noop

In a high availability deployment of PgCat, it is possible that a client may land on a container of PgCat that is very busy with clients and as such the new client might be perpetually stuck in checkout failure loop because all connections are used by other clients. This is specially true in session mode pools with long-lived client connections (e.g. FDW connections).

One way to fix this issue is to close client connections after they encounter some number of checkout failure. This will force the client to hit the Network load balancer again, land on a different process/container, try to checkout a connection on the new process/container. if it fails, it is disconnected and tries with another one.

This mechanism is guaranteed to eventually land on a balanced state where all clients are able to find connections provided that the overall number of connections across all containers matches the number of clients.

I was able to reproduce this issue in a control environment and was able to show this PR is able to fix it.

.circleci/pgcat.toml

@@ -59,6 +59,7 @@ admin_password = "admin_pass"
 # session: one server connection per connected client
 # transaction: one server connection per client transaction
 pool_mode = "transaction"
+prepared_statements_cache_size = 500
 
 # If the client doesn't specify, route traffic to
 # this role by default.
@@ -141,6 +142,7 @@ query_parser_enabled = true
 query_parser_read_write_splitting = true
 primary_reads_enabled = true
 sharding_function = "pg_bigint_hash"
+prepared_statements_cache_size = 500
 
 [pools.simple_db.users.0]
 username = "simple_user"

.circleci/run_tests.sh

@@ -26,6 +26,7 @@ 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
 
 # Start Toxiproxy
+kill -9 $(pgrep toxiproxy) || true
 LOG_LEVEL=error toxiproxy-server &
 sleep 1
 
@@ -106,10 +107,26 @@ cd ../..
 # These tests will start and stop the pgcat server so it will need to be restarted after the tests
 #
 pip3 install -r tests/python/requirements.txt
-python3 tests/python/tests.py || exit 1
+pytest || exit 1
+
+
+#
+# Go tests
+# Starts its own pgcat server
+#
+pushd tests/go
+/usr/local/go/bin/go test || exit 1
+popd
 
 start_pgcat "info"
 
+#
+# Rust tests
+#
+cd tests/rust
+cargo run
+cd ../../
+
 # Admin tests
 export PGPASSWORD=admin_pass
 psql -U admin_user -e -h 127.0.0.1 -p 6432 -d pgbouncer -c 'SHOW STATS' > /dev/null
@@ -161,3 +178,6 @@ killall pgcat -s SIGINT
 
 # Allow for graceful shutdown
 sleep 1
+
+kill -9 $(pgrep toxiproxy)
+sleep 1

.github/dependabot.yml

@@ -10,3 +10,7 @@ updates:
     commit-message:
       prefix: "chore(deps)"
     open-pull-requests-limit: 10
+  - package-ecosystem: "github-actions"
+    directory: "/"
+    schedule:
+      interval: "weekly"

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

@@ -2,7 +2,9 @@ name: Build and Push
 
 on:
   push:
-    branches:    
+    paths:
+      - '!charts/**.md'
+    branches:
       - main
     tags:
       - v*
@@ -21,14 +23,17 @@ jobs:
 
     steps:
       - name: Checkout Repository
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
+
+      - name: Set up QEMU
+        uses: docker/setup-qemu-action@v3
 
       - name: Set up Docker Buildx
-        uses: docker/setup-buildx-action@v2
+        uses: docker/setup-buildx-action@v3
 
       - name: Determine tags
         id: metadata
-        uses: docker/metadata-action@v4
+        uses: docker/metadata-action@v5
         with:
           images: ${{ env.registry }}/${{ env.image-name }}
           tags: |
@@ -40,15 +45,18 @@ jobs:
             type=raw,value=latest,enable={{ is_default_branch }}
 
       - name: Log in to the Container registry
-        uses: docker/login-action@v2.1.0
+        uses: docker/login-action@v3
         with:
           registry: ${{ env.registry }}
           username: ${{ github.actor }}
           password: ${{ secrets.GITHUB_TOKEN }}
 
       - name: Build and push ${{ env.image-name }}
-        uses: docker/build-push-action@v3
+        uses: docker/build-push-action@v6
         with:
+          context: .
+          platforms: linux/amd64,linux/arm64
+          provenance: false
           push: true
           tags: ${{ steps.metadata.outputs.tags }}
           labels: ${{ steps.metadata.outputs.labels }}

.github/workflows/chart-lint-test.yaml

@@ -0,0 +1,50 @@
+name: Lint and Test Charts
+
+on:
+  pull_request:
+    paths:
+      - charts/**
+      - '!charts/**.md'
+jobs:
+  lint-test:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v3.1.0
+        with:
+          fetch-depth: 0
+
+      - name: Set up Helm
+        uses: azure/setup-helm@v3
+        with:
+          version: v3.8.1
+
+      # Python is required because `ct lint` runs Yamale (https://github.com/23andMe/Yamale) and
+      # yamllint (https://github.com/adrienverge/yamllint) which require Python
+      - name: Set up Python
+        uses: actions/setup-python@v5.1.0
+        with:
+          python-version: 3.7
+
+      - name: Set up chart-testing
+        uses: helm/chart-testing-action@v2.2.1
+        with:
+          version: v3.5.1
+
+      - name: Run chart-testing (list-changed)
+        id: list-changed
+        run: |
+          changed=$(ct list-changed --config ct.yaml)
+          if [[ -n "$changed" ]]; then
+            echo "changed=true" >> $GITHUB_OUTPUT
+          fi
+
+      - name: Run chart-testing (lint)
+        run: ct lint --config ct.yaml
+
+      - name: Create kind cluster
+        uses: helm/kind-action@v1.10.0
+        if: steps.list-changed.outputs.changed == 'true'
+
+      - name: Run chart-testing (install)
+        run: ct install --config ct.yaml

.github/workflows/chart-release.yaml

@@ -0,0 +1,40 @@
+name: Release Charts
+
+on:
+  push:
+    paths:
+      - charts/**
+      - '!**.md'
+    branches:
+      - main
+
+jobs:
+  release:
+    runs-on: ubuntu-latest
+
+    permissions:
+      contents: write
+
+    steps:
+      - name: Checkout
+        uses: actions/checkout@8ade135a41bc03ea155e62e844d188df1ea18608 # v4.1.0
+        with:
+          fetch-depth: 0
+
+      - name: Configure Git
+        run: |
+          git config user.name "$GITHUB_ACTOR"
+          git config user.email "$GITHUB_ACTOR@users.noreply.github.com"
+
+      - name: Install Helm
+        uses: azure/setup-helm@5119fcb9089d432beecbf79bb2c7915207344b78 # v3.5
+        with:
+          version: v3.13.0
+
+      - name: Run chart-releaser
+        uses: helm/chart-releaser-action@a917fd15b20e8b64b94d9158ad54cd6345335584 # v1.6.0
+        with:
+          charts_dir: charts
+          config: cr.yaml
+        env:
+          CR_TOKEN: "${{ secrets.GITHUB_TOKEN }}"

.github/workflows/generate-chart-readme.yaml

@@ -0,0 +1,48 @@
+name: '[CI/CD] Update README metadata'
+
+on:
+  pull_request_target:
+    branches:
+      - main
+    paths:
+      - 'charts/*/values.yaml'
+# Remove all permissions by default
+permissions: {}
+jobs:
+  update-readme-metadata:
+    runs-on: ubuntu-latest
+    permissions:
+      contents: write
+    steps:
+      - name: Install readme-generator-for-helm
+        run: npm install -g @bitnami/readme-generator-for-helm
+      - name: Checkout
+        uses: actions/checkout@8ade135a41bc03ea155e62e844d188df1ea18608
+        with:
+          path: charts
+          ref: ${{github.event.pull_request.head.ref}}
+          repository: ${{github.event.pull_request.head.repo.full_name}}
+          token: ${{ secrets.GITHUB_TOKEN }}
+      - name: Execute readme-generator-for-helm
+        env:
+          DIFF_URL: "${{github.event.pull_request.diff_url}}"
+          TEMP_FILE: "${{runner.temp}}/pr-${{github.event.number}}.diff"
+        run: |
+          # This request doesn't consume API calls.
+          curl -Lkso $TEMP_FILE $DIFF_URL
+          files_changed="$(sed -nr 's/[\-\+]{3} [ab]\/(.*)/\1/p' $TEMP_FILE | sort | uniq)"
+          # Adding || true to avoid "Process exited with code 1" errors
+          charts_dirs_changed="$(echo "$files_changed" | xargs dirname | grep -o "pgcat/[^/]*" | sort | uniq || true)"
+          for chart in ${charts_dirs_changed}; do
+            echo "Updating README.md for ${chart}"
+            readme-generator --values "charts/${chart}/values.yaml" --readme "charts/${chart}/README.md" --schema "/tmp/schema.json"
+          done
+      - name: Push changes
+        run: |
+          # Push all the changes
+          cd charts
+          if git status -s | grep pgcat; then
+            git config user.name "$GITHUB_ACTOR"
+            git config user.email "$GITHUB_ACTOR@users.noreply.github.com"
+            git add . && git commit -am "Update README.md with readme-generator-for-helm" --signoff && git push
+          fi

.github/workflows/publish-deb-package.yml

@@ -1,6 +1,9 @@
 name: pgcat package (deb)
 
 on:
+  push:
+    tags:
+      - v*
   workflow_dispatch:
     inputs:
       packageVersion:
@@ -16,6 +19,14 @@ jobs:
     runs-on: ${{ matrix.os }}
     steps:
     - uses: actions/checkout@v3
+    - name: Set package version
+      if: github.event_name == 'push'  # For push event
+      run: |
+        TAG=${{ github.ref_name }}
+        echo "packageVersion=${TAG#v}" >> "$GITHUB_ENV"
+    - name: Set package version (manual dispatch)
+      if: github.event_name == 'workflow_dispatch'  # For manual dispatch
+      run: echo "packageVersion=${{ github.event.inputs.packageVersion }}" >> "$GITHUB_ENV"
     - uses: actions-rs/toolchain@v1
       with:
         toolchain: stable
@@ -39,10 +50,10 @@ jobs:
           export ARCH=arm64
         fi
 
-        bash utilities/deb.sh ${{ inputs.packageVersion }}
+        bash utilities/deb.sh ${{ env.packageVersion }}
 
         deb-s3 upload \
           --lock \
           --bucket apt.postgresml.org \
-          pgcat-${{ inputs.packageVersion }}-ubuntu22.04-${ARCH}.deb \
+          pgcat-${{ env.packageVersion }}-ubuntu22.04-${ARCH}.deb \
           --codename $(lsb_release -cs)

.gitignore

@@ -10,4 +10,6 @@ lcov.info
 dev/.bash_history
 dev/cache
 !dev/cache/.keepme
-.venv
+.venv
+**/__pycache__
+.bundle

CONFIG.md

@@ -36,10 +36,11 @@ Port at which prometheus exporter listens on.
 ### connect_timeout
 ```
 path: general.connect_timeout
-default: 5000 # milliseconds
+default: 1000 # milliseconds
 ```
 
-How long to wait before aborting a server connection (ms).
+How long the client waits to obtain a server connection before aborting (ms).
+This is similar to PgBouncer's `query_wait_timeout`.
 
 ### idle_timeout
 ```
@@ -297,6 +298,19 @@ Load balancing mode
 `random` selects the server at random
 `loc` selects the server with the least outstanding busy connections
 
+### checkout_failure_limit
+```
+path: pools.<pool_name>.checkout_failure_limit
+default: 0 (disabled)
+```
+
+`Maximum number of checkout failures a client is allowed before it
+gets disconnected. This is needed to prevent persistent client/server
+imbalance in high availability setups where multiple PgCat instances are placed
+behind a single load balancer. If for any reason a client lands on a PgCat instance that has 
+a large number of connected clients, it might get stuck in perpetual checkout failure loop especially
+in session mode
+`
 ### default_role
 ```
 path: pools.<pool_name>.default_role
@@ -308,6 +322,45 @@ If the client doesn't specify, PgCat routes traffic to this role by default.
 `replica` round-robin between replicas only without touching the primary,
 `primary` all queries go to the primary unless otherwise specified.
 
+### db_activity_based_routing
+```
+path: pools.<pool_name>.db_activity_based_routing
+default: false
+```
+
+If enabled, PgCat will route queries to the primary if the queried table was recently written to.
+Only relevant when `query_parser_enabled` *and* `query_parser_read_write_splitting` is enabled.
+
+##### Considerations:
+- *This feature is experimental and may not work as expected.*
+- This feature only works when the same PgCat instance is used for both reads and writes to the database.
+- This feature is not relevant when the primary is not part of the pool of databases used for load balancing of read queries.
+- If more than one PgCat instance is used for HA purposes, this feature will not work as expected. A way to still make it work is by using sticky sessions.
+
+### db_activity_based_ms_init_delay
+```
+path: pools.<pool_name>.db_activity_based_ms_init_delay
+default: 100
+```
+
+The delay in milliseconds before the first activity-based routing check is performed.
+
+### db_activity_ttl
+```
+path: pools.<pool_name>.db_activity_ttl
+default: 900
+```
+
+The time in seconds after which a DB is considered inactive when no queries/updates are performed to it.
+
+### table_mutation_cache_ms_ttl
+```
+path: pools.<pool_name>.table_mutation_cache_ms_ttl
+default: 50
+```
+
+The time in milliseconds after a write to a table that all queries to that table will be routed to the primary.
+
 ### prepared_statements_cache_size
 ```
 path: general.prepared_statements_cache_size
@@ -462,10 +515,18 @@ path: pools.<pool_name>.users.<user_index>.pool_size
 default: 9
 ```
 
-Maximum number of server connections that can be established for this user
+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.
 
