']))
+            expect(conn.exec("SELECT 1 + 2")).not_to be_nil
+          end
+        end
+
+        context 'and with cleartext passwords not set' do
+          let(:config_user) { { 'username' => 'sharding_user' } }
+
+          it 'it uses obtained passwords' do
+            connection_string = processes.pgcat.connection_string("sharded_db0", pg_user['username'], pg_user['password'])
+            conn = PG.connect(connection_string)
+            expect(conn.async_exec("SELECT 1 + 2")).not_to be_nil
+            connection_string = processes.pgcat.connection_string("sharded_db1", pg_user['username'], pg_user['password'])
+            conn = PG.connect(connection_string)
+            expect(conn.async_exec("SELECT 1 + 2")).not_to be_nil
+          end
+        end
+
+      end
+    end
+  end
+end

tests/ruby/helpers/auth_query_helper.rb

@@ -0,0 +1,173 @@
+module Helpers
+  module AuthQuery
+    def self.single_shard_auth_query(
+          pg_user:,
+          config_user:,
+          pool_name:,
+          extra_conf: {},
+          log_level: 'debug',
+          wait_until_ready: true
+        )
+
+      user = {
+        "pool_size" => 10,
+        "statement_timeout" => 0,
+      }
+
+      pgcat = PgcatProcess.new(log_level)
+      pgcat_cfg = pgcat.current_config.deep_merge(extra_conf)
+
+      primary  = PgInstance.new(5432,  pg_user["username"], pg_user["password"], "shard0")
+      replica  = PgInstance.new(10432, pg_user["username"], pg_user["password"], "shard0")
+
+      # Main proxy configs
+      pgcat_cfg["pools"] = {
+        "#{pool_name}" => {
+          "default_role" => "any",
+          "pool_mode" => "transaction",
+          "load_balancing_mode" => "random",
+          "primary_reads_enabled" => false,
+          "query_parser_enabled" => false,
+          "sharding_function" => "pg_bigint_hash",
+          "shards" => {
+            "0" => {
+              "database" => "shard0",
+              "servers" => [
+                ["localhost", primary.port.to_s, "primary"],
+                ["localhost", replica.port.to_s, "replica"],
+              ]
+            },
+          },
+          "users" => { "0" => user.merge(config_user) }
+        }
+      }
+      pgcat_cfg["general"]["port"] = pgcat.port
+      pgcat.update_config(pgcat_cfg)
+      pgcat.start
+      
+      pgcat.wait_until_ready(
+        pgcat.connection_string(
+          "sharded_db",
+          pg_user['username'],
+          pg_user['password']
+        )
+      ) if wait_until_ready
+
+      OpenStruct.new.tap do |struct|
+        struct.pgcat = pgcat
+        struct.primary = primary
+        struct.replicas = [replica]
+        struct.all_databases = [primary]
+      end
+    end
+
+    def self.two_pools_auth_query(
+          pg_user:,
+          config_user:,
+          pool_names:,
+          extra_conf: {},
+          log_level: 'debug'
+        )
+
+      user = {
+        "pool_size" => 10,
+        "statement_timeout" => 0,
+      }
+
+      pgcat = PgcatProcess.new(log_level)
+      pgcat_cfg = pgcat.current_config
+
+      primary  = PgInstance.new(5432,  pg_user["username"], pg_user["password"], "shard0")
+      replica  = PgInstance.new(10432, pg_user["username"], pg_user["password"], "shard0")
+
+      pool_template = Proc.new do |database|
+        {
+          "default_role" => "any",
+          "pool_mode" => "transaction",
+          "load_balancing_mode" => "random",
+          "primary_reads_enabled" => false,
+          "query_parser_enabled" => false,
+          "sharding_function" => "pg_bigint_hash",
+          "shards" => {
+            "0" => {
+              "database" => database,
+              "servers" => [
+                ["localhost", primary.port.to_s, "primary"],
+                ["localhost", replica.port.to_s, "replica"],
+              ]
+            },
+          },
+          "users" => { "0" => user.merge(config_user) }
+        }                                  
+      end
+      # Main proxy configs
+      pgcat_cfg["pools"] = {
+        "#{pool_names[0]}" => pool_template.call("shard0"),
+        "#{pool_names[1]}" => pool_template.call("shard1")
+      }
+
+      pgcat_cfg["general"]["port"] = pgcat.port
+      pgcat.update_config(pgcat_cfg.deep_merge(extra_conf))
+      pgcat.start
+      
+      pgcat.wait_until_ready(pgcat.connection_string("sharded_db0", pg_user['username'], pg_user['password']))
+
+      OpenStruct.new.tap do |struct|
+        struct.pgcat = pgcat
+        struct.primary = primary
+        struct.replicas = [replica]
+        struct.all_databases = [primary]
+      end
+    end
+
+    def self.create_query_auth_function(user)
+      return <<-SQL
+CREATE OR REPLACE FUNCTION public.user_lookup(in i_username text, out uname text, out phash text)
+RETURNS record AS $$
+BEGIN
+    SELECT usename, passwd FROM pg_catalog.pg_shadow
+    WHERE usename = i_username INTO uname, phash;
+    RETURN;
+END;
+$$ LANGUAGE plpgsql SECURITY DEFINER;
+
+GRANT EXECUTE ON FUNCTION public.user_lookup(text) TO #{user};
+SQL
+    end
+
+    def self.exec_in_instances(query:, instance_ports: [ 5432, 10432 ], database: 'postgres', user: 'postgres', password: 'postgres')
+      instance_ports.each do |port|
+        c = PG.connect("postgres://#{user}:#{password}@localhost:#{port}/#{database}")
+        c.exec(query)
+        c.close
+      end
+    end
+
+    def self.set_up_auth_query_for_user(user:, password:, instance_ports: [ 5432, 10432 ], database: 'shard0' )
+      instance_ports.each do |port|
+        connection = PG.connect("postgres://postgres:postgres@localhost:#{port}/#{database}")
+        connection.exec(self.drop_query_auth_function(user)) rescue PG::UndefinedFunction
+        connection.exec("DROP ROLE #{user}") rescue PG::UndefinedObject
+        connection.exec("CREATE ROLE #{user} ENCRYPTED PASSWORD '#{password}' LOGIN;")
+        connection.exec(self.create_query_auth_function(user))
+        connection.close
+      end
+    end
+
+    def self.tear_down_auth_query_for_user(user:, password:, instance_ports: [ 5432, 10432 ], database: 'shard0' )
+      instance_ports.each do |port|
+        connection = PG.connect("postgres://postgres:postgres@localhost:#{port}/#{database}")
+        connection.exec(self.drop_query_auth_function(user)) rescue PG::UndefinedFunction
+        connection.exec("DROP ROLE #{user}")
+        connection.close
+      end
+    end
+
+    def self.drop_query_auth_function(user)
+      return <<-SQL
+REVOKE ALL ON FUNCTION public.user_lookup(text) FROM public, #{user};
+DROP FUNCTION public.user_lookup(in i_username text, out uname text, out phash text);
+SQL
+    end
+  end
+end