+### min_pool_size
+```
+path: pools.<pool_name>.users.<user_index>.min_pool_size
+default: 0
+```
+
+Minimum number of idle server connections to retain for this pool.
+
 ### statement_timeout
 ```
 path: pools.<pool_name>.users.<user_index>.statement_timeout
@@ -475,6 +536,16 @@ default: 0
 Maximum query duration. Dangerous, but protects against DBs that died in a non-obvious way.
 0 means it is disabled.
 
+### connect_timeout
+```
+path: pools.<pool_name>.users.<user_index>.connect_timeout
+default: <UNSET> # milliseconds
+```
+
+How long the client waits to obtain a server connection before aborting (ms).
+This is similar to PgBouncer's `query_wait_timeout`.
+If unset, uses the `connect_timeout` defined globally.
+
 ## `pools.<pool_name>.shards.<shard_index>` Section
 
 ### servers
@@ -502,4 +573,3 @@ default: "shard0"
 ```
 
 Database name (e.g. "postgres")
-

CONTRIBUTING.md

@@ -6,6 +6,32 @@ Thank you for contributing! Just a few tips here:
 2. Run the test suite (e.g. `pgbench`) to make sure everything still works. The tests are in `.circleci/run_tests.sh`.
 3. Performance is important, make sure there are no regressions in your branch vs. `main`.
 
+## How to run the integration tests locally and iterate on them
+We have integration tests written in Ruby, Python, Go and Rust.
+Below are the steps to run them in a developer-friendly way that allows iterating and quick turnaround.
+Hear me out, this should be easy, it will involve opening a shell into a container with all the necessary dependancies available for you and you can modify the test code and immediately rerun your test in the interactive shell.
+
+
+Quite simply, make sure you have docker installed and then run
+`./start_test_env.sh`
+
+That is it!
+
+Within this test environment you can modify the file in your favorite IDE and rerun the tests without having to bootstrap the entire environment again.
+
+Once the environment is ready, you can run the tests by running
+Ruby:   `cd /app/tests/ruby && bundle exec ruby <test_name>.rb --format documentation`
+Python: `cd /app/ && pytest`
+Rust:   `cd /app/tests/rust && cargo run`
+Go:     `cd /app/tests/go && /usr/local/go/bin/go test`
+
+You can also rebuild PgCat directly within the environment and the tests will run against the newly built binary
+To rebuild PgCat, just run `cargo build` within the container under `/app`
+
+![Animated gif showing how to run tests](https://github.com/user-attachments/assets/2258fde3-2aed-4efb-bdc5-e4f12dcd4d33)
+
+
+
 Happy hacking!
 
 ## TODOs

Cargo.lock

@@ -132,7 +132,7 @@ checksum = "16e62a023e7c117e27523144c5d2459f4397fcc3cab0085af8e2224f643a0193"
 dependencies = [
  "proc-macro2",
  "quote",
- "syn 2.0.26",
+ "syn 2.0.87",
 ]
 
 [[package]]
@@ -143,9 +143,15 @@ checksum = "a564d521dd56509c4c47480d00b80ee55f7e385ae48db5744c67ad50c92d2ebf"
 dependencies = [
  "proc-macro2",
  "quote",
- "syn 2.0.26",
+ "syn 2.0.87",
 ]
 
+[[package]]
+name = "atomic-waker"
+version = "1.1.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "1505bd5d3d116872e7271a6d4e16d81d0c8570876c8de68093a09ac269d8aac0"
+
 [[package]]
 name = "atomic_enum"
 version = "0.2.0"
@@ -186,12 +192,11 @@ checksum = "604178f6c5c21f02dc555784810edfb88d34ac2c73b2eae109655649ee73ce3d"
 
 [[package]]
 name = "bb8"
-version = "0.8.1"
+version = "0.8.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "98b4b0f25f18bcdc3ac72bdb486ed0acf7e185221fd4dc985bc15db5800b0ba2"
+checksum = "d89aabfae550a5c44b43ab941844ffcd2e993cb6900b342debf59e9ea74acdb8"
 dependencies = [
  "async-trait",
- "futures-channel",
  "futures-util",
  "parking_lot",
  "tokio",
@@ -224,6 +229,12 @@ version = "3.13.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "a3e2c3daef883ecc1b5d58c15adae93470a91d425f3532ba1695849656af3fc1"
 
+[[package]]
+name = "bytecount"
+version = "0.6.8"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5ce89b21cab1437276d2650d57e971f9d548a2d9037cc231abdc0562b97498ce"
+
 [[package]]
 name = "byteorder"
 version = "1.4.3"
@@ -236,6 +247,37 @@ version = "1.4.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "89b2fd2a0dcf38d7971e2194b6b6eebab45ae01067456a7fd93d5547a61b70be"
 
+[[package]]
+name = "camino"
+version = "1.1.9"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "8b96ec4966b5813e2c0507c1f86115c8c5abaadc3980879c3424042a02fd1ad3"
+dependencies = [
+ "serde",
+]
+
+[[package]]
+name = "cargo-platform"
+version = "0.1.8"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "24b1f0365a6c6bb4020cd05806fd0d33c44d38046b8bd7f0e40814b9763cabfc"
+dependencies = [
+ "serde",
+]
+
+[[package]]
+name = "cargo_metadata"
+version = "0.14.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "4acbb09d9ee8e23699b9634375c72795d095bf268439da88562cf9b501f181fa"
+dependencies = [
+ "camino",
+ "cargo-platform",
+ "semver",
+ "serde",
+ "serde_json",
+]
+
 [[package]]
 name = "cc"
 version = "1.0.79"
@@ -295,7 +337,7 @@ dependencies = [
  "heck",
  "proc-macro2",
  "quote",
- "syn 2.0.26",
+ "syn 2.0.87",
 ]
 
 [[package]]
@@ -325,6 +367,21 @@ dependencies = [
  "libc",
 ]
 
+[[package]]
+name = "crossbeam-channel"
+version = "0.5.13"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "33480d6946193aa8033910124896ca395333cae7e2d1113d1fef6c3272217df2"
+dependencies = [
+ "crossbeam-utils",
+]
+
+[[package]]
+name = "crossbeam-utils"
+version = "0.8.20"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "22ec99545bb0ed0ea7bb9b8e1e9122ea386ff8a48c0922e43f36d45ab09e0e80"
+
 [[package]]
 name = "crypto-common"
 version = "0.1.6"
@@ -335,6 +392,19 @@ dependencies = [
  "typenum",
 ]
 
+[[package]]
+name = "dashmap"
+version = "5.5.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "978747c1d849a7d2ee5e8adc0159961c48fb7e5db2f06af6723b80123bb53856"
+dependencies = [
+ "cfg-if",
+ "hashbrown",
+ "lock_api",
+ "once_cell",
+ "parking_lot_core",
+]
+
 [[package]]
 name = "data-encoding"
 version = "2.4.0"
@@ -397,6 +467,15 @@ dependencies = [
  "libc",
 ]
 
+[[package]]
+name = "error-chain"
+version = "0.12.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "2d2f06b9cac1506ece98fe3231e3cc9c4410ec3d5b1f24ae1c8946f0742cdefc"
+dependencies = [
+ "version_check",
+]
+
 [[package]]
 name = "exitcode"
 version = "1.1.2"
@@ -409,6 +488,12 @@ version = "0.2.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "4443176a9f2c162692bd3d352d745ef9413eec5782a80d8fd6f8a1ac692a07f7"
 
+[[package]]
+name = "fastrand"
+version = "2.1.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e8c02a5121d4ea3eb16a80748c74f5549a5665e4c21333c6098f283870fbdea6"
+
 [[package]]
 name = "fnv"
 version = "1.0.7"
@@ -480,7 +565,7 @@ checksum = "89ca545a94061b6365f2c7355b4b32bd20df3ff95f02da9329b34ccc3bd6ee72"
 dependencies = [
  "proc-macro2",
  "quote",
- "syn 2.0.26",
+ "syn 2.0.87",
 ]
 
 [[package]]
@@ -541,30 +626,30 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "b6c80984affa11d98d1b88b66ac8853f143217b399d3c74116778ff8fdb4ed2e"
 
 [[package]]
-name = "h2"
-version = "0.3.20"
+name = "glob"
+version = "0.3.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "97ec8491ebaf99c8eaa73058b045fe58073cd6be7f596ac993ced0b0a0c01049"
+checksum = "d2fabcfbdc87f4758337ca535fb41a6d701b65693ce38287d856d1674551ec9b"
+
+[[package]]
+name = "h2"
+version = "0.4.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "524e8ac6999421f49a846c2d4411f337e53497d8ec55d67753beffa43c5d9205"
 dependencies = [
+ "atomic-waker",
  "bytes",
  "fnv",
  "futures-core",
  "futures-sink",
- "futures-util",
  "http",
- "indexmap 1.9.3",
+ "indexmap",
  "slab",
  "tokio",
  "tokio-util",
  "tracing",
 ]
 
-[[package]]
-name = "hashbrown"
-version = "0.12.3"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "8a9ee70c43aaf417c914396645a0fa852624801b24ebb7ae78fe8272889ac888"
-
 [[package]]
 name = "hashbrown"
 version = "0.14.0"
@@ -609,9 +694,9 @@ dependencies = [
 
 [[package]]
 name = "http"
-version = "0.2.9"
+version = "1.1.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "bd6effc99afb63425aff9b05836f029929e345a6148a14b7ecd5ab67af944482"
+checksum = "21b9ddb458710bc376481b842f5da65cdf31522de232c1ca8146abce2a358258"
 dependencies = [
  "bytes",
  "fnv",
@@ -620,12 +705,24 @@ dependencies = [
 
 [[package]]
 name = "http-body"
-version = "0.4.5"
+version = "1.0.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d5f38f16d184e36f2408a55281cd658ecbd3ca05cce6d6510a176eca393e26d1"
+checksum = "1efedce1fb8e6913f23e0c92de8e62cd5b772a67e7b3946df930a62566c93184"
 dependencies = [
  "bytes",
  "http",
+]
+
+[[package]]
+name = "http-body-util"
+version = "0.1.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "793429d76616a256bcb62c2a2ec2bed781c8307e797e2598c50010f2bee2544f"
+dependencies = [
+ "bytes",
+ "futures-util",
+ "http",
+ "http-body",
  "pin-project-lite",
 ]
 
@@ -643,13 +740,12 @@ checksum = "c4a1e36c821dbe04574f602848a19f742f4fb3c98d40449f11bcad18d6b17421"
 
 [[package]]
 name = "hyper"
-version = "0.14.27"
+version = "1.4.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ffb1cfd654a8219eaef89881fdb3bb3b1cdc5fa75ded05d6933b2b382e395468"
+checksum = "50dfd22e0e76d0f662d429a5f80fcaf3855009297eab6a0a9f8543834744ba05"
 dependencies = [
  "bytes",
  "futures-channel",
- "futures-core",
  "futures-util",
  "h2",
  "http",
@@ -658,13 +754,26 @@ dependencies = [
  "httpdate",
  "itoa",
  "pin-project-lite",
- "socket2 0.4.9",
+ "smallvec",
  "tokio",
- "tower-service",
- "tracing",
  "want",
 ]
 
+[[package]]
+name = "hyper-util"
+version = "0.1.7"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "cde7055719c54e36e95e8719f95883f22072a48ede39db7fc17a4e1d5281e9b9"
+dependencies = [
+ "bytes",
+ "futures-util",
+ "http",
+ "http-body",
+ "hyper",
+ "pin-project-lite",
+ "tokio",
+]
+
 [[package]]
 name = "iana-time-zone"
 version = "0.1.57"
@@ -709,16 +818,6 @@ dependencies = [
  "unicode-normalization",
 ]
 
-[[package]]
-name = "indexmap"
-version = "1.9.3"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "bd070e393353796e801d209ad339e89596eb4c8d430d18ede6a1cced8fafbd99"
-dependencies = [
- "autocfg",
- "hashbrown 0.12.3",
-]
-
 [[package]]
 name = "indexmap"
 version = "2.0.0"
@@ -726,7 +825,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "d5477fe2230a79769d8dc68e0eabf5437907c0457a5614a9e8dddb67f65eb65d"
 dependencies = [
  "equivalent",
- "hashbrown 0.14.0",
+ "hashbrown",
 ]
 
 [[package]]
@@ -848,7 +947,7 @@ version = "0.12.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "1efa59af2ddfad1854ae27d75009d538d0998b4b2fd47083e743ac1a10e46c60"
 dependencies = [
- "hashbrown 0.14.0",
+ "hashbrown",
 ]
 
 [[package]]
@@ -905,6 +1004,21 @@ dependencies = [
  "autocfg",
 ]
 
+[[package]]
+name = "mini-moka"
+version = "0.10.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c325dfab65f261f386debee8b0969da215b3fa0037e74c8a1234db7ba986d803"
+dependencies = [
+ "crossbeam-channel",
+ "crossbeam-utils",
+ "dashmap",
+ "skeptic",
+ "smallvec",
+ "tagptr",
+ "triomphe",
+]
+
 [[package]]
 name = "miniz_oxide"
 version = "0.7.1"
@@ -979,9 +1093,9 @@ dependencies = [
 
 [[package]]
 name = "once_cell"
-version = "1.18.0"
+version = "1.20.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "dd8b5dd2ae5ed71462c540258bedcb51965123ad7e7ccf4b9a8cafaa4a63576d"
+checksum = "1261fe7e33c73b354eab43b1273a57c8f967d0391e80353e51f764ac02cf6775"
 
 [[package]]
 name = "overload"
@@ -1020,7 +1134,7 @@ checksum = "9b2a4787296e9989611394c33f193f676704af1686e70b8f8033ab5ba9a35a94"
 
 [[package]]
 name = "pgcat"
-version = "1.1.2-dev1"
+version = "1.3.0"
 dependencies = [
  "arc-swap",
  "async-trait",
@@ -1034,12 +1148,15 @@ dependencies = [
  "fallible-iterator",
  "futures",
  "hmac",
+ "http-body-util",
  "hyper",
+ "hyper-util",
  "itertools",
  "jemallocator",
  "log",
  "lru",
  "md-5",
+ "mini-moka",
  "nix",
  "num_cpus",
  "once_cell",
@@ -1054,6 +1171,7 @@ dependencies = [
  "serde",
  "serde_derive",
  "serde_json",
+ "serial_test",
  "sha-1",
  "sha2",
  "socket2 0.4.9",
@@ -1099,7 +1217,7 @@ dependencies = [
  "phf_shared",
  "proc-macro2",
  "quote",
- "syn 2.0.26",
+ "syn 2.0.87",
 ]
 
 [[package]]
@@ -1128,7 +1246,7 @@ checksum = "ec2e072ecce94ec471b13398d5402c188e76ac03cf74dd1a975161b23a3f6d9c"
 dependencies = [
  "proc-macro2",
  "quote",
- "syn 2.0.26",
+ "syn 2.0.87",
 ]
 
 [[package]]
@@ -1169,13 +1287,24 @@ checksum = "5b40af805b3121feab8a3c29f04d8ad262fa8e0561883e7653e024ae4479e6de"
 
 [[package]]
 name = "proc-macro2"
-version = "1.0.66"
+version = "1.0.89"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "18fb31db3f9bddb2ea821cde30a9f70117e3f119938b5ee630b7403aa6e2ead9"
+checksum = "f139b0662de085916d1fb67d2b4169d1addddda1919e696f3252b740b629986e"
 dependencies = [
  "unicode-ident",
 ]
 
+[[package]]
+name = "pulldown-cmark"
+version = "0.9.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "57206b407293d2bcd3af849ce869d52068623f19e1b5ff8e8778e3309439682b"
+dependencies = [
+ "bitflags 2.3.3",
+ "memchr",
+ "unicase",
+]
+
 [[package]]
 name = "quick-error"
 version = "1.2.3"
@@ -1184,9 +1313,9 @@ checksum = "a1d01941d82fa2ab50be1e79e6714289dd7cde78eba4c074bc5a4374f650dfe0"
 
 [[package]]
 name = "quote"
-version = "1.0.31"
+version = "1.0.37"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "5fe8a65d69dd0808184ebb5f836ab526bb259db23c657efa38711b1072ee47f0"
+checksum = "b5b9d34b8991d19d98081b46eacdd8eb58c6f2b201139f7c5f643cc155a633af"
 dependencies = [
  "proc-macro2",
 ]
@@ -1365,6 +1494,24 @@ version = "1.0.15"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "1ad4cc8da4ef723ed60bced201181d83791ad433213d8c24efffda1eec85d741"
 
+[[package]]
+name = "same-file"
+version = "1.0.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "93fc1dc3aaa9bfed95e02e6eadabb4baf7e3078b0bd1b4d7b6b0b68378900502"
+dependencies = [
+ "winapi-util",
+]
+
+[[package]]
+name = "scc"
+version = "2.2.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d8d25269dd3a12467afe2e510f69fb0b46b698e5afb296b59f2145259deaf8e8"
+dependencies = [
+ "sdd",
+]
+
 [[package]]
 name = "scopeguard"
 version = "1.2.0"
@@ -1382,23 +1529,38 @@ dependencies = [
 ]
 
 [[package]]
-name = "serde"
-version = "1.0.171"
+name = "sdd"
+version = "3.0.4"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "30e27d1e4fd7659406c492fd6cfaf2066ba8773de45ca75e855590f856dc34a9"
+checksum = "49c1eeaf4b6a87c7479688c6d52b9f1153cedd3c489300564f932b065c6eab95"
+
+[[package]]
+name = "semver"
+version = "1.0.23"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "61697e0a1c7e512e84a621326239844a24d8207b4669b41bc18b32ea5cbf988b"
+dependencies = [
+ "serde",
+]
+
+[[package]]
+name = "serde"
+version = "1.0.214"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f55c3193aca71c12ad7890f1785d2b73e1b9f63a0bbc353c08ef26fe03fc56b5"
 dependencies = [
  "serde_derive",
 ]
 
 [[package]]
 name = "serde_derive"
-version = "1.0.171"
+version = "1.0.214"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "389894603bd18c46fa56231694f8d827779c0951a667087194cf9de94ed24682"
+checksum = "de523f781f095e28fa605cdce0f8307e451cc0fd14e2eb4cd2e98a355b147766"
 dependencies = [
  "proc-macro2",
  "quote",
- "syn 2.0.26",
+ "syn 2.0.87",
 ]
 
 [[package]]
@@ -1421,6 +1583,31 @@ dependencies = [
  "serde",
 ]
 
+[[package]]
+name = "serial_test"
+version = "3.1.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "4b4b487fe2acf240a021cf57c6b2b4903b1e78ca0ecd862a71b71d2a51fed77d"
+dependencies = [
+ "futures",
+ "log",
+ "once_cell",
+ "parking_lot",
+ "scc",
+ "serial_test_derive",
+]
+
+[[package]]
+name = "serial_test_derive"
+version = "3.1.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "82fe9db325bcef1fbcde82e078a5cc4efdf787e96b3b9cf45b50b529f2083d67"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn 2.0.87",
+]
+
 [[package]]
 name = "sha-1"
 version = "0.10.1"
@@ -1467,6 +1654,21 @@ version = "0.3.10"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "7bd3e3206899af3f8b12af284fafc038cc1dc2b41d1b89dd17297221c5d225de"
 
+[[package]]
+name = "skeptic"
+version = "0.13.7"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "16d23b015676c90a0f01c197bfdc786c20342c73a0afdda9025adb0bc42940a8"
+dependencies = [
+ "bytecount",
+ "cargo_metadata",
+ "error-chain",
+ "glob",
+ "pulldown-cmark",
+ "tempfile",
+ "walkdir",
+]
+
 [[package]]
 name = "slab"
 version = "0.4.8"
@@ -1478,9 +1680,9 @@ dependencies = [
 
 [[package]]
 name = "smallvec"
-version = "1.11.0"
+version = "1.13.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "62bb4feee49fdd9f707ef802e22365a35de4b7b299de4763d44bfea899442ff9"
+checksum = "3c5e1a9a646d36c3599cd173a41282daf47c44583ad367b8e6837255952e5c67"
 
 [[package]]
 name = "socket2"
@@ -1510,9 +1712,9 @@ checksum = "6e63cff320ae2c57904679ba7cb63280a3dc4613885beafb148ee7bf9aa9042d"
 
 [[package]]
 name = "sqlparser"
-version = "0.34.0"
+version = "0.52.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "37d3706eefb17039056234df6b566b0014f303f867f2656108334a55b8096f59"
+checksum = "9a875d8cd437cc8a97e9aeaeea352ec9a19aea99c23e9effb17757291de80b08"
 dependencies = [
  "log",
  "sqlparser_derive",
@@ -1520,13 +1722,13 @@ dependencies = [
 
 [[package]]
 name = "sqlparser_derive"
-version = "0.1.1"
+version = "0.2.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "55fe75cb4a364c7f7ae06c7dbbc8d84bddd85d6cdf9975963c3935bc1991761e"
+checksum = "01b2e185515564f15375f593fb966b5718bc624ba77fe49fa4616ad619690554"
 dependencies = [
  "proc-macro2",
  "quote",
- "syn 1.0.109",
+ "syn 2.0.87",
 ]
 
 [[package]]
@@ -1570,15 +1772,34 @@ dependencies = [
 
 [[package]]
 name = "syn"
-version = "2.0.26"
+version = "2.0.87"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "45c3457aacde3c65315de5031ec191ce46604304d2446e803d71ade03308d970"
+checksum = "25aa4ce346d03a6dcd68dd8b4010bcb74e54e62c90c573f394c46eae99aba32d"
 dependencies = [
  "proc-macro2",
  "quote",
  "unicode-ident",
 ]
 
+[[package]]
+name = "tagptr"
+version = "0.2.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "7b2093cf4c8eb1e67749a6762251bc9cd836b6fc171623bd0a9d324d37af2417"
+
+[[package]]
+name = "tempfile"
+version = "3.8.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "cb94d2f3cc536af71caac6b6fcebf65860b347e7ce0cc9ebe8f70d3e521054ef"
+dependencies = [
+ "cfg-if",
+ "fastrand",
+ "redox_syscall",
+ "rustix",
+ "windows-sys",
+]
+
 [[package]]
 name = "thiserror"
 version = "1.0.43"
@@ -1596,7 +1817,7 @@ checksum = "463fe12d7993d3b327787537ce8dd4dfa058de32fc2b195ef3cde03dc4771e8f"
 dependencies = [
  "proc-macro2",
  "quote",
- "syn 2.0.26",
+ "syn 2.0.87",
 ]
 
 [[package]]
@@ -1663,7 +1884,7 @@ checksum = "630bdcf245f78637c13ec01ffae6187cca34625e8c63150d424b59e55af2675e"
 dependencies = [
  "proc-macro2",
  "quote",
- "syn 2.0.26",
+ "syn 2.0.87",
 ]
 
 [[package]]
@@ -1741,19 +1962,13 @@ version = "0.19.14"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "f8123f27e969974a3dfba720fdb560be359f57b44302d280ba72e76a74480e8a"
 dependencies = [
- "indexmap 2.0.0",
+ "indexmap",
  "serde",
  "serde_spanned",
  "toml_datetime",
  "winnow",
 ]
 
-[[package]]
-name = "tower-service"
-version = "0.3.2"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b6bc1c9ce2b5135ac7f93c72918fc37feb872bdc6a5533a8b85eb4b86bfdae52"
-
 [[package]]
 name = "tracing"
 version = "0.1.37"
@@ -1774,7 +1989,7 @@ checksum = "5f4f31f56159e98206da9efd823404b79b6ef3143b4a7ab76e67b1751b25a4ab"
 dependencies = [
  "proc-macro2",
  "quote",
- "syn 2.0.26",
+ "syn 2.0.87",
 ]
 
 [[package]]
@@ -1829,6 +2044,12 @@ dependencies = [
  "tracing-serde",
 ]
 
+[[package]]
+name = "triomphe"
+version = "0.1.11"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "859eb650cfee7434994602c3a68b25d77ad9e68c8a6cd491616ef86661382eb3"
+
 [[package]]
 name = "trust-dns-proto"
 version = "0.22.0"
@@ -1886,6 +2107,12 @@ version = "1.16.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "497961ef93d974e23eb6f433eb5fe1b7930b659f06d12dec6fc44a8f554c0bba"
 
+[[package]]
+name = "unicase"
+version = "2.8.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "7e51b68083f157f853b6379db119d1c1be0e6e4dec98101079dec41f6f5cf6df"
+
 [[package]]
 name = "unicode-bidi"
 version = "0.3.13"
@@ -1942,6 +2169,16 @@ version = "0.9.4"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "49874b5167b65d7193b8aba1567f5c7d93d001cafc34600cee003eda787e483f"
 
+[[package]]
+name = "walkdir"
+version = "2.5.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "29790946404f91d9c5d06f9874efddea1dc06c5efe94541a7d6863108e3a5e4b"
+dependencies = [
+ "same-file",
+ "winapi-util",
+]
+
 [[package]]
 name = "want"
 version = "0.3.1"
@@ -1984,7 +2221,7 @@ dependencies = [
  "once_cell",
  "proc-macro2",
  "quote",
- "syn 2.0.26",
+ "syn 2.0.87",
  "wasm-bindgen-shared",
 ]
 
@@ -2006,7 +2243,7 @@ checksum = "54681b18a46765f095758388f2d0cf16eb8d4169b639ab575a8f5693af210c7b"
 dependencies = [
  "proc-macro2",
  "quote",
- "syn 2.0.26",
+ "syn 2.0.87",
  "wasm-bindgen-backend",
  "wasm-bindgen-shared",
 ]
@@ -2058,6 +2295,15 @@ version = "0.4.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "ac3b87c63620426dd9b991e5ce0329eff545bccbbb34f3be09ff6fb6ab51b7b6"
 
+[[package]]
+name = "winapi-util"
+version = "0.1.9"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "cf221c93e13a30d793f7645a0e7762c55d169dbb0a49671918a2319d289b10bb"
+dependencies = [
+ "windows-sys",
+]
+
 [[package]]
 name = "winapi-x86_64-pc-windows-gnu"
 version = "0.4.0"

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "pgcat"
-version = "1.1.2-dev1"
+version = "1.3.0"
 edition = "2021"
 
 # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
@@ -8,7 +8,7 @@ edition = "2021"
 tokio = { version = "1", features = ["full"] }
 bytes = "1"
 md-5 = "0.10"
-bb8 = "0.8.1"
+bb8 = "=0.8.6"
 async-trait = "0.1"
 rand = "0.8"
 chrono = "0.4"
@@ -19,7 +19,7 @@ serde_derive = "1"
 regex = "1"
 num_cpus = "1"
 once_cell = "1"
-sqlparser = {version = "0.34", features = ["visitor"] }
+sqlparser = { version = "0.52", features = ["visitor"] }
 log = "0.4"
 arc-swap = "1"
 parking_lot = "0.12.1"
@@ -29,7 +29,9 @@ base64 = "0.21"
 stringprep = "0.1"
 tokio-rustls = "0.24"
 rustls-pemfile = "1"
-hyper = { version = "0.14", features = ["full"] }
+http-body-util = "0.1.2"
+hyper = { version = "1.4.1", features = ["full"] }
+hyper-util = { version = "0.1.7", features = ["tokio"] }
 phf = { version = "0.11.1", features = ["macros"] }
 exitcode = "1.1.2"
 futures = "0.3"
@@ -47,9 +49,16 @@ serde_json = "1"
 itertools = "0.10"
 clap = { version = "4.3.1", features = ["derive", "env"] }
 tracing = "0.1.37"
-tracing-subscriber = { version = "0.3.17", features = ["json", "env-filter", "std"]}
+tracing-subscriber = { version = "0.3.17", features = [
+    "json",
+    "env-filter",
+    "std",
+] }
 lru = "0.12.0"
+mini-moka = "0.10.3"
 