tests/ruby/helpers/pg_instance.rb

@@ -38,6 +38,8 @@ class PgInstance
   def reset
     reset_toxics
     reset_stats
+    drop_connections
+    sleep 0.1
   end
 
   def toxiproxy
@@ -66,12 +68,22 @@ class PgInstance
 
   def reset_toxics
     Toxiproxy[@toxiproxy_name].toxics.each(&:destroy)
+    sleep 0.1
   end
 
   def reset_stats
     with_connection { |c| c.async_exec("SELECT pg_stat_statements_reset()") }
   end
 
+  def drop_connections
+    username = with_connection { |c| c.async_exec("SELECT current_user")[0]["current_user"] }
+    with_connection { |c| c.async_exec("SELECT pg_terminate_backend(pid) FROM pg_stat_activity WHERE pid <> pg_backend_pid() AND usename='#{username}'") }
+  end
+
+  def count_connections
+    with_connection { |c| c.async_exec("SELECT COUNT(*) as count FROM pg_stat_activity")[0]["count"].to_i }
+  end
+
   def count_query(query)
     with_connection { |c| c.async_exec("SELECT SUM(calls) FROM pg_stat_statements WHERE query = '#{query}'")[0]["sum"].to_i }
   end

tests/ruby/helpers/pg_socket.rb

@@ -0,0 +1,259 @@
+require 'socket'
+require 'digest/md5'
+
+BACKEND_MESSAGE_CODES = {
+  'Z' => "ReadyForQuery",
+  'C' => "CommandComplete",
+  'T' => "RowDescription",
+  'D' => "DataRow",
+  '1' => "ParseComplete",
+  '2' => "BindComplete",
+  'E' => "ErrorResponse",
+  's' => "PortalSuspended",
+}
+
+class PostgresSocket
+  def initialize(host, port)
+    @port = port
+    @host = host
+    @socket = TCPSocket.new @host, @port
+    @parameters = {}
+    @verbose = true
+  end
+
+  def send_md5_password_message(username, password, salt)
+    m = Digest::MD5.hexdigest(password + username)
+    m = Digest::MD5.hexdigest(m + salt.map(&:chr).join(""))
+    m = 'md5' + m
+    bytes = (m.split("").map(&:ord) + [0]).flatten
+    message_size = bytes.count + 4
+
+    message = []
+
+    message << 'p'.ord
+    message << [message_size].pack('l>').unpack('CCCC') # 4
+    message << bytes
+    message.flatten!
+
+
+    @socket.write(message.pack('C*'))
+  end
+
+  def send_startup_message(username, database, password)
+    message = []
+
+    message << [196608].pack('l>').unpack('CCCC') # 4
+    message << "user".split('').map(&:ord) # 4, 8
+    message << 0 # 1, 9
+    message << username.split('').map(&:ord) # 2, 11
+    message << 0 # 1, 12
+    message << "database".split('').map(&:ord) # 8, 20
+    message << 0 # 1, 21
+    message << database.split('').map(&:ord) # 2, 23
+    message << 0 # 1, 24
+    message << 0 # 1, 25
+    message.flatten!
+
+    total_message_size = message.size + 4
+
+    message_len = [total_message_size].pack('l>').unpack('CCCC')
+
+    @socket.write([message_len + message].flatten.pack('C*'))
+
+    sleep 0.1
+
+    read_startup_response(username, password)
+  end
+
+  def read_startup_response(username, password)
+    message_code, message_len = @socket.recv(5).unpack("al>")
+    while message_code == 'R'
+      auth_code = @socket.recv(4).unpack('l>').pop
+      case auth_code
+      when 5 # md5
+        salt = @socket.recv(4).unpack('CCCC')
+        send_md5_password_message(username, password, salt)
+        message_code, message_len = @socket.recv(5).unpack("al>")
+      when 0 # trust
+        break
+      end
+    end
+    loop do
+      message_code, message_len = @socket.recv(5).unpack("al>")
+      if message_code == 'Z'
+        @socket.recv(1).unpack("a") # most likely I
+        break # We are good to go
+      end
+      if message_code == 'S'
+        actual_message = @socket.recv(message_len - 4).unpack("C*")
+        k,v = actual_message.pack('U*').split(/\x00/)
+        @parameters[k] = v
+      end
+      if message_code == 'K'
+        process_id, secret_key = @socket.recv(message_len - 4).unpack("l>l>")
+        @parameters["process_id"] = process_id
+        @parameters["secret_key"] = secret_key
+      end
+    end
+    return @parameters
+  end
+
+  def cancel_query
+    socket = TCPSocket.new @host, @port
+    process_key = @parameters["process_id"]
+    secret_key =  @parameters["secret_key"]
+    message = []
+    message << [16].pack('l>').unpack('CCCC') # 4
+    message << [80877102].pack('l>').unpack('CCCC') # 4
+    message << [process_key.to_i].pack('l>').unpack('CCCC') # 4
+    message << [secret_key.to_i].pack('l>').unpack('CCCC') # 4
+    message.flatten!
+    socket.write(message.flatten.pack('C*'))
+    socket.close
+    log "[F] Sent CancelRequest message"
+  end
+
+  def send_query_message(query)
+    query_size = query.length
+    message_size = 1 + 4 + query_size
+    message = []
+    message << "Q".ord
+    message << [message_size].pack('l>').unpack('CCCC') # 4
+    message << query.split('').map(&:ord) # 2, 11
+    message << 0 # 1, 12
+    message.flatten!
+    @socket.write(message.flatten.pack('C*'))
+    log "[F] Sent Q message (#{query})"
+  end
+
+  def send_parse_message(query)
+    query_size = query.length
+    message_size = 2 + 2 + 4 + query_size
+    message = []
+    message << "P".ord
+    message << [message_size].pack('l>').unpack('CCCC') # 4
+    message << 0 # unnamed statement
+    message << query.split('').map(&:ord) # 2, 11
+    message << 0 # 1, 12
+    message << [0, 0]
+    message.flatten!
+    @socket.write(message.flatten.pack('C*'))
+    log "[F] Sent P message (#{query})"
+  end
+
+  def send_bind_message
+    message = []
+    message << "B".ord
+    message << [12].pack('l>').unpack('CCCC') # 4
+    message << 0 # unnamed statement
+    message << 0 # unnamed statement
+    message << [0, 0] # 2
+    message << [0, 0] # 2
+    message << [0, 0] # 2
+    message.flatten!
+    @socket.write(message.flatten.pack('C*'))
+    log "[F] Sent B message"
+  end
+
+  def send_describe_message(mode)
+    message = []
+    message << "D".ord
+    message << [6].pack('l>').unpack('CCCC') # 4
+    message << mode.ord
+    message << 0 # unnamed statement
+    message.flatten!
+    @socket.write(message.flatten.pack('C*'))
+    log "[F] Sent D message"
+  end
+
+  def send_execute_message(limit=0)
+    message = []
+    message << "E".ord
+    message << [9].pack('l>').unpack('CCCC') # 4
+    message << 0 # unnamed statement
+    message << [limit].pack('l>').unpack('CCCC') # 4
+    message.flatten!
+    @socket.write(message.flatten.pack('C*'))
+    log "[F] Sent E message"
+  end
+
+  def send_sync_message
+    message = []
+    message << "S".ord
+    message << [4].pack('l>').unpack('CCCC') # 4
+    message.flatten!
+    @socket.write(message.flatten.pack('C*'))
+    log "[F] Sent S message"
+  end
+
+  def send_copydone_message
+    message = []
+    message << "c".ord
+    message << [4].pack('l>').unpack('CCCC') # 4
+    message.flatten!
+    @socket.write(message.flatten.pack('C*'))
+    log "[F] Sent c message"
+  end
+
+  def send_copyfail_message
+    message = []
+    message << "f".ord
+    message << [5].pack('l>').unpack('CCCC') # 4
+    message << 0
+    message.flatten!
+    @socket.write(message.flatten.pack('C*'))
+    log "[F] Sent f message"
+  end
+
+  def send_flush_message
+    message = []
+    message << "H".ord
+    message << [4].pack('l>').unpack('CCCC') # 4
+    message.flatten!
+    @socket.write(message.flatten.pack('C*'))
+    log "[F] Sent H message"
+  end
+
+  def read_from_server()
+    output_messages = []
+    retry_count = 0
+    message_code = nil
+    message_len = 0
+    loop do
+      begin
+        message_code, message_len = @socket.recv_nonblock(5).unpack("al>")
+      rescue IO::WaitReadable
+        return output_messages if retry_count > 50
+
+        retry_count += 1
+        sleep(0.01)
+        next
+      end
+      message = {
+        code: message_code,
+        len: message_len,
+        bytes: []
+      }
+      log "[B] #{BACKEND_MESSAGE_CODES[message_code] || ('UnknownMessage(' + message_code + ')')}"
+
+      actual_message_length = message_len - 4
+      if actual_message_length > 0
+        message[:bytes] = @socket.recv(message_len - 4).unpack("C*")
+        log "\t#{message[:bytes].join(",")}"
+        log "\t#{message[:bytes].map(&:chr).join(" ")}"
+      end
+      output_messages << message
+      return output_messages if message_code == 'Z'
+    end
+  end
+
+  def log(msg)
+    return unless @verbose
+
+    puts msg
+  end
+
+  def close
+    @socket.close
+  end
+end