 [target.'cfg(not(target_env = "msvc"))'.dependencies]
 jemallocator = "0.5.0"
 
+[dev-dependencies]
+serial_test = "*"

Dockerfile

@@ -1,4 +1,4 @@
-FROM rust:1-slim-bookworm AS builder
+FROM rust:1.81.0-slim-bookworm AS builder
 
 RUN apt-get update && \
     apt-get install -y build-essential
@@ -19,3 +19,4 @@ COPY --from=builder /app/pgcat.toml /etc/pgcat/pgcat.toml
 WORKDIR /etc/pgcat
 ENV RUST_LOG=info
 CMD ["pgcat"]
+STOPSIGNAL SIGINT

Dockerfile.ci

@@ -1,4 +1,4 @@
-FROM cimg/rust:1.67.1
+FROM cimg/rust:1.81.0
 COPY --from=sclevine/yj /bin/yj /bin/yj
 RUN /bin/yj -h
 RUN sudo apt-get update && \
@@ -9,6 +9,9 @@ RUN sudo apt-get update && \
 	sudo apt-get upgrade curl && \
 	cargo install cargo-binutils rustfilt && \
 	rustup component add llvm-tools-preview && \
- 	pip3 install psycopg2 && sudo gem install bundler && \
+	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_$(dpkg --print-architecture).deb && \
 	sudo dpkg -i /tmp/toxiproxy-2.4.0.deb
+RUN wget -O /tmp/go1.21.3.linux-$(dpkg --print-architecture).tar.gz https://go.dev/dl/go1.21.3.linux-$(dpkg --print-architecture).tar.gz && \
+    sudo tar -C /usr/local -xzf /tmp/go1.21.3.linux-$(dpkg --print-architecture).tar.gz && \
+    rm /tmp/go1.21.3.linux-$(dpkg --print-architecture).tar.gz