tests/ruby/helpers/pgcat_helper.rb

@@ -2,10 +2,18 @@ require 'json'
 require 'ostruct'
 require_relative 'pgcat_process'
 require_relative 'pg_instance'
+require_relative 'pg_socket'
+
+class ::Hash
+    def deep_merge(second)
+        merger = proc { |key, v1, v2| Hash === v1 && Hash === v2 ? v1.merge(v2, &merger) : v2 }
+        self.merge(second, &merger)
+    end
+end
 
 module Helpers
   module Pgcat
-    def self.three_shard_setup(pool_name, pool_size, pool_mode="transaction")
+    def self.three_shard_setup(pool_name, pool_size, pool_mode="transaction", lb_mode="random", log_level="info")
       user = {
         "password" => "sharding_user",
         "pool_size" => pool_size,
@@ -13,7 +21,7 @@ module Helpers
         "username" => "sharding_user"
       }
 
-      pgcat    = PgcatProcess.new("info")
+      pgcat    = PgcatProcess.new(log_level)
       primary0 = PgInstance.new(5432, user["username"], user["password"], "shard0")
       primary1 = PgInstance.new(7432, user["username"], user["password"], "shard1")
       primary2 = PgInstance.new(8432, user["username"], user["password"], "shard2")
@@ -23,8 +31,10 @@ module Helpers
         "#{pool_name}" => {
           "default_role" => "any",
           "pool_mode" => pool_mode,
-          "primary_reads_enabled" => false,
-          "query_parser_enabled" => false,
+          "load_balancing_mode" => lb_mode,
+          "primary_reads_enabled" => true,
+          "query_parser_enabled" => true,
+          "automatic_sharding_key" => "data.id",
           "sharding_function" => "pg_bigint_hash",
           "shards" => {
             "0" => { "database" => "shard0", "servers" => [["localhost", primary0.port.to_s, "primary"]] },
@@ -46,7 +56,7 @@ module Helpers
       end
     end
 
-    def self.single_instance_setup(pool_name, pool_size, pool_mode="transaction")
+    def self.single_instance_setup(pool_name, pool_size, pool_mode="transaction", lb_mode="random", log_level="trace")
       user = {
         "password" => "sharding_user",
         "pool_size" => pool_size,
@@ -54,7 +64,7 @@ module Helpers
         "username" => "sharding_user"
       }
 
-      pgcat = PgcatProcess.new("trace")
+      pgcat = PgcatProcess.new(log_level)
       pgcat_cfg = pgcat.current_config
 
       primary  = PgInstance.new(5432, user["username"], user["password"], "shard0")
@@ -64,6 +74,7 @@ module Helpers
         "#{pool_name}" => {
           "default_role" => "primary",
           "pool_mode" => pool_mode,
+          "load_balancing_mode" => lb_mode,
           "primary_reads_enabled" => false,
           "query_parser_enabled" => false,
           "sharding_function" => "pg_bigint_hash",
@@ -90,7 +101,7 @@ module Helpers
       end
     end
 
-    def self.single_shard_setup(pool_name, pool_size, pool_mode="transaction")
+    def self.single_shard_setup(pool_name, pool_size, pool_mode="transaction", lb_mode="random", log_level="info")
       user = {
         "password" => "sharding_user",
         "pool_size" => pool_size,
@@ -98,7 +109,7 @@ module Helpers
         "username" => "sharding_user"
       }
 
-      pgcat = PgcatProcess.new("info")
+      pgcat = PgcatProcess.new(log_level)
       pgcat_cfg = pgcat.current_config
 
       primary  = PgInstance.new(5432, user["username"], user["password"], "shard0")
@@ -111,6 +122,7 @@ module Helpers
         "#{pool_name}" => {
           "default_role" => "any",
           "pool_mode" => pool_mode,
+          "load_balancing_mode" => lb_mode,
           "primary_reads_enabled" => false,
           "query_parser_enabled" => false,
           "sharding_function" => "pg_bigint_hash",

tests/ruby/helpers/pgcat_process.rb

@@ -8,7 +8,11 @@ class PgcatProcess
   attr_reader :pid
 
   def self.finalize(pid, log_filename, config_filename)
-    `kill #{pid}`
+    if pid
+      Process.kill("TERM", pid)
+      Process.wait(pid)
+    end
+
     File.delete(config_filename) if File.exist?(config_filename)
     File.delete(log_filename) if File.exist?(log_filename)
   end
@@ -20,7 +24,13 @@ class PgcatProcess
     @log_filename = "/tmp/pgcat_log_#{SecureRandom.urlsafe_base64}.log"
     @config_filename = "/tmp/pgcat_cfg_#{SecureRandom.urlsafe_base64}.toml"
 
-    @command = "../../target/debug/pgcat #{@config_filename}"
+    command_path = if ENV['CARGO_TARGET_DIR'] then
+                     "#{ENV['CARGO_TARGET_DIR']}/debug/pgcat"
+                   else
+                     '../../target/debug/pgcat'
+                   end
+
+    @command = "#{command_path} #{@config_filename}"
 
     FileUtils.cp("../../pgcat.toml", @config_filename)
     cfg = current_config
@@ -38,34 +48,40 @@ class PgcatProcess
     @original_config = current_config
     output_to_write = TOML::Generator.new(config_hash).body
     output_to_write = output_to_write.gsub(/,\s*["|'](\d+)["|']\s*,/, ',\1,')
+    output_to_write = output_to_write.gsub(/,\s*["|'](\d+)["|']\s*\]/, ',\1]')
     File.write(@config_filename, output_to_write)
   end
 
   def current_config
-    old_cfg = File.read(@config_filename)
-    loadable_string = old_cfg.gsub(/,\s*(\d+)\s*,/, ', "\1",')
+    loadable_string = File.read(@config_filename)
+    loadable_string = loadable_string.gsub(/,\s*(\d+)\s*,/,  ', "\1",')
+    loadable_string = loadable_string.gsub(/,\s*(\d+)\s*\]/, ', "\1"]')
     TOML.load(loadable_string)
   end
 
   def reload_config
     `kill -s HUP #{@pid}`
-    sleep 0.1
+    sleep 0.5
   end
 
   def start
     raise StandardError, "Process is already started" unless @pid.nil?
     @pid = Process.spawn(@env, @command, err: @log_filename, out: @log_filename)
+    Process.detach(@pid)
     ObjectSpace.define_finalizer(@log_filename, proc { PgcatProcess.finalize(@pid, @log_filename, @config_filename) })
 
     return self
   end
 
-  def wait_until_ready
+  def wait_until_ready(connection_string = nil)
     exc = nil
     10.times do
-      PG::connect(example_connection_string).close
+      Process.kill 0, @pid
+      PG::connect(connection_string || example_connection_string).close
 
       return self
+    rescue Errno::ESRCH
+      raise StandardError, "Process #{@pid} died. #{logs}"
     rescue => e
       exc = e
       sleep(0.5)
@@ -75,8 +91,11 @@ class PgcatProcess
   end
 
   def stop
-    `kill #{@pid}`
-    sleep 0.1
+    return unless @pid
+
+    Process.kill("TERM", @pid)
+    Process.wait(@pid)
+    @pid = nil
   end
 
   def shutdown
@@ -93,13 +112,10 @@ class PgcatProcess
     "postgresql://#{username}:#{password}@0.0.0.0:#{@port}/pgcat"
   end
 
-  def connection_string(pool_name, username)
+  def connection_string(pool_name, username, password = nil)
     cfg = current_config
-
     user_idx, user_obj = cfg["pools"][pool_name]["users"].detect { |k, user| user["username"] == username }
-    password = user_obj["password"]
-
-    "postgresql://#{username}:#{password}@0.0.0.0:#{@port}/#{pool_name}"
+    "postgresql://#{username}:#{password || user_obj["password"]}@0.0.0.0:#{@port}/#{pool_name}"
   end
 
   def example_connection_string

tests/ruby/load_balancing_spec.rb

@@ -1,7 +1,7 @@
 # frozen_string_literal: true
 require_relative 'spec_helper'
 
-describe "Load Balancing" do
+describe "Random Load Balancing" do
   let(:processes) { Helpers::Pgcat.single_shard_setup("sharded_db", 5) }
   after do
     processes.all_databases.map(&:reset)
@@ -46,7 +46,110 @@ describe "Load Balancing" do
         end
       end
 