README.md

@@ -40,7 +40,7 @@ PgCat is stable and used in production to serve hundreds of thousands of queries
       </a>
     </td>
     <td>
-      <a href="https://postgresml.org/blog/scaling-postgresml-to-one-million-requests-per-second">
+      <a href="https://postgresml.org/blog/scaling-postgresml-to-1-million-requests-per-second">
         <img src="./images/postgresml.webp" height="70" width="auto">
       </a>
     </td>
@@ -57,7 +57,7 @@ PgCat is stable and used in production to serve hundreds of thousands of queries
       </a>
     </td>
     <td>
-      <a href="https://postgresml.org/blog/scaling-postgresml-to-one-million-requests-per-second">
+      <a href="https://postgresml.org/blog/scaling-postgresml-to-1-million-requests-per-second">
         PostgresML
       </a>
     </td>
@@ -231,7 +231,7 @@ User.find_by_email("test@example.com")
 ```sql
 -- Grab a bunch of users from shard 1
 SET SHARD TO '1';
-SELECT * FROM users LIMT 10;
+SELECT * FROM users LIMIT 10;
 
 -- Find by id
 SET SHARDING KEY TO '1234';
@@ -268,6 +268,8 @@ psql -h 127.0.0.1 -p 6432 -d pgbouncer -c 'SHOW DATABASES'
 
 Additionally, Prometheus statistics are available at `/metrics` via HTTP.
 
+We also have a [basic Grafana dashboard](https://github.com/postgresml/pgcat/blob/main/grafana_dashboard.json) based on Prometheus metrics that you can import into Grafana and build on it or use it for monitoring.
+
 ### 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. All settings except the `host` and `port` can be reloaded without restarting the pooler, including sharding and replicas configurations.

charts/pgcat/.helmignore

@@ -0,0 +1,23 @@
+# Patterns to ignore when building packages.
+# This supports shell glob matching, relative path matching, and
+# negation (prefixed with !). Only one pattern per line.
+.DS_Store
+# Common VCS dirs
+.git/
+.gitignore
+.bzr/
+.bzrignore
+.hg/
+.hgignore
+.svn/
+# Common backup files
+*.swp
+*.bak
+*.tmp
+*.orig
+*~
+# Various IDEs
+.project
+.idea/
+*.tmproj
+.vscode/

charts/pgcat/Chart.yaml

@@ -0,0 +1,8 @@
+apiVersion: v2
+name: pgcat
+description: A Helm chart for PgCat a PostgreSQL pooler and proxy (like PgBouncer) with support for sharding, load balancing, failover and mirroring.
+maintainers:
+  - name: PostgresML
+    email: team@postgresml.org
+appVersion: "1.3.0"
+version: 0.2.5

charts/pgcat/templates/NOTES.txt

@@ -0,0 +1,22 @@
+1. Get the application URL by running these commands:
+{{- if .Values.ingress.enabled }}
+{{- range $host := .Values.ingress.hosts }}
+  {{- range .paths }}
+  http{{ if $.Values.ingress.tls }}s{{ end }}://{{ $host.host }}{{ .path }}
+  {{- end }}
+{{- end }}
+{{- else if contains "NodePort" .Values.service.type }}
+  export NODE_PORT=$(kubectl get --namespace {{ .Release.Namespace }} -o jsonpath="{.spec.ports[0].nodePort}" services {{ include "pgcat.fullname" . }})
+  export NODE_IP=$(kubectl get nodes --namespace {{ .Release.Namespace }} -o jsonpath="{.items[0].status.addresses[0].address}")
+  echo http://$NODE_IP:$NODE_PORT
+{{- else if contains "LoadBalancer" .Values.service.type }}
+     NOTE: It may take a few minutes for the LoadBalancer IP to be available.
+           You can watch the status of by running 'kubectl get --namespace {{ .Release.Namespace }} svc -w {{ include "pgcat.fullname" . }}'
+  export SERVICE_IP=$(kubectl get svc --namespace {{ .Release.Namespace }} {{ include "pgcat.fullname" . }} --template "{{"{{ range (index .status.loadBalancer.ingress 0) }}{{.}}{{ end }}"}}")
+  echo http://$SERVICE_IP:{{ .Values.service.port }}
+{{- else if contains "ClusterIP" .Values.service.type }}
+  export POD_NAME=$(kubectl get pods --namespace {{ .Release.Namespace }} -l "app.kubernetes.io/name={{ include "pgcat.name" . }},app.kubernetes.io/instance={{ .Release.Name }}" -o jsonpath="{.items[0].metadata.name}")
+  export CONTAINER_PORT=$(kubectl get pod --namespace {{ .Release.Namespace }} $POD_NAME -o jsonpath="{.spec.containers[0].ports[0].containerPort}")
+  echo "Visit http://127.0.0.1:8080 to use your application"
+  kubectl --namespace {{ .Release.Namespace }} port-forward $POD_NAME 8080:$CONTAINER_PORT
+{{- end }}

charts/pgcat/templates/_config.tpl

@@ -0,0 +1,3 @@
+{{/*
+  Configuration template definition
+*/}}

charts/pgcat/templates/_helpers.tpl

@@ -0,0 +1,62 @@
+{{/*
+Expand the name of the chart.
+*/}}
+{{- define "pgcat.name" -}}
+{{- default .Chart.Name .Values.nameOverride | trunc 63 | trimSuffix "-" }}
+{{- end }}
+
+{{/*
+Create a default fully qualified app name.
+We truncate at 63 chars because some Kubernetes name fields are limited to this (by the DNS naming spec).
+If release name contains chart name it will be used as a full name.
+*/}}
+{{- define "pgcat.fullname" -}}
+{{- if .Values.fullnameOverride }}
+{{- .Values.fullnameOverride | trunc 63 | trimSuffix "-" }}
+{{- else }}
+{{- $name := default .Chart.Name .Values.nameOverride }}
+{{- if contains $name .Release.Name }}
+{{- .Release.Name | trunc 63 | trimSuffix "-" }}
+{{- else }}
+{{- printf "%s-%s" .Release.Name $name | trunc 63 | trimSuffix "-" }}
+{{- end }}
+{{- end }}
+{{- end }}
+
+{{/*
+Create chart name and version as used by the chart label.
+*/}}
+{{- define "pgcat.chart" -}}
+{{- printf "%s-%s" .Chart.Name .Chart.Version | replace "+" "_" | trunc 63 | trimSuffix "-" }}
+{{- end }}
+
+{{/*
+Common labels
+*/}}
+{{- define "pgcat.labels" -}}
+helm.sh/chart: {{ include "pgcat.chart" . }}
+{{ include "pgcat.selectorLabels" . }}
+{{- if .Chart.AppVersion }}
+app.kubernetes.io/version: {{ .Chart.AppVersion | quote }}
+{{- end }}
+app.kubernetes.io/managed-by: {{ .Release.Service }}
+{{- end }}
+
+{{/*
+Selector labels
+*/}}
+{{- define "pgcat.selectorLabels" -}}
+app.kubernetes.io/name: {{ include "pgcat.name" . }}
+app.kubernetes.io/instance: {{ .Release.Name }}
+{{- end }}
+
+{{/*
+Create the name of the service account to use
+*/}}
+{{- define "pgcat.serviceAccountName" -}}
+{{- if .Values.serviceAccount.create }}
+{{- default (include "pgcat.fullname" .) .Values.serviceAccount.name }}
+{{- else }}
+{{- default "default" .Values.serviceAccount.name }}
+{{- end }}
+{{- end }}

charts/pgcat/templates/deployment.yaml

@@ -0,0 +1,66 @@
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: {{ include "pgcat.fullname" . }}
+  labels:
+    {{- include "pgcat.labels" . | nindent 4 }}
+spec:
+  replicas: {{ .Values.replicaCount }}
+  selector:
+    matchLabels:
+      {{- include "pgcat.selectorLabels" . | nindent 6 }}
+  template:
+    metadata:
+      annotations:
+        checksum/secret: {{ include (print $.Template.BasePath "/secret.yaml") . | sha256sum }}
+      {{- with .Values.podAnnotations }}
+        {{- toYaml . | nindent 8 }}
+      {{- end }}
+      labels:
+        {{- include "pgcat.selectorLabels" . | nindent 8 }}
+    spec:
+      {{- with .Values.image.pullSecrets }}
+      imagePullSecrets:
+        {{- toYaml . | nindent 8 }}
+      {{- end }}
+      serviceAccountName: {{ include "pgcat.serviceAccountName" . }}
+      securityContext:
+        {{- toYaml .Values.podSecurityContext | nindent 8 }}
+      containers:
+        - name: {{ .Chart.Name }}
+          securityContext:
+            {{- toYaml .Values.containerSecurityContext | nindent 12 }}
+          image: "{{ .Values.image.repository }}:{{ .Values.image.tag | default .Chart.AppVersion }}"
+          imagePullPolicy: {{ .Values.image.pullPolicy }}
+          ports:
+            - name: pgcat
+              containerPort: {{ .Values.configuration.general.port }}
+              protocol: TCP
+          livenessProbe:
+            tcpSocket:
+              port: pgcat
+          readinessProbe:
+            tcpSocket:
+              port: pgcat
+          resources:
+            {{- toYaml .Values.resources | nindent 12 }}
+          volumeMounts:
+          - mountPath: /etc/pgcat
+            name: config
+      {{- with .Values.nodeSelector }}
+      nodeSelector:
+        {{- toYaml . | nindent 8 }}
+      {{- end }}
+      {{- with .Values.affinity }}
+      affinity:
+        {{- toYaml . | nindent 8 }}
+      {{- end }}
+      {{- with .Values.tolerations }}
+      tolerations:
+        {{- toYaml . | nindent 8 }}
+      {{- end }}
+      volumes:
+      - secret:
+          defaultMode: 420
+          secretName: {{ include "pgcat.fullname" . }}
+        name: config

charts/pgcat/templates/ingress.yaml

@@ -0,0 +1,61 @@
+{{- if .Values.ingress.enabled -}}
+{{- $fullName := include "pgcat.fullname" . -}}
+{{- $svcPort := .Values.service.port -}}
+{{- if and .Values.ingress.className (not (semverCompare ">=1.18-0" .Capabilities.KubeVersion.GitVersion)) }}
+  {{- if not (hasKey .Values.ingress.annotations "kubernetes.io/ingress.class") }}
+  {{- $_ := set .Values.ingress.annotations "kubernetes.io/ingress.class" .Values.ingress.className}}
+  {{- end }}
+{{- end }}
+{{- if semverCompare ">=1.19-0" .Capabilities.KubeVersion.GitVersion -}}
+apiVersion: networking.k8s.io/v1
+{{- else if semverCompare ">=1.14-0" .Capabilities.KubeVersion.GitVersion -}}
+apiVersion: networking.k8s.io/v1beta1
+{{- else -}}
+apiVersion: extensions/v1beta1
+{{- end }}
+kind: Ingress
+metadata:
+  name: {{ $fullName }}
+  labels:
+    {{- include "pgcat.labels" . | nindent 4 }}
+  {{- with .Values.ingress.annotations }}
+  annotations:
+    {{- toYaml . | nindent 4 }}
+  {{- end }}
+spec:
+  {{- if and .Values.ingress.className (semverCompare ">=1.18-0" .Capabilities.KubeVersion.GitVersion) }}
+  ingressClassName: {{ .Values.ingress.className }}
+  {{- end }}
+  {{- if .Values.ingress.tls }}
+  tls:
+    {{- range .Values.ingress.tls }}
+    - hosts:
+        {{- range .hosts }}
+        - {{ . | quote }}
+        {{- end }}
+      secretName: {{ .secretName }}
+    {{- end }}
+  {{- end }}
+  rules:
+    {{- range .Values.ingress.hosts }}
+    - host: {{ .host | quote }}
+      http:
+        paths:
+          {{- range .paths }}
+          - path: {{ .path }}
+            {{- if and .pathType (semverCompare ">=1.18-0" $.Capabilities.KubeVersion.GitVersion) }}
+            pathType: {{ .pathType }}
+            {{- end }}
+            backend:
+              {{- if semverCompare ">=1.19-0" $.Capabilities.KubeVersion.GitVersion }}
+              service:
+                name: {{ $fullName }}
+                port:
+                  number: {{ $svcPort }}
+              {{- else }}
+              serviceName: {{ $fullName }}
+              servicePort: {{ $svcPort }}
+              {{- end }}
+          {{- end }}
+    {{- end }}
+{{- end }}

charts/pgcat/templates/secret.yaml

@@ -0,0 +1,101 @@
+apiVersion: v1
+kind: Secret
+metadata:
+  name: {{ include "pgcat.fullname" . }}
+  labels:
+    {{- include "pgcat.labels" . | nindent 4 }}
+type: Opaque
+stringData:
+  pgcat.toml: |
+    [general]
+    host = {{ .Values.configuration.general.host | quote }}
+    port = {{ .Values.configuration.general.port }}
+    enable_prometheus_exporter = {{ .Values.configuration.general.enable_prometheus_exporter }}
+    prometheus_exporter_port = {{ .Values.configuration.general.prometheus_exporter_port }}
+    connect_timeout = {{ .Values.configuration.general.connect_timeout }}
+    idle_timeout = {{ .Values.configuration.general.idle_timeout | int }}
+    server_lifetime = {{ .Values.configuration.general.server_lifetime | int }}
+    server_tls = {{ .Values.configuration.general.server_tls }}
+    idle_client_in_transaction_timeout = {{ .Values.configuration.general.idle_client_in_transaction_timeout | int }}
+    healthcheck_timeout = {{ .Values.configuration.general.healthcheck_timeout }}
+    healthcheck_delay = {{ .Values.configuration.general.healthcheck_delay }}
+    shutdown_timeout = {{ .Values.configuration.general.shutdown_timeout }}
+    ban_time = {{ .Values.configuration.general.ban_time }}
+    log_client_connections = {{ .Values.configuration.general.log_client_connections }}
+    log_client_disconnections = {{ .Values.configuration.general.log_client_disconnections }}
+    tcp_keepalives_idle = {{ .Values.configuration.general.tcp_keepalives_idle }}
+    tcp_keepalives_count = {{ .Values.configuration.general.tcp_keepalives_count }}
+    tcp_keepalives_interval = {{ .Values.configuration.general.tcp_keepalives_interval }}
+    {{- if and (ne .Values.configuration.general.tls_certificate "-") (ne .Values.configuration.general.tls_private_key "-") }}
+    tls_certificate = "{{ .Values.configuration.general.tls_certificate }}"
+    tls_private_key = "{{ .Values.configuration.general.tls_private_key }}"
+    {{- end }}
+    admin_username = {{ .Values.configuration.general.admin_username | quote }}
+    admin_password = {{ .Values.configuration.general.admin_password | quote }}
+    {{- if and .Values.configuration.general.auth_query_user .Values.configuration.general.auth_query_password .Values.configuration.general.auth_query }}
+    auth_query = {{ .Values.configuration.general.auth_query | quote }}
+    auth_query_user = {{ .Values.configuration.general.auth_query_user | quote }}
+    auth_query_password = {{ .Values.configuration.general.auth_query_password | quote }}
+    {{- end }}
+
+    {{- range $pool := .Values.configuration.pools }}
+
+    ##
+    ## pool for {{ $pool.name }}
+    ##
+    [pools.{{ $pool.name | quote }}]
+    pool_mode = {{ default "transaction" $pool.pool_mode | quote }}
+    load_balancing_mode = {{ default "random" $pool.load_balancing_mode | quote }}
+    default_role = {{ default "any" $pool.default_role | quote }}
+    prepared_statements_cache_size = {{ default 500 $pool.prepared_statements_cache_size }}
+    query_parser_enabled = {{ default true $pool.query_parser_enabled }}
+    query_parser_read_write_splitting = {{ default true $pool.query_parser_read_write_splitting }}
+    primary_reads_enabled = {{ default true $pool.primary_reads_enabled }}
+    db_activity_based_routing = {{ default false $pool.db_activity_based_routing }}
+    db_activity_based_ms_init_delay = {{ default 100 $pool.db_activity_based_ms_init_delay }}
+    db_activity_ttl = {{ default 900 $pool.db_activity_ttl }}
+    table_mutation_cache_ttl = {{ default 50 $pool.table_mutation_cache_ttl }}
+    sharding_function = {{ default "pg_bigint_hash" $pool.sharding_function | quote }}
+
+    {{-   range $index, $user := $pool.users }}
+
+    ## pool {{ $pool.name }} user {{ $user.username | quote }}
+    ##
+    [pools.{{ $pool.name | quote }}.users.{{ $index }}]
+    username = {{ $user.username | quote }}
+    {{- if $user.password }}
+    password = {{ $user.password | quote }}
+    {{- else if and $user.passwordSecret.name $user.passwordSecret.key }}
+    {{- $secret := (lookup "v1" "Secret" $.Release.Namespace $user.passwordSecret.name) }}
+    {{- if $secret }}
+    {{- $password := index $secret.data $user.passwordSecret.key | b64dec }}
+    password = {{ $password | quote }}
+    {{- end }}
+    {{- end }}
+    pool_size = {{ $user.pool_size }}
+    statement_timeout = {{ default 0 $user.statement_timeout }}
+    min_pool_size = {{ default 3 $user.min_pool_size }}
+    {{- if $user.server_lifetime }}
+    server_lifetime = {{ $user.server_lifetime }}
+    {{- end }}
+    {{-     if and $user.server_username $user.server_password }}
+    server_username = {{ $user.server_username | quote }}
+    server_password = {{ $user.server_password | quote }}
+    {{-     end }}
+    {{-   end }}
+
+    {{-   range $index, $shard := $pool.shards }}
+
+    ## pool {{ $pool.name }} database {{ $shard.database }}
+    ##
+    [pools.{{ $pool.name | quote }}.shards.{{ $index }}]
+    {{-     if gt (len $shard.servers) 0}}
+    servers = [
+    {{-       range  $server := $shard.servers }}
+        [ {{ $server.host | quote }}, {{ $server.port }}, {{ $server.role | quote }} ],
+    {{-       end }}
+    ]
+    {{-     end }}
+    database = {{ $shard.database | quote }}
+    {{-   end }}
+    {{- end }}

charts/pgcat/templates/service.yaml

@@ -0,0 +1,15 @@
+apiVersion: v1
+kind: Service
+metadata:
+  name: {{ include "pgcat.fullname" . }}
+  labels:
+    {{- include "pgcat.labels" . | nindent 4 }}
+spec:
+  type: {{ .Values.service.type }}
+  ports:
+    - port: {{ .Values.service.port }}
+      targetPort: pgcat
+      protocol: TCP
+      name: pgcat
+  selector:
+    {{- include "pgcat.selectorLabels" . | nindent 4 }}

charts/pgcat/templates/serviceaccount.yaml

@@ -0,0 +1,12 @@
+{{- if .Values.serviceAccount.create -}}
+apiVersion: v1
+kind: ServiceAccount
+metadata:
+  name: {{ include "pgcat.serviceAccountName" . }}
+  labels:
+    {{- include "pgcat.labels" . | nindent 4 }}
+  {{- with .Values.serviceAccount.annotations }}
+  annotations:
+    {{- toYaml . | nindent 4 }}
+  {{- end }}
+{{- end }}

charts/pgcat/values.yaml

@@ -0,0 +1,390 @@
+## String to partially override aspnet-core.fullname template (will maintain the release name)
+## @param nameOverride String to partially override common.names.fullname
+##
+nameOverride: ""
+
+## String to fully override aspnet-core.fullname template
+## @param fullnameOverride String to fully override common.names.fullname
+##
+fullnameOverride: ""
+
+## Number of PgCat replicas to deploy
+## @param replicaCount Number of PgCat replicas to deploy
+replicaCount: 1
+
+## Bitnami PgCat image version
+## ref: https://hub.docker.com/r/bitnami/kubewatch/tags/
+##
+## @param image.registry PgCat image registry
+## @param image.repository PgCat image name
+## @param image.tag PgCat image tag
+## @param image.pullPolicy PgCat image tag
+## @param image.pullSecrets Specify docker-registry secret names as an array
+image:
+  repository: ghcr.io/postgresml/pgcat
+  # Overrides the image tag whose default is the chart appVersion.
+  tag: "main"
+  ## Specify a imagePullPolicy
+  ## Defaults to 'Always' if image tag is 'latest', else set to 'IfNotPresent'
+  ## ref: http://kubernetes.io/docs/user-guide/images/#pre-pulling-images
+  ##
+  pullPolicy: IfNotPresent
+  ## Optionally specify an array of imagePullSecrets.
+  ## Secrets must be manually created in the namespace.