-      expect(failed_count).to eq(2)
+      processes.all_databases.each do |instance|
+        queries_routed = instance.count_select_1_plus_2
+        if processes.replicas[0..1].include?(instance)
+          expect(queries_routed).to eq(0)
+        else
+          expect(queries_routed).to be_within(expected_share * MARGIN_OF_ERROR).of(expected_share)
+        end
+      end
+    end
+  end
+end
+
+describe "Least Outstanding Queries Load Balancing" do
+  let(:processes) { Helpers::Pgcat.single_shard_setup("sharded_db", 1, "transaction", "loc") }
+  after do
+    processes.all_databases.map(&:reset)
+    processes.pgcat.shutdown
+  end
+
+  context "under homogeneous load" do
+    it "balances query volume between all instances" do
+      conn = PG.connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
+
+      query_count = QUERY_COUNT
+      expected_share = query_count / processes.all_databases.count
+      failed_count = 0
+
+      query_count.times do
+        conn.async_exec("SELECT 1 + 2")
+      rescue
+        failed_count += 1
+      end
+
+      expect(failed_count).to eq(0)
+      processes.all_databases.map(&:count_select_1_plus_2).each do |instance_share|
+        expect(instance_share).to be_within(expected_share * MARGIN_OF_ERROR).of(expected_share)
+      end
+    end
+  end
+
+  context "under heterogeneous load" do
+    xit "balances query volume between all instances based on how busy they are" do
+      slow_query_count = 2
+      threads = Array.new(slow_query_count) do
+        Thread.new do
+          conn = PG.connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
+          conn.async_exec("BEGIN")
+        end
+      end
+
+      conn = PG.connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
+
+      query_count = QUERY_COUNT
+      expected_share = query_count / (processes.all_databases.count - slow_query_count)
+      failed_count = 0
+
+      query_count.times do
+        conn.async_exec("SELECT 1 + 2")
+      rescue
+        failed_count += 1
+      end
+
+      expect(failed_count).to eq(0)
+      # Under LOQ, we expect replicas running the slow pg_sleep
+      # to get no selects
+      expect(
+        processes.
+          all_databases.
+          map(&:count_select_1_plus_2).
+          count { |instance_share| instance_share == 0 }
+      ).to eq(slow_query_count)
+
+      # We also expect the quick queries to be spread across
+      # the idle servers only
+      processes.
+        all_databases.
+        map(&:count_select_1_plus_2).
+        reject { |instance_share| instance_share == 0 }.
+        each do |instance_share|
+          expect(instance_share).to be_within(expected_share * MARGIN_OF_ERROR).of(expected_share)
+      end
+
+      threads.map(&:join)
+    end
+  end
+
+  context "when some replicas are down" do
+    it "balances query volume between working instances" do
+      conn = PG.connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
+      expected_share = QUERY_COUNT / (processes.all_databases.count - 2)
+      failed_count = 0
+
+      processes[:replicas][0].take_down do
+        processes[:replicas][1].take_down do
+          QUERY_COUNT.times do
+            conn.async_exec("SELECT 1 + 2")
+          rescue
+            conn = PG.connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
+            failed_count += 1
+          end
+        end
+      end
+
+      expect(failed_count).to be <= 2
       processes.all_databases.each do |instance|
         queries_routed = instance.count_select_1_plus_2
         if processes.replicas[0..1].include?(instance)

tests/ruby/mirrors_spec.rb

@@ -0,0 +1,90 @@
+# frozen_string_literal: true
+require 'uri'
+require_relative 'spec_helper'
+
+describe "Query Mirroing" do
+  let(:processes) { Helpers::Pgcat.single_instance_setup("sharded_db", 10) }
+  let(:mirror_pg) { PgInstance.new(8432, "sharding_user", "sharding_user", "shard2")}
+  let(:pgcat_conn_str) { processes.pgcat.connection_string("sharded_db", "sharding_user") }
+  let(:mirror_host) { "localhost" }
+
+  before do
+    new_configs = processes.pgcat.current_config
+    new_configs["pools"]["sharded_db"]["shards"]["0"]["mirrors"] = [
+      [mirror_host, mirror_pg.port.to_s, "0"],
+      [mirror_host, mirror_pg.port.to_s, "0"],
+      [mirror_host, mirror_pg.port.to_s, "0"],
+    ]
+    processes.pgcat.update_config(new_configs)
+    processes.pgcat.reload_config
+  end
+
+  after do
+    processes.all_databases.map(&:reset)
+    mirror_pg.reset
+    processes.pgcat.shutdown
+  end
+
+  xit "can mirror a query" do
+    conn = PG.connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
+    runs = 15
+    runs.times { conn.async_exec("SELECT 1 + 2") }
+    sleep 0.5
+    expect(processes.all_databases.first.count_select_1_plus_2).to eq(runs)
+    expect(mirror_pg.count_select_1_plus_2).to eq(runs * 3)
+  end
+
+  context "when main server connection is closed" do
+    it "closes the mirror connection" do
+      baseline_count = processes.all_databases.first.count_connections
+      5.times do |i|
+        # Force pool cycling to detect zombie mirror connections
+        new_configs = processes.pgcat.current_config
+        new_configs["pools"]["sharded_db"]["idle_timeout"] = 5000 + i
+        new_configs["pools"]["sharded_db"]["shards"]["0"]["mirrors"] = [
+          [mirror_host, mirror_pg.port.to_s, "0"],
+          [mirror_host, mirror_pg.port.to_s, "0"],
+          [mirror_host, mirror_pg.port.to_s, "0"],
+        ]
+        processes.pgcat.update_config(new_configs)
+        processes.pgcat.reload_config
+      end
+      conn = PG.connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
+      conn.async_exec("SELECT 1 + 2")
+      sleep 0.5
+      # Expect same number of connection even after pool cycling
+      expect(processes.all_databases.first.count_connections).to be < baseline_count + 2
+    end
+  end
+
+  xcontext "when mirror server goes down temporarily" do
+    it "continues to transmit queries after recovery" do
+      conn = PG.connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
+      mirror_pg.take_down do
+        conn.async_exec("SELECT 1 + 2")
+        sleep 0.1
+      end
+      10.times { conn.async_exec("SELECT 1 + 2") }
+      sleep 1
+      expect(mirror_pg.count_select_1_plus_2).to be >= 2
+    end
+  end
+
+  context "when a mirror is down" do
+    let(:mirror_host) { "badhost" }
+
+    it "does not fail to send the main query" do
+      conn = PG.connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
+      # No Errors here
+      conn.async_exec("SELECT 1 + 2")
+      expect(processes.all_databases.first.count_select_1_plus_2).to eq(1)
+    end
+
+    it "does not fail to send the main query (even after thousands of mirror attempts)" do
+      conn = PG.connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
+      # No Errors here
+      1000.times { conn.async_exec("SELECT 1 + 2") }
+      expect(processes.all_databases.first.count_select_1_plus_2).to eq(1000)
+    end
+  end
+end

tests/ruby/misc_spec.rb

@@ -8,6 +8,100 @@ describe "Miscellaneous" do
     processes.pgcat.shutdown
   end
 
+  context "when adding then removing instance using RELOAD" do
+    it "works correctly" do
+      admin_conn = PG::connect(processes.pgcat.admin_connection_string)
+
+      current_configs = processes.pgcat.current_config
+      correct_count = current_configs["pools"]["sharded_db"]["shards"]["0"]["servers"].count
+      expect(admin_conn.async_exec("SHOW DATABASES").count).to eq(correct_count)
+
+      extra_replica = current_configs["pools"]["sharded_db"]["shards"]["0"]["servers"].last.clone
+      extra_replica[0] = "127.0.0.1"
+      current_configs["pools"]["sharded_db"]["shards"]["0"]["servers"] << extra_replica
+
+      processes.pgcat.update_config(current_configs) # with replica added
+      processes.pgcat.reload_config
+      correct_count = current_configs["pools"]["sharded_db"]["shards"]["0"]["servers"].count
+      expect(admin_conn.async_exec("SHOW DATABASES").count).to eq(correct_count)
+
+      current_configs["pools"]["sharded_db"]["shards"]["0"]["servers"].pop
+
+      processes.pgcat.update_config(current_configs) # with replica removed again
+      processes.pgcat.reload_config
+      correct_count = current_configs["pools"]["sharded_db"]["shards"]["0"]["servers"].count
+      expect(admin_conn.async_exec("SHOW DATABASES").count).to eq(correct_count)
+    end
+  end
+
+  context "when removing then adding instance back using RELOAD" do
+    it "works correctly" do
+      admin_conn = PG::connect(processes.pgcat.admin_connection_string)
+
+      current_configs = processes.pgcat.current_config
+      correct_count = current_configs["pools"]["sharded_db"]["shards"]["0"]["servers"].count
+      expect(admin_conn.async_exec("SHOW DATABASES").count).to eq(correct_count)
+
+      removed_replica = current_configs["pools"]["sharded_db"]["shards"]["0"]["servers"].pop
+      processes.pgcat.update_config(current_configs) # with replica removed
+      processes.pgcat.reload_config
+      correct_count = current_configs["pools"]["sharded_db"]["shards"]["0"]["servers"].count
+      expect(admin_conn.async_exec("SHOW DATABASES").count).to eq(correct_count)
+
+      current_configs["pools"]["sharded_db"]["shards"]["0"]["servers"] << removed_replica
+
+      processes.pgcat.update_config(current_configs) # with replica added again
+      processes.pgcat.reload_config
+      correct_count = current_configs["pools"]["sharded_db"]["shards"]["0"]["servers"].count
+      expect(admin_conn.async_exec("SHOW DATABASES").count).to eq(correct_count)
+    end
+  end
+
+  describe "TCP Keepalives" do
+    # Ideally, we should block TCP traffic to the database using
+    # iptables to mimic passive (connection is dropped without a RST packet)
+    # but we cannot do this in CircleCI because iptables requires NET_ADMIN
+    # capability that we cannot enable in CircleCI
+    # Toxiproxy won't work either because it does not block keepalives
+    # so our best bet is to query the OS keepalive params set on the socket
+
+    context "default settings" do
+      it "applies default keepalive settings" do
+        # We query ss command to verify that we have correct keepalive values set
+        # we can only verify the keepalives_idle parameter but that's good enough
+        # example output
+        #Recv-Q Send-Q Local Address:Port  Peer Address:Port Process
+        #0      0          127.0.0.1:60526    127.0.0.1:18432 timer:(keepalive,1min59sec,0)
+        #0      0          127.0.0.1:60664    127.0.0.1:19432 timer:(keepalive,4.123ms,0)
+
+        port_search_criteria = processes.all_databases.map { |d| "dport = :#{d.port}"}.join(" or ")
+        results = `ss -t4 state established -o -at '( #{port_search_criteria}  )'`.lines
+        results.shift
+        results.each { |line| expect(line).to match(/timer:\(keepalive,.*ms,0\)/) }
+      end
+    end
+
+    context "changed settings" do
+      it "applies keepalive settings from config" do
+        new_configs = processes.pgcat.current_config
+
+        new_configs["general"]["tcp_keepalives_idle"] = 120
+        new_configs["general"]["tcp_keepalives_count"] = 1
+        new_configs["general"]["tcp_keepalives_interval"] = 1
+        processes.pgcat.update_config(new_configs)
+        # We need to kill the old process that was using the default configs
+        processes.pgcat.stop
+        processes.pgcat.start
+        processes.pgcat.wait_until_ready
+
+        port_search_criteria = processes.all_databases.map { |d| "dport = :#{d.port}"}.join(" or ")
+        results = `ss -t4 state established -o -at '( #{port_search_criteria}  )'`.lines
+        results.shift
+        results.each { |line| expect(line).to include("timer:(keepalive,1min") }
+      end
+    end
+  end
+
   describe "Extended Protocol handling" do
     it "does not send packets that client does not expect during extended protocol sequence" do
       new_configs = processes.pgcat.current_config
@@ -189,5 +283,84 @@ describe "Miscellaneous" do
         expect(processes.primary.count_query("DISCARD ALL")).to eq(10)
       end
     end
+
+    context "transaction mode with transactions" do
+      let(:processes) { Helpers::Pgcat.single_shard_setup("sharded_db", 5, "transaction") }
+      it "Does not clear set statement state when declared in a transaction" do
+        10.times do
+          conn = PG::connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
+          conn.async_exec("SET SERVER ROLE to 'primary'")
+          conn.async_exec("BEGIN")
+          conn.async_exec("SET statement_timeout to 1000")
+          conn.async_exec("COMMIT")
+          conn.close
+        end
+        expect(processes.primary.count_query("DISCARD ALL")).to eq(0)
+
+        10.times do
+          conn = PG::connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
+          conn.async_exec("SET SERVER ROLE to 'primary'")
+          conn.async_exec("BEGIN")
+          conn.async_exec("SET LOCAL statement_timeout to 1000")
+          conn.async_exec("COMMIT")
+          conn.close
+        end
+        expect(processes.primary.count_query("DISCARD ALL")).to eq(0)
+      end
+    end
+  end
+
+  describe "Idle client timeout" do
+    context "idle transaction timeout set to 0" do
+      before do
+        current_configs = processes.pgcat.current_config
+        correct_idle_client_transaction_timeout = current_configs["general"]["idle_client_in_transaction_timeout"]
+        puts(current_configs["general"]["idle_client_in_transaction_timeout"])
+  
+        current_configs["general"]["idle_client_in_transaction_timeout"] = 0
+  
+        processes.pgcat.update_config(current_configs) # with timeout 0
+        processes.pgcat.reload_config
+      end
+
+      it "Allow client to be idle in transaction" do
+        conn = PG::connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
+        conn.async_exec("BEGIN")
+        conn.async_exec("SELECT 1")
+        sleep(2)
+        conn.async_exec("COMMIT")
+        conn.close
+      end
+    end
+
+    context "idle transaction timeout set to 500ms" do
+      before do
+        current_configs = processes.pgcat.current_config
+        correct_idle_client_transaction_timeout = current_configs["general"]["idle_client_in_transaction_timeout"]  
+        current_configs["general"]["idle_client_in_transaction_timeout"] = 500
+  
+        processes.pgcat.update_config(current_configs) # with timeout 500
+        processes.pgcat.reload_config
+      end
+
+      it "Allow client to be idle in transaction below timeout" do
+        conn = PG::connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
+        conn.async_exec("BEGIN")
+        conn.async_exec("SELECT 1")
+        sleep(0.4) # below 500ms
+        conn.async_exec("COMMIT")
+        conn.close
+      end
+
+      it "Error when client idle in transaction time exceeds timeout" do
+        conn = PG::connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
+        conn.async_exec("BEGIN")
+        conn.async_exec("SELECT 1")
+        sleep(1) # above 500ms
+        expect{ conn.async_exec("COMMIT") }.to raise_error(PG::SystemError, /idle transaction timeout/) 
+        conn.async_exec("SELECT 1") # should be able to send another query
+        conn.close
+      end
+    end
   end
 end