+  ## ref: https://kubernetes.io/docs/tasks/configure-pod-container/pull-image-private-registry/
+  ## Example:
+  ## pullSecrets:
+  ##   - myRegistryKeySecretName
+  ##
+  pullSecrets: []
+
+## Specifies whether a ServiceAccount should be created
+##
+## @param serviceAccount.create Enable the creation of a ServiceAccount for PgCat pods
+## @param serviceAccount.name Name of the created ServiceAccount
+##
+serviceAccount:
+  ## Specifies whether a service account should be created
+  create: true
+  ## Annotations to add to the service account
+  annotations: {}
+  ## The name of the service account to use.
+  ## If not set and create is true, a name is generated using the fullname template
+  name: ""
+
+## Annotations for server pods.
+## ref: https://kubernetes.io/docs/concepts/overview/working-with-objects/annotations/
+##
+## @param podAnnotations  Annotations for PgCat pods
+##
+podAnnotations: {}
+
+## PgCat containers' SecurityContext
+## ref: https://kubernetes.io/docs/tasks/configure-pod-container/security-context/#set-the-security-context-for-a-pod
+##
+## @param podSecurityContext.enabled Enabled PgCat pods' Security Context
+## @param podSecurityContext.fsGroup Set PgCat pod's Security Context fsGroup
+##
+podSecurityContext: {}
+  # fsGroup: 2000
+
+## PgCat pods' Security Context
+## ref: https://kubernetes.io/docs/tasks/configure-pod-container/security-context/#set-the-security-context-for-a-container
+##
+## @param containerSecurityContext.enabled Enabled PgCat containers' Security Context
+## @param containerSecurityContext.runAsUser Set PgCat container's Security Context runAsUser
+## @param containerSecurityContext.runAsNonRoot Set PgCat container's Security Context runAsNonRoot
+##
+containerSecurityContext: {}
+  # capabilities:
+  #   drop:
+  #   - ALL
+  # readOnlyRootFilesystem: true
+  # runAsNonRoot: true
+  # runAsUser: 1000
+
+## PgCat service
+##
+## @param service.type PgCat service type
+## @param service.port PgCat service port
+service:
+  type: ClusterIP
+  port: 6432
+
+ingress:
+  enabled: false
+  className: ""
+  annotations: {}
+    # kubernetes.io/ingress.class: nginx
+    # kubernetes.io/tls-acme: "true"
+  hosts:
+    - host: chart-example.local
+      paths:
+        - path: /
+          pathType: ImplementationSpecific
+  tls: []
+  #  - secretName: chart-example-tls
+  #    hosts:
+  #      - chart-example.local
+
+## PgCat resource requests and limits
+## ref: http://kubernetes.io/docs/user-guide/compute-resources/
+##
+## @skip resources Optional description
+## @disabled-param resources.limits  The resources limits for the PgCat container
+## @disabled-param resources.requests  The requested resources for the PgCat container
+##
+resources:
+  # We usually recommend not to specify default resources and to leave this as a conscious
+  # choice for the user. This also increases chances charts run on environments with little
+  # resources, such as Minikube. If you do want to specify resources, uncomment the following
+  # lines, adjust them as necessary, and remove the curly braces after 'resources:'.
+  limits: {}
+  #   cpu: 100m
+  #   memory: 128Mi
+  requests: {}
+  #   cpu: 100m
+  #   memory: 128Mi
+
+## Node labels for pod assignment. Evaluated as a template.
+## ref: https://kubernetes.io/docs/user-guide/node-selection/
+##
+## @param nodeSelector  Node labels for pod assignment
+##
+nodeSelector: {}
+
+## Tolerations for pod assignment. Evaluated as a template.
+## ref: https://kubernetes.io/docs/concepts/configuration/taint-and-toleration/
+##
+## @param tolerations  Tolerations for pod assignment
+##
+tolerations: []
+
+## Affinity for pod assignment. Evaluated as a template.
+## ref: https://kubernetes.io/docs/concepts/configuration/assign-pod-node/#affinity-and-anti-affinity
+## Note: podAffinityPreset, podAntiAffinityPreset, and nodeAffinityPreset will be ignored when it's set
+##
+## @param affinity  Affinity for pod assignment
+##
+affinity: {}
+
+## PgCat configuration
+## @param configuration [object]
+configuration:
+  ## General pooler settings
+  ## @param [object]
+  general:
+    ## @param configuration.general.host What IP to run on, 0.0.0.0 means accessible from everywhere.
+    host: "0.0.0.0"
+
+    ## @param configuration.general.port Port to run on, same as PgBouncer used in this example.
+    port: 6432
+
+    ## @param configuration.general.enable_prometheus_exporter Whether to enable prometheus exporter or not.
+    enable_prometheus_exporter: false
+
+    ## @param configuration.general.prometheus_exporter_port Port at which prometheus exporter listens on.
+    prometheus_exporter_port: 9930
+
+    # @param configuration.general.connect_timeout How long to wait before aborting a server connection (ms).
+    connect_timeout: 5000
+
+    # 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
+
+    # Whether to use TLS for server connections or not.
+    server_tls: false
+
+    # How long a client is allowed to be idle while in a transaction (ms).
+    idle_client_in_transaction_timeout: 0  # milliseconds
+
+    # @param configuration.general.healthcheck_timeout How much time to give `SELECT 1` health check query to return with a result (ms).
+    healthcheck_timeout: 1000
+
+    # @param configuration.general.healthcheck_delay How long to keep connection available for immediate re-use, without running a healthcheck query on it
+    healthcheck_delay: 30000
+
+    # @param configuration.general.shutdown_timeout How much time to give clients during shutdown before forcibly killing client connections (ms).
+    shutdown_timeout: 60000
+
+    # @param configuration.general.ban_time For how long to ban a server if it fails a health check (seconds).
+    ban_time: 60    # seconds
+
+    # @param configuration.general.log_client_connections If we should log client connections
+    log_client_connections: false
+
+    # @param configuration.general.log_client_disconnections If we should log client disconnections
+    log_client_disconnections: false
+
+    # TLS
+    # tls_certificate: "server.cert"
+    # tls_private_key: "server.key"
+    tls_certificate: "-"
+    tls_private_key: "-"
+
+    # Credentials to access the virtual administrative database (pgbouncer or pgcat)
+    # Connecting to that database allows running commands like `SHOW POOLS`, `SHOW DATABASES`, etc..
+    admin_username: "postgres"
+    admin_password: "postgres"
+
+    # 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: null
+
+    # 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.
+    #
+    # @param configuration.general.auth_query_user
+    auth_query_user: null
+
+    # 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.
+    #
+    # @param configuration.general.auth_query_password
+    auth_query_password: null
+
+    # 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
+
+  ## 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"
+  ## @param [object]
+  pools:
+    [{
+      name: "simple", pool_mode: "transaction",
+      users: [{username: "user", password: "pass", pool_size: 5, statement_timeout: 0}],
+      shards: [{
+        servers: [{host: "postgres", port: 5432, role: "primary"}],
+        database: "postgres"
+      }]
+    }]
+    # - ## default values
+    #   ##
+    #   ##
+    #   ##
+    #   name: "db"
+
+    #  ## Pool mode (see PgBouncer docs for more).
+    #  ## session: one server connection per connected client
+    #  ## transaction: one server connection per client transaction
+    #  ## @param configuration.poolsPostgres.pool_mode
+    #  pool_mode: "transaction"
+
+    #  ## Load balancing mode
+    #  ## `random` selects the server at random
+    #  ## `loc` selects the server with the least outstanding busy connections
+    #  ##
+    #  ## @param configuration.poolsPostgres.load_balancing_mode
+    #  load_balancing_mode: "random"
+
+    #  ## Prepared statements cache size.
+    #  ## TODO: update documentation
+    #  ##
+    #  ## @param configuration.poolsPostgres.prepared_statements_cache_size
+    #  prepared_statements_cache_size: 500
+
+    #  ## 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.
+    #  ## @param configuration.poolsPostgres.default_role
+    #  default_role: "any"
+
+    #  ## Query parser. If 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.
+    #  ## @param configuration.poolsPostgres.query_parser_enabled
+    #  query_parser_enabled: true
+
+    #  ## If the query parser is enabled and this setting is enabled, we'll attempt to
+    #  ## infer the role from the query itself.
+    #  ## @param configuration.poolsPostgres.query_parser_read_write_splitting
+    #  query_parser_read_write_splitting: true
+
+    # ## Db activity based routing. If enabled, we'll route queries to the primary if the table was recently mutated.
+    # ## @param configuration.poolsPostgres.db_activity_based_routing
+    # db_activity_based_routing: false
+
+    # ## DB activity based init delay. How long to wait before starting to route queries to the primary after a table mutation.
+    # ## @param configuration.poolsPostgres.db_activity_based_ms_init_delay
+    # db_activity_based_ms_init_delay: 100
+
+    # ## DB activity TTL. How long before marking the DB as inactive after no mutations or queries.
+    # ## @param configuration.poolsPostgres.db_activity_ttl
+    # db_activity_ttl: 900
+
+    # ## Table mutation cache TTL. How long to keep track of table mutations.
+    # ## @param configuration.poolsPostgres.table_mutation_cache_ttl
+    # table_mutation_cache_ttl: 50
+
+    #  ## 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.
+    #  ## @param configuration.poolsPostgres.primary_reads_enabled
+    #  primary_reads_enabled: true
+
+    #  ## 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
+    #  ##
+    #  ## @param configuration.poolsPostgres.sharding_function
+    #  sharding_function: "pg_bigint_hash"
+
+    #  ## Credentials for users that may connect to this cluster
+    #  ## @param users [array]
+    #  ## @param users[0].username Name of the env var (required)
+    #  ## @param users[0].password Value for the env var (required) leave empty to use existing secret see passwordSecret.name and passwordSecret.key
+    #  ## @param users[0].passwordSecret.name Name of the secret containing the password
+    #  ## @param users[0].passwordSecret.key Key in the secret containing the password
+    #  ## @param users[0].pool_size Maximum number of server connections that can be established for this user
+    #  ## @param users[0].statement_timeout Maximum query duration. Dangerous, but protects against DBs that died in a non-obvious way.
+    #  users: []
+    #    # - username: "user"
+    #    #   password: "pass"
+    #    #
+    #    #  # 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.
+    #    #  statement_timeout: 0
+    #    #
+    #    #  # PostgreSQL username used to connect to the server.
+    #    #  server_username: "postgres
+    #    #
+    #    #  # PostgreSQL password used to connect to the server.
+    #    #  server_password: "postgres
+
+    #  ## @param shards [array]
+    #  ## @param shards[0].server[0].host Host for this shard
+    #  ## @param shards[0].server[0].port Port for this shard
+    #  ## @param shards[0].server[0].role Role for this shard
+    #  shards: []
+    #      # [ host, port, role ]
+    #    # - servers:
+    #    #     - host: "postgres"
+    #    #       port: 5432
+    #    #       role: "primary"
+    #    #     - host: "postgres"
+    #    #       port: 5432
+    #    #       role: "replica"
+    #    #   database: "postgres"
+    #    #   # [ host, port, role ]
+    #    # - servers:
+    #    #     - host: "postgres"
+    #    #       port: 5432
+    #    #       role: "primary"
+    #    #     - host: "postgres"
+    #    #       port: 5432
+    #    #       role: "replica"
+    #    #   database: "postgres"
+    #    #   # [ host, port, role ]
+    #    # - servers:
+    #    #     - host: "postgres"
+    #    #       port: 5432
+    #    #       role: "primary"
+    #    #     - host: "postgres"
+    #    #       port: 5432
+    #    #       role: "replica"
+    #    #   database: "postgres"

cr.yaml

@@ -0,0 +1,2 @@
+sign: false
+pages_branch: main

ct.yaml

@@ -0,0 +1,5 @@
+remote: origin
+target-branch: main
+chart-dirs:
+  - charts
+

dev/Dockerfile

@@ -1,4 +1,4 @@
-FROM rust:1.70-bullseye
+FROM rust:bullseye
 
 # Dependencies
 COPY --from=sclevine/yj /bin/yj /bin/yj

pgcat.service

@@ -11,6 +11,7 @@ RestartSec=1
 Environment=RUST_LOG=info
 LimitNOFILE=65536
 ExecStart=/usr/bin/pgcat /etc/pgcat.toml
+ExecReload=/bin/kill -SIGHUP $MAINPID
 
 [Install]
 WantedBy=multi-user.target

pgcat.toml

@@ -179,7 +179,7 @@ primary_reads_enabled = true
 # `random`: picks a shard at random
 # `random_healthy`: picks a shard at random favoring shards with the least number of recent errors
 # `shard_<number>`: e.g. shard_0, shard_4, etc. picks a specific shard, everytime
-# no_shard_specified_behavior = "shard_0"
+# default_shard = "shard_0"
 
 # 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?
@@ -301,6 +301,8 @@ username = "other_user"
 password = "other_user"
 pool_size = 21
 statement_timeout = 15000
+connect_timeout = 1000
+idle_timeout = 1000
 
 # Shard configs are structured as pool.<pool_name>.shards.<shard_id>
 # Each shard config contains a list of servers that make up the shard

postinst

@@ -7,3 +7,7 @@ systemctl enable pgcat
 if ! id pgcat 2> /dev/null; then
 	useradd -s /usr/bin/false pgcat
 fi
+
+if [ -f /etc/pgcat.toml ]; then
+	systemctl start pgcat
+fi

src/admin.rs

@@ -55,7 +55,12 @@ where
 
     let query_parts: Vec<&str> = query.trim_end_matches(';').split_whitespace().collect();
 
-    match query_parts[0].to_ascii_uppercase().as_str() {
+    match query_parts
+        .first()
+        .unwrap_or(&"")
+        .to_ascii_uppercase()
+        .as_str()
+    {
         "BAN" => {
             trace!("BAN");
             ban(stream, query_parts).await
@@ -84,7 +89,12 @@ where
             trace!("SHUTDOWN");
             shutdown(stream).await
         }
-        "SHOW" => match query_parts[1].to_ascii_uppercase().as_str() {
+        "SHOW" => match query_parts
+            .get(1)
+            .unwrap_or(&"")
+            .to_ascii_uppercase()
+            .as_str()
+        {
             "HELP" => {
                 trace!("SHOW HELP");
                 show_help(stream).await
@@ -690,6 +700,8 @@ where
         ("query_count", DataType::Numeric),
         ("error_count", DataType::Numeric),
         ("age_seconds", DataType::Numeric),
+        ("maxwait", DataType::Numeric),
+        ("maxwait_us", DataType::Numeric),
     ];
 
     let new_map = get_client_stats();
@@ -697,6 +709,7 @@ where
     res.put(row_description(&columns));
 
     for (_, client) in new_map {
+        let max_wait = client.max_wait_time.load(Ordering::Relaxed);
         let row = vec![
             format!("{:#010X}", client.client_id()),
             client.pool_name(),
@@ -710,6 +723,8 @@ where
                 .duration_since(client.connect_time())
                 .as_secs()
                 .to_string(),
+            (max_wait / 1_000_000).to_string(),
+            (max_wait % 1_000_000).to_string(),
         ];
 
         res.put(data_row(&row));

src/auth_passthrough.rs

@@ -1,3 +1,4 @@
+use crate::config::AuthType;
 use crate::errors::Error;
 use crate::pool::ConnectionPool;
 use crate::server::Server;
@@ -71,6 +72,7 @@ impl AuthPassthrough {
     pub async fn fetch_hash(&self, address: &crate::config::Address) -> Result<String, Error> {
         let auth_user = crate::config::User {
             username: self.user.clone(),
+            auth_type: AuthType::MD5,
             password: Some(self.password.clone()),
             server_username: None,
             server_password: None,
@@ -79,6 +81,8 @@ impl AuthPassthrough {
             pool_mode: None,
             server_lifetime: None,
             min_pool_size: None,
+            connect_timeout: None,
+            idle_timeout: None,
         };
 
         let user = &address.username;

src/client.rs

@@ -14,7 +14,9 @@ use tokio::sync::mpsc::Sender;
 
 use crate::admin::{generate_server_parameters_for_admin, handle_admin};
 use crate::auth_passthrough::refetch_auth_hash;
-use crate::config::{get_config, get_idle_client_in_transaction_timeout, Address, PoolMode};
+use crate::config::{
+    get_config, get_idle_client_in_transaction_timeout, Address, AuthType, PoolMode,
+};
 use crate::constants::*;
 use crate::messages::*;
 use crate::plugins::PluginOutput;
@@ -463,8 +465,8 @@ where
             .count()
             == 1;
 
-        // Kick any client that's not admin while we're in admin-only mode.
         if !admin && admin_only {
+            // Kick any client that's not admin while we're in admin-only mode.
             debug!(
                 "Rejecting non-admin connection to {} when in admin only mode",
                 pool_name
@@ -481,72 +483,76 @@ where
         let process_id: i32 = rand::random();
         let secret_key: i32 = rand::random();
 
-        // Perform MD5 authentication.
-        // TODO: Add SASL support.
-        let salt = md5_challenge(&mut write).await?;
-
-        let code = match read.read_u8().await {
-            Ok(p) => p,
-            Err(_) => {
-                return Err(Error::ClientSocketError(
-                    "password code".into(),
-                    client_identifier,
-                ))
-            }
-        };
-
-        // PasswordMessage
-        if code as char != 'p' {
-            return Err(Error::ProtocolSyncError(format!(
-                "Expected p, got {}",
-                code as char
-            )));
-        }
-
-        let len = match read.read_i32().await {
-            Ok(len) => len,
-            Err(_) => {
-                return Err(Error::ClientSocketError(
-                    "password message length".into(),
-                    client_identifier,
-                ))
-            }
-        };
-
-        let mut password_response = vec![0u8; (len - 4) as usize];
-
-        match read.read_exact(&mut password_response).await {
-            Ok(_) => (),
-            Err(_) => {
-                return Err(Error::ClientSocketError(
-                    "password message".into(),
-                    client_identifier,
-                ))
-            }
-        };
-
         let mut prepared_statements_enabled = false;
 
         // Authenticate admin user.
         let (transaction_mode, mut server_parameters) = if admin {
             let config = get_config();
+            // TODO: Add SASL support.
+            // Perform MD5 authentication.
+            match config.general.admin_auth_type {
+                AuthType::Trust => (),
+                AuthType::MD5 => {
+                    let salt = md5_challenge(&mut write).await?;
 
-            // Compare server and client hashes.
-            let password_hash = md5_hash_password(
-                &config.general.admin_username,
-                &config.general.admin_password,
-                &salt,
-            );
+                    let code = match read.read_u8().await {
+                        Ok(p) => p,
+                        Err(_) => {
+                            return Err(Error::ClientSocketError(
+                                "password code".into(),
+                                client_identifier,
+                            ))
+                        }
+                    };
 
-            if password_hash != password_response {
-                let error = Error::ClientGeneralError("Invalid password".into(), client_identifier);
+                    // PasswordMessage
+                    if code as char != 'p' {
+                        return Err(Error::ProtocolSyncError(format!(
+                            "Expected p, got {}",
+                            code as char
+                        )));
+                    }
 
-                warn!("{}", error);
-                wrong_password(&mut write, username).await?;
+                    let len = match read.read_i32().await {
+                        Ok(len) => len,
+                        Err(_) => {
+                            return Err(Error::ClientSocketError(
+                                "password message length".into(),
+                                client_identifier,
+                            ))
+                        }
+                    };
 
-                return Err(error);
+                    let mut password_response = vec![0u8; (len - 4) as usize];
+
+                    match read.read_exact(&mut password_response).await {
+                        Ok(_) => (),
+                        Err(_) => {
+                            return Err(Error::ClientSocketError(
+                                "password message".into(),
+                                client_identifier,
+                            ))
+                        }
+                    };
+
+                    // Compare server and client hashes.
+                    let password_hash = md5_hash_password(
+                        &config.general.admin_username,
+                        &config.general.admin_password,
+                        &salt,
+                    );
+
+                    if password_hash != password_response {
+                        let error =
+                            Error::ClientGeneralError("Invalid password".into(), client_identifier);
+
+                        warn!("{}", error);
+                        wrong_password(&mut write, username).await?;
+
+                        return Err(error);
+                    }
+                }
             }
-
             (false, generate_server_parameters_for_admin())
         }
         // Authenticate normal user.
@@ -573,92 +579,143 @@ where
             // Obtain the hash to compare, we give preference to that written in cleartext in config
             // if there is nothing set in cleartext and auth passthrough (auth_query) is configured, we use the hash obtained
             // when the pool was created. If there is no hash there, we try to fetch it one more time.
-            let password_hash = if let Some(password) = &pool.settings.user.password {
-                Some(md5_hash_password(username, password, &salt))
-            } else {
-                if !get_config().is_auth_query_configured() {
-                    wrong_password(&mut write, username).await?;
-                    return Err(Error::ClientAuthImpossible(username.into()));
-                }
+            match pool.settings.user.auth_type {
+                AuthType::Trust => (),
+                AuthType::MD5 => {
+                    // Perform MD5 authentication.
+                    // TODO: Add SASL support.
+                    let salt = md5_challenge(&mut write).await?;
 
-                let mut hash = (*pool.auth_hash.read()).clone();
+                    let code = match read.read_u8().await {
+                        Ok(p) => p,
+                        Err(_) => {
+                            return Err(Error::ClientSocketError(
+                                "password code".into(),
+                                client_identifier,
+                            ))
+                        }
+                    };
 
-                if hash.is_none() {
-                    warn!(
-                        "Query auth configured \
-                          but no hash password found \
-                          for pool {}. Will try to refetch it.",
-                        pool_name
-                    );
+                    // PasswordMessage
+                    if code as char != 'p' {
+                        return Err(Error::ProtocolSyncError(format!(
+                            "Expected p, got {}",
+                            code as char
+                        )));
+                    }
 
-                    match refetch_auth_hash(&pool).await {
-                        Ok(fetched_hash) => {
-                            warn!("Password for {}, obtained. Updating.", client_identifier);
+                    let len = match read.read_i32().await {
+                        Ok(len) => len,
+                        Err(_) => {
+                            return Err(Error::ClientSocketError(
+                                "password message length".into(),
+                                client_identifier,
+                            ))
+                        }
+                    };
+
+                    let mut password_response = vec![0u8; (len - 4) as usize];
+
+                    match read.read_exact(&mut password_response).await {
+                        Ok(_) => (),
+                        Err(_) => {
+                            return Err(Error::ClientSocketError(
+                                "password message".into(),
+                                client_identifier,
+                            ))
+                        }
+                    };
+
+                    let password_hash = if let Some(password) = &pool.settings.user.password {
+                        Some(md5_hash_password(username, password, &salt))
+                    } else {
+                        if !get_config().is_auth_query_configured() {
+                            wrong_password(&mut write, username).await?;
+                            return Err(Error::ClientAuthImpossible(username.into()));
+                        }
+
+                        let mut hash = (*pool.auth_hash.read()).clone();
+
+                        if hash.is_none() {
+                            warn!(
+                                "Query auth configured \
+                                  but no hash password found \
+                                  for pool {}. Will try to refetch it.",
+                                pool_name
+                            );
+
+                            match refetch_auth_hash(&pool).await {
+                                Ok(fetched_hash) => {
+                                    warn!(
+                                        "Password for {}, obtained. Updating.",
+                                        client_identifier
+                                    );
+
+                                    {
+                                        let mut pool_auth_hash = pool.auth_hash.write();
+                                        *pool_auth_hash = Some(fetched_hash.clone());
+                                    }
+
+                                    hash = Some(fetched_hash);
+                                }
+
+                                Err(err) => {
+                                    wrong_password(&mut write, username).await?;
+
+                                    return Err(Error::ClientAuthPassthroughError(
+                                        err.to_string(),
+                                        client_identifier,
+                                    ));
+                                }
+                            }
+                        };
+
+                        Some(md5_hash_second_pass(&hash.unwrap(), &salt))
+                    };
+
+                    // Once we have the resulting hash, we compare with what the client gave us.
+                    // If they do not match and auth query is set up, we try to refetch the hash one more time
+                    // to see if the password has changed since the pool was created.
+                    //
+                    // @TODO: we could end up fetching again the same password twice (see above).
+                    if password_hash.unwrap() != password_response {
+                        warn!(
+                            "Invalid password {}, will try to refetch it.",
+                            client_identifier
+                        );
+
+                        let fetched_hash = match refetch_auth_hash(&pool).await {
+                            Ok(fetched_hash) => fetched_hash,
+                            Err(err) => {
+                                wrong_password(&mut write, username).await?;
+
+                                return Err(err);
+                            }
+                        };
+
+                        let new_password_hash = md5_hash_second_pass(&fetched_hash, &salt);
+
+                        // Ok password changed in server an auth is possible.
+                        if new_password_hash == password_response {
+                            warn!(
+                                "Password for {}, changed in server. Updating.",
+                                client_identifier
+                            );
 
                             {
                                 let mut pool_auth_hash = pool.auth_hash.write();
-                                *pool_auth_hash = Some(fetched_hash.clone());
+                                *pool_auth_hash = Some(fetched_hash);
                             }
-
-                            hash = Some(fetched_hash);
-                        }
-
-                        Err(err) => {
+                        } else {
                             wrong_password(&mut write, username).await?;
-
-                            return Err(Error::ClientAuthPassthroughError(
-                                err.to_string(),
+                            return Err(Error::ClientGeneralError(
+                                "Invalid password".into(),
                                 client_identifier,
                             ));
                         }
                     }
-                };
-
-                Some(md5_hash_second_pass(&hash.unwrap(), &salt))
-            };
-
-            // Once we have the resulting hash, we compare with what the client gave us.
-            // If they do not match and auth query is set up, we try to refetch the hash one more time
-            // to see if the password has changed since the pool was created.
-            //
-            // @TODO: we could end up fetching again the same password twice (see above).
-            if password_hash.unwrap() != password_response {
-                warn!(
-                    "Invalid password {}, will try to refetch it.",
-                    client_identifier
-                );
-
-                let fetched_hash = match refetch_auth_hash(&pool).await {
-                    Ok(fetched_hash) => fetched_hash,
-                    Err(err) => {
-                        wrong_password(&mut write, username).await?;
-
-                        return Err(err);
-                    }
-                };
-
-                let new_password_hash = md5_hash_second_pass(&fetched_hash, &salt);
-
-                // Ok password changed in server an auth is possible.
-                if new_password_hash == password_response {
-                    warn!(
-                        "Password for {}, changed in server. Updating.",
-                        client_identifier
-                    );
-
-                    {
-                        let mut pool_auth_hash = pool.auth_hash.write();
-                        *pool_auth_hash = Some(fetched_hash);
-                    }
-                } else {
-                    wrong_password(&mut write, username).await?;
-                    return Err(Error::ClientGeneralError(
-                        "Invalid password".into(),
-                        client_identifier,
-                    ));
                 }
             }
-
             let transaction_mode = pool.settings.pool_mode == PoolMode::Transaction;
             prepared_statements_enabled =
                 transaction_mode && pool.prepared_statement_cache.is_some();
@@ -802,6 +859,8 @@ where
         // e.g. primary, replica, which shard.
         let mut query_router = QueryRouter::new();
 
+        let mut checkout_failure_count: u64 = 0;
+
         self.stats.register(self.stats.clone());
 
         // Result returned by one of the plugins.
@@ -824,6 +883,7 @@ where
         };
 
         query_router.update_pool_settings(&pool.settings);
+        query_router.set_default_role();
 
         // Our custom protocol loop.
         // We expect the client to either start a transaction with regular queries
@@ -1050,7 +1110,25 @@ where
                         query_router.role(),
                         err
                     );
-
+                    checkout_failure_count += 1;
+                    if let Some(limit) = pool.settings.checkout_failure_limit {
+                        if checkout_failure_count >= limit {
+                            error!(
+                                "Checkout failure limit reached ({} / {}) - disconnecting client",
+                                checkout_failure_count, limit
+                            );
+                            error_response_terminal(
+                                &mut self.write,
+                                &format!(
+                                    "checkout failure limit reached ({} / {})",
+                                    checkout_failure_count, limit
+                                ),
+                            )
+                            .await?;
+                            self.stats.disconnect();
+                            return Ok(());
+                        }
+                    }
                     continue;
                 }
             };
@@ -1149,7 +1227,7 @@ where
                 // This reads the first byte without advancing the internal pointer and mutating the bytes
                 let code = *message.first().unwrap() as char;
 
-                trace!("Message: {}", code);
+                trace!("Client message: {}", code);
 
                 match code {
                     // Query
@@ -1188,6 +1266,7 @@ where
                                 };
                             }
                         }
+
                         debug!("Sending query to server");
 
                         self.send_and_receive_loop(
@@ -1252,7 +1331,7 @@ where
                         self.buffer_describe(message).await?;
                     }
 
-                    // Execute
+                    // Execute2
                     // Execute a prepared statement prepared in `P` and bound in `B`.
                     'E' => {
                         self.extended_protocol_data_buffer
@@ -1320,6 +1399,7 @@ where
                         {
                             match protocol_data {
                                 ExtendedProtocolData::Parse { data, metadata } => {
+                                    debug!("Have parse in extended buffer");
                                     let (parse, hash) = match metadata {
                                         Some(metadata) => metadata,
                                         None => {
@@ -1435,7 +1515,7 @@ where
                             .await
                             {
                                 // We might be in some kind of error/in between protocol state
-                                server.mark_bad();
+                                server.mark_bad(err.to_string().as_str());
                                 return Err(err);
                             }
 
@@ -1502,7 +1582,7 @@ where
                         match write_all_flush(&mut self.write, &response).await {
                             Ok(_) => (),
                             Err(err) => {
-                                server.mark_bad();
+                                server.mark_bad(err.to_string().as_str());
                                 return Err(err);
                             }
                         };
@@ -1656,11 +1736,25 @@ where
     ) -> Result<(), Error> {
         match self.prepared_statements.get(&client_name) {
             Some((parse, hash)) => {
-                debug!("Prepared statement `{}` found in cache", parse.name);