tests/ruby/protocol_spec.rb

@@ -0,0 +1,155 @@
+# frozen_string_literal: true
+require_relative 'spec_helper'
+
+
+describe "Portocol handling" do
+  let(:processes) { Helpers::Pgcat.single_instance_setup("sharded_db", 1, "session") }
+  let(:sequence) { [] }
+  let(:pgcat_socket) { PostgresSocket.new('localhost', processes.pgcat.port) }
+  let(:pgdb_socket) { PostgresSocket.new('localhost', processes.all_databases.first.port) }
+
+  after do
+    pgdb_socket.close
+    pgcat_socket.close
+    processes.all_databases.map(&:reset)
+    processes.pgcat.shutdown
+  end
+
+  def run_comparison(sequence, socket_a, socket_b)
+    sequence.each do |msg, *args|
+      socket_a.send(msg, *args)
+      socket_b.send(msg, *args)
+
+      compare_messages(
+        socket_a.read_from_server,
+        socket_b.read_from_server
+      )
+    end
+  end
+
+  def compare_messages(msg_arr0, msg_arr1)
+    if msg_arr0.count != msg_arr1.count
+      error_output = []
+
+      error_output << "#{msg_arr0.count} : #{msg_arr1.count}"
+      error_output << "PgCat Messages"
+      error_output += msg_arr0.map { |message| "\t#{message[:code]} - #{message[:bytes].map(&:chr).join(" ")}" }
+      error_output << "PgServer Messages"
+      error_output += msg_arr1.map { |message| "\t#{message[:code]} - #{message[:bytes].map(&:chr).join(" ")}" }
+      error_desc = error_output.join("\n")
+      raise StandardError, "Message count mismatch #{error_desc}"
+    end
+
+    (0..msg_arr0.count - 1).all? do |i|
+      msg0 = msg_arr0[i]
+      msg1 = msg_arr1[i]
+
+      result = [
+        msg0[:code] == msg1[:code],
+        msg0[:len] == msg1[:len],
+        msg0[:bytes] == msg1[:bytes],
+      ].all?
+
+      next result if result
+
+      if result == false
+        error_string = []
+        if msg0[:code] != msg1[:code]
+          error_string << "code #{msg0[:code]} != #{msg1[:code]}"
+        end
+        if msg0[:len] != msg1[:len]
+          error_string << "len #{msg0[:len]} != #{msg1[:len]}"
+        end
+        if msg0[:bytes] != msg1[:bytes]
+          error_string << "bytes #{msg0[:bytes]} != #{msg1[:bytes]}"
+        end
+        err = error_string.join("\n")
+
+        raise StandardError, "Message mismatch #{err}"
+      end
+    end
+  end
+
+  RSpec.shared_examples "at parity with database" do
+    before do
+      pgcat_socket.send_startup_message("sharding_user", "sharded_db", "sharding_user")
+      pgdb_socket.send_startup_message("sharding_user", "shard0", "sharding_user")
+    end
+
+    it "works" do
+      run_comparison(sequence, pgcat_socket, pgdb_socket)
+    end
+  end
+
+  context "Cancel Query" do
+    let(:sequence) {
+      [
+        [:send_query_message, "SELECT pg_sleep(5)"],
+        [:cancel_query]
+      ]
+    }
+
+    it_behaves_like "at parity with database"
+  end
+
+  xcontext "Simple query after parse" do
+    let(:sequence) {
+      [
+        [:send_parse_message, "SELECT 5"],
+        [:send_query_message, "SELECT 1"],
+        [:send_bind_message],
+        [:send_describe_message, "P"],
+        [:send_execute_message],
+        [:send_sync_message],
+      ]
+    }
+
+    # Known to fail due to PgCat not supporting flush
+    it_behaves_like "at parity with database"
+  end
+
+  xcontext "Flush message" do
+    let(:sequence) {
+      [
+        [:send_parse_message, "SELECT 1"],
+        [:send_flush_message]
+      ]
+    }
+
+    # Known to fail due to PgCat not supporting flush
+    it_behaves_like "at parity with database"
+  end
+
+  xcontext "Bind without parse" do
+    let(:sequence) {
+      [
+        [:send_bind_message]
+      ]
+    }
+    # This is known to fail.
+    # Server responds immediately, Proxy buffers the message
+    it_behaves_like "at parity with database"
+  end
+
+  context "Simple message" do
+    let(:sequence) {
+      [[:send_query_message, "SELECT 1"]]
+    }
+
+    it_behaves_like "at parity with database"
+  end
+
+  context "Extended protocol" do
+    let(:sequence) {
+      [
+        [:send_parse_message, "SELECT 1"],
+        [:send_bind_message],
+        [:send_describe_message, "P"],
+        [:send_execute_message],
+        [:send_sync_message],
+      ]
+    }
+
+    it_behaves_like "at parity with database"
+  end
+end

tests/ruby/sharding_spec.rb

@@ -0,0 +1,51 @@
+# frozen_string_literal: true
+require_relative 'spec_helper'
+
+
+describe "Sharding" do
+  let(:processes) { Helpers::Pgcat.three_shard_setup("sharded_db", 5) }
+
+  before do
+    conn = PG.connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
+
+    # Setup the sharding data
+    3.times do |i|
+      conn.exec("SET SHARD TO '#{i}'")
+      conn.exec("DELETE FROM data WHERE id > 0")
+    end
+
+    18.times do |i|
+      i = i + 1
+      conn.exec("SET SHARDING KEY TO '#{i}'")
+      conn.exec("INSERT INTO data (id, value) VALUES (#{i}, 'value_#{i}')")
+    end
+  end
+
+  after do
+
+    processes.all_databases.map(&:reset)
+    processes.pgcat.shutdown
+  end
+
+  describe "automatic routing of extended protocol" do
+    it "can do it" do
+      conn = PG.connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
+      conn.exec("SET SERVER ROLE TO 'auto'")
+
+      18.times do |i|
+        result = conn.exec_params("SELECT * FROM data WHERE id = $1", [i + 1])
+        expect(result.ntuples).to eq(1)
+      end
+    end
+
+    it "can do it with multiple parameters" do
+      conn = PG.connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
+      conn.exec("SET SERVER ROLE TO 'auto'")
+
+      18.times do |i|
+        result = conn.exec_params("SELECT * FROM data WHERE id = $1 AND id = $2", [i + 1, i + 1])
+        expect(result.ntuples).to eq(1)
+      end
+    end
+  end
+end

tests/ruby/spec_helper.rb

@@ -4,7 +4,7 @@ require 'pg'
 require_relative 'helpers/pgcat_helper'
 
 QUERY_COUNT = 300
-MARGIN_OF_ERROR = 0.30
+MARGIN_OF_ERROR = 0.35
 
 def with_captured_stdout_stderr
   sout = STDOUT.clone
@@ -19,3 +19,10 @@ ensure
   STDOUT.reopen(sout)
   STDERR.reopen(serr)
 end
+
+def clients_connected_to_pool(pool_index: 0, processes:)
+  admin_conn = PG::connect(processes.pgcat.admin_connection_string)
+  results = admin_conn.async_exec("SHOW POOLS")[pool_index]
+  admin_conn.close
+  results['cl_idle'].to_i + results['cl_active'].to_i + results['cl_waiting'].to_i
+end

utilities/generate_config_docs.py

@@ -0,0 +1,92 @@
+import re
+import tomli
+
+class DocGenerator:
+    def __init__(self, filename):
+        self.doc = []
+        self.current_section = ""
+        self.current_comment = []
+        self.current_field_name = ""
+        self.current_field_value = []
+        self.current_field_unset = False
+        self.filename = filename
+
+    def write(self):
+        with open("../CONFIG.md", "w") as text_file:
+            text_file.write("# PgCat Configurations \n")
+            for entry in self.doc:
+                if entry["name"] == "__section__":
+                    text_file.write("## `" + entry["section"] + "` Section" + "\n")
+                    text_file.write("\n")
+                    continue
+                text_file.write("### " + entry["name"]+ "\n")
+                text_file.write("```"+ "\n")
+                text_file.write("path: " + entry["fqdn"]+ "\n")
+                text_file.write("default: " + entry["defaults"].strip()+ "\n")
+                if entry["example"] is not None:
+                    text_file.write("example: " + entry["example"].strip()+ "\n")
+                text_file.write("```"+ "\n")
+                text_file.write("\n")
+                text_file.write(entry["comment"]+ "\n")
+                text_file.write("\n")
+
+    def save_entry(self):
+        if len(self.current_field_name) == 0:
+            return
+        if len(self.current_comment) == 0:
+            return
+        self.current_section = self.current_section.replace("sharded_db", "<pool_name>")
+        self.current_section = self.current_section.replace("simple_db", "<pool_name>")
+        self.current_section = self.current_section.replace("users.0", "users.<user_index>")
+        self.current_section = self.current_section.replace("users.1", "users.<user_index>")
+        self.current_section = self.current_section.replace("shards.0", "shards.<shard_index>")
+        self.current_section = self.current_section.replace("shards.1", "shards.<shard_index>")
+        self.doc.append(
+            {
+                "name": self.current_field_name,
+                "fqdn": self.current_section + "." + self.current_field_name,
+                "section": self.current_section,
+                "comment": "\n".join(self.current_comment),
+                "defaults": self.current_field_value if not self.current_field_unset else "<UNSET>",
+                "example": self.current_field_value  if self.current_field_unset  else None
+            }
+        )
+        self.current_comment = []
+        self.current_field_name = ""
+        self.current_field_value = []
+    def parse(self):
+        with open("../pgcat.toml", "r") as f:
+            for line in f.readlines():
+                line = line.strip()
+                if len(line) == 0:
+                    self.save_entry()
+
+                if line.startswith("["):
+                    self.current_section = line[1:-1]
+                    self.current_field_name = "__section__"
+                    self.current_field_unset = False
+                    self.save_entry()
+
+                elif line.startswith("#"):
+                    results = re.search("^#\s*([A-Za-z0-9_]+)\s*=(.+)$", line)
+                    if results is not None:
+                        self.current_field_name = results.group(1)
+                        self.current_field_value = results.group(2)
+                        self.current_field_unset = True
+                        self.save_entry()
+                    else:
+                        self.current_comment.append(line[1:].strip())
+                else:
+                    results = re.search("^\s*([A-Za-z0-9_]+)\s*=(.+)$", line)
+                    if results is None:
+                        continue
+                    self.current_field_name = results.group(1)
+                    self.current_field_value = results.group(2)
+                    self.current_field_unset = False
+                    self.save_entry()
+        self.save_entry()
+        return self
+
+
+DocGenerator("../pgcat.toml").parse().write()
+

utilities/requirements.txt

@@ -0,0 +1 @@
+tomli