+                debug!("Prepared statement `{}` found in cache", client_name);
                 // In this case we want to send the parse message to the server
                 // since pgcat is initiating the prepared statement on this specific server
-                self.register_parse_to_server_cache(true, hash, parse, pool, server, address)
-                    .await?;
+                match self
+                    .register_parse_to_server_cache(true, hash, parse, pool, server, address)
+                    .await
+                {
+                    Ok(_) => (),
+                    Err(err) => match err {
+                        Error::PreparedStatementError => {
+                            debug!("Removed {} from client cache", client_name);
+                            self.prepared_statements.remove(&client_name);
+                        }
+
+                        _ => {
+                            return Err(err);
+                        }
+                    },
+                }
             }
 
             None => {
@@ -1689,11 +1783,20 @@ where
         // We want to promote this in the pool's LRU
         pool.promote_prepared_statement_hash(hash);
 
+        debug!("Checking for prepared statement {}", parse.name);
+
         if let Err(err) = server
             .register_prepared_statement(parse, should_send_parse_to_server)
             .await
         {
-            pool.ban(address, BanReason::MessageSendFailed, Some(&self.stats));
+            match err {
+                // Don't ban for this.
+                Error::PreparedStatementError => (),
+                _ => {
+                    pool.ban(address, BanReason::MessageSendFailed, Some(&self.stats));
+                }
+            };
+
             return Err(err);
         }
 
@@ -1704,18 +1807,14 @@ where
     /// and also the pool's statement cache. Add it to extended protocol data.
     fn buffer_parse(&mut self, message: BytesMut, pool: &ConnectionPool) -> Result<(), Error> {
         // Avoid parsing if prepared statements not enabled
-        let client_given_name = match self.prepared_statements_enabled {
-            true => Parse::get_name(&message)?,
-            false => "".to_string(),
-        };
-
-        if client_given_name.is_empty() {
+        if !self.prepared_statements_enabled {
             debug!("Anonymous parse message");
             self.extended_protocol_data_buffer
                 .push_back(ExtendedProtocolData::create_new_parse(message, None));
             return Ok(());
         }
 
+        let client_given_name = Parse::get_name(&message)?;
         let parse: Parse = (&message).try_into()?;
 
         // Compute the hash of the parse statement
@@ -1753,18 +1852,15 @@ where
     /// saved in the client cache.
     async fn buffer_bind(&mut self, message: BytesMut) -> Result<(), Error> {
         // Avoid parsing if prepared statements not enabled
-        let client_given_name = match self.prepared_statements_enabled {
-            true => Bind::get_name(&message)?,
-            false => "".to_string(),
-        };
-
-        if client_given_name.is_empty() {
+        if !self.prepared_statements_enabled {
             debug!("Anonymous bind message");
             self.extended_protocol_data_buffer
                 .push_back(ExtendedProtocolData::create_new_bind(message, None));
             return Ok(());
         }
 
+        let client_given_name = Bind::get_name(&message)?;
+
         match self.prepared_statements.get(&client_given_name) {
             Some((rewritten_parse, _)) => {
                 let message = Bind::rename(message, &rewritten_parse.name)?;
@@ -1807,12 +1903,7 @@ where
     /// saved in the client cache.
     async fn buffer_describe(&mut self, message: BytesMut) -> Result<(), Error> {
         // Avoid parsing if prepared statements not enabled
-        let describe: Describe = match self.prepared_statements_enabled {
-            true => (&message).try_into()?,
-            false => Describe::empty_new(),
-        };
-
-        if describe.anonymous() {
+        if !self.prepared_statements_enabled {
             debug!("Anonymous describe message");
             self.extended_protocol_data_buffer
                 .push_back(ExtendedProtocolData::create_new_describe(message, None));
@@ -1820,6 +1911,15 @@ where
             return Ok(());
         }
 
+        let describe: Describe = (&message).try_into()?;
+        if describe.target == 'P' {
+            debug!("Portal describe message");
+            self.extended_protocol_data_buffer
+                .push_back(ExtendedProtocolData::create_new_describe(message, None));
+
+            return Ok(());
+        }
+
         let client_given_name = describe.statement_name.clone();
 
         match self.prepared_statements.get(&client_given_name) {
@@ -1904,7 +2004,7 @@ where
                 Ok(_) => (),
                 Err(err) => {
                     // We might be in some kind of error/in between protocol state, better to just kill this server
-                    server.mark_bad();
+                    server.mark_bad(err.to_string().as_str());
                     return Err(err);
                 }
             };
@@ -1971,11 +2071,13 @@ where
                 }
             },
             Err(_) => {
-                error!(
-                    "Statement timeout while talking to {:?} with user {}",
-                    address, pool.settings.user.username
+                server.mark_bad(
+                    format!(
+                        "Statement timeout while talking to {:?} with user {}",
+                        address, pool.settings.user.username
+                    )
+                    .as_str(),
                 );
-                server.mark_bad();
                 pool.ban(address, BanReason::StatementTimeout, Some(client_stats));
                 error_response_terminal(&mut self.write, "pool statement timeout").await?;
                 Err(Error::StatementTimeout)

src/config.rs

@@ -38,12 +38,12 @@ pub enum Role {
     Mirror,
 }
 
-impl ToString for Role {
-    fn to_string(&self) -> String {
-        match *self {
-            Role::Primary => "primary".to_string(),
-            Role::Replica => "replica".to_string(),
-            Role::Mirror => "mirror".to_string(),
+impl std::fmt::Display for Role {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        match self {
+            Role::Primary => write!(f, "primary"),
+            Role::Replica => write!(f, "replica"),
+            Role::Mirror => write!(f, "mirror"),
         }
     }
 }
@@ -208,6 +208,9 @@ impl Address {
 pub struct User {
     pub username: String,
     pub password: Option<String>,
+
+    #[serde(default = "User::default_auth_type")]
+    pub auth_type: AuthType,
     pub server_username: Option<String>,
     pub server_password: Option<String>,
     pub pool_size: u32,
@@ -216,6 +219,8 @@ pub struct User {
     pub server_lifetime: Option<u64>,
     #[serde(default)] // 0
     pub statement_timeout: u64,
+    pub connect_timeout: Option<u64>,
+    pub idle_timeout: Option<u64>,
 }
 
 impl Default for User {
@@ -223,6 +228,7 @@ impl Default for User {
         User {
             username: String::from("postgres"),
             password: None,
+            auth_type: AuthType::MD5,
             server_username: None,
             server_password: None,
             pool_size: 15,
@@ -230,11 +236,17 @@ impl Default for User {
             statement_timeout: 0,
             pool_mode: None,
             server_lifetime: None,
+            connect_timeout: None,
+            idle_timeout: None,
         }
     }
 }
 
 impl User {
+    pub fn default_auth_type() -> AuthType {
+        AuthType::MD5
+    }
+
     fn validate(&self) -> Result<(), Error> {
         if let Some(min_pool_size) = self.min_pool_size {
             if min_pool_size > self.pool_size {
@@ -330,6 +342,9 @@ pub struct General {
     pub admin_username: String,
     pub admin_password: String,
 
+    #[serde(default = "General::default_admin_auth_type")]
+    pub admin_auth_type: AuthType,
+
     #[serde(default = "General::default_validate_config")]
     pub validate_config: bool,
 
@@ -344,6 +359,10 @@ impl General {
         "0.0.0.0".into()
     }
 
+    pub fn default_admin_auth_type() -> AuthType {
+        AuthType::MD5
+    }
+
     pub fn default_port() -> u16 {
         5432
     }
@@ -452,6 +471,7 @@ impl Default for General {
             verify_server_certificate: false,
             admin_username: String::from("admin"),
             admin_password: String::from("admin"),
+            admin_auth_type: AuthType::MD5,
             validate_config: true,
             auth_query: None,
             auth_query_user: None,
@@ -472,11 +492,20 @@ pub enum PoolMode {
     Session,
 }
 
-impl ToString for PoolMode {
-    fn to_string(&self) -> String {
-        match *self {
-            PoolMode::Transaction => "transaction".to_string(),
-            PoolMode::Session => "session".to_string(),
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Eq, Copy, Hash)]
+pub enum AuthType {
+    #[serde(alias = "trust", alias = "Trust")]
+    Trust,
+
+    #[serde(alias = "md5", alias = "MD5")]
+    MD5,
+}
+
+impl std::fmt::Display for PoolMode {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        match self {
+            PoolMode::Transaction => write!(f, "transaction"),
+            PoolMode::Session => write!(f, "session"),
         }
     }
 }
@@ -489,12 +518,13 @@ pub enum LoadBalancingMode {
     #[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(),
+
+impl std::fmt::Display for LoadBalancingMode {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        match self {
+            LoadBalancingMode::Random => write!(f, "random"),
             LoadBalancingMode::LeastOutstandingConnections => {
-                "least_outstanding_connections".to_string()
+                write!(f, "least_outstanding_connections")
             }
         }
     }
@@ -528,6 +558,14 @@ pub struct Pool {
     /// Close idle connections that have been opened for longer than this.
     pub idle_timeout: Option<u64>,
 
+    /// Maximum number of checkout failures a client is allowed before it
+    /// gets disconnected. This is needed to prevent persistent client/server
+    /// imbalance in high availability setups where multiple PgCat instances are placed
+    /// behind a single load balancer. If for any reason a client lands on a PgCat instance that has
+    /// a large number of connected clients, it might get stuck in perpetual checkout failure loop especially
+    /// in session mode
+    pub checkout_failure_limit: Option<u64>,
+
     /// Close server connections that have been opened for longer than this.
     /// Only applied to idle connections. If the connection is actively used for
     /// longer than this period, the pool will not interrupt it.
@@ -559,6 +597,19 @@ pub struct Pool {
     #[serde(default = "Pool::default_prepared_statements_cache_size")]
     pub prepared_statements_cache_size: usize,
 
+    // Support for query routing based on database activity
+    #[serde(default = "Pool::default_db_activity_based_routing")]
+    pub db_activity_based_routing: bool,
+
+    #[serde(default = "Pool::default_db_activity_init_delay")]
+    pub db_activity_init_delay: u64,
+
+    #[serde(default = "Pool::default_db_activity_ttl")]
+    pub db_activity_ttl: u64,
+
+    #[serde(default = "Pool::default_table_mutation_cache_ms_ttl")]
+    pub table_mutation_cache_ms_ttl: u64,
+
     pub plugins: Option<Plugins>,
     pub shards: BTreeMap<String, Shard>,
     pub users: BTreeMap<String, User>,
@@ -612,6 +663,25 @@ impl Pool {
         0
     }
 
+    pub fn default_db_activity_based_routing() -> bool {
+        false
+    }
+
+    pub fn default_db_activity_init_delay() -> u64 {
+        // 100 milliseconds
+        100
+    }
+
+    pub fn default_db_activity_ttl() -> u64 {
+        // 15 minutes
+        15 * 60
+    }
+
+    pub fn default_table_mutation_cache_ms_ttl() -> u64 {
+        // 50 milliseconds
+        50
+    }
+
     pub fn validate(&mut self) -> Result<(), Error> {
         match self.default_role.as_ref() {
             "any" => (),
@@ -694,6 +764,23 @@ impl Pool {
             user.validate()?;
         }
 
+        if self.db_activity_based_routing {
+            if self.db_activity_init_delay == 0 {
+                error!("db_activity_init_delay must be greater than 0");
+                return Err(Error::BadConfig);
+            }
+
+            if self.table_mutation_cache_ms_ttl == 0 {
+                error!("table_mutation_cache_ms_ttl must be greater than 0");
+                return Err(Error::BadConfig);
+            }
+
+            if self.db_activity_ttl == 0 {
+                error!("db_activity_ttl must be greater than 0");
+                return Err(Error::BadConfig);
+            }
+        }
+
         Ok(())
     }
 }
@@ -703,6 +790,7 @@ impl Default for Pool {
         Pool {
             pool_mode: Self::default_pool_mode(),
             load_balancing_mode: Self::default_load_balancing_mode(),
+            checkout_failure_limit: None,
             default_role: String::from("any"),
             query_parser_enabled: false,
             query_parser_max_length: None,
@@ -723,6 +811,10 @@ impl Default for Pool {
             cleanup_server_connections: true,
             log_client_parameter_status_changes: false,
             prepared_statements_cache_size: Self::default_prepared_statements_cache_size(),
+            db_activity_based_routing: Self::default_db_activity_based_routing(),
+            db_activity_init_delay: Self::default_db_activity_init_delay(),
+            db_activity_ttl: Self::default_db_activity_ttl(),
+            table_mutation_cache_ms_ttl: Self::default_table_mutation_cache_ms_ttl(),
             plugins: None,
             shards: BTreeMap::from([(String::from("1"), Shard::default())]),
             users: BTreeMap::default(),
@@ -995,15 +1087,17 @@ impl Config {
     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();
+                pool.auth_query.clone_from(&self.general.auth_query);
             }
 
             if pool.auth_query_user.is_none() {
-                pool.auth_query_user = self.general.auth_query_user.clone();
+                pool.auth_query_user
+                    .clone_from(&self.general.auth_query_user);
             }
 
             if pool.auth_query_password.is_none() {
-                pool.auth_query_password = self.general.auth_query_password.clone();
+                pool.auth_query_password
+                    .clone_from(&self.general.auth_query_password);
             }
         }
     }
@@ -1151,7 +1245,7 @@ impl Config {
             "Default max server lifetime: {}ms",
             self.general.server_lifetime
         );
-        info!("Sever round robin: {}", self.general.server_round_robin);
+        info!("Server round robin: {}", self.general.server_round_robin);
         match self.general.tls_certificate.clone() {
             Some(tls_certificate) => {
                 info!("TLS certificate: {}", tls_certificate);
@@ -1213,6 +1307,17 @@ impl Config {
                 None => self.general.idle_timeout,
             };
             info!("[pool: {}] Idle timeout: {}ms", pool_name, idle_timeout);
+            match pool_config.checkout_failure_limit {
+                Some(checkout_failure_limit) => {
+                    info!(
+                        "[pool: {}] Checkout failure limit: {}",
+                        pool_name, checkout_failure_limit
+                    );
+                }
+                None => {
+                    info!("[pool: {}] Checkout failure limit: not set", pool_name);
+                }
+            };
             info!(
                 "[pool: {}] Sharding function: {}",
                 pool_name,
@@ -1257,6 +1362,22 @@ impl Config {
                 "[pool: {}] Cleanup server connections: {}",
                 pool_name, pool_config.cleanup_server_connections
             );
+            info!(
+                "[pool: {}] DB activity based routing: {}",
+                pool_name, pool_config.db_activity_based_routing
+            );
+            info!(
+                "[pool: {}] DB activity init delay: {}",
+                pool_name, pool_config.db_activity_init_delay
+            );
+            info!(
+                "[pool: {}] DB activity TTL: {}",
+                pool_name, pool_config.db_activity_ttl
+            );
+            info!(
+                "[pool: {}] Table mutation cache TTL: {}",
+                pool_name, pool_config.table_mutation_cache_ms_ttl
+            );
             info!(
                 "[pool: {}] Log client parameter status changes: {}",
                 pool_name, pool_config.log_client_parameter_status_changes
@@ -1307,6 +1428,24 @@ impl Config {
                         None => "default".to_string(),
                     }
                 );
+                info!(
+                    "[pool: {}][user: {}] Connection timeout: {}",
+                    pool_name,
+                    user.1.username,
+                    match user.1.connect_timeout {
+                        Some(connect_timeout) => format!("{}ms", connect_timeout),
+                        None => "not set".to_string(),
+                    }
+                );
+                info!(
+                    "[pool: {}][user: {}] Idle timeout: {}",
+                    pool_name,
+                    user.1.username,
+                    match user.1.idle_timeout {
+                        Some(idle_timeout) => format!("{}ms", idle_timeout),
+                        None => "not set".to_string(),
+                    }
+                );
             }
         }
     }

src/errors.rs

@@ -29,6 +29,7 @@ pub enum Error {
     QueryRouterParserError(String),
     QueryRouterError(String),
     InvalidShardId(usize),
+    PreparedStatementError,
 }
 
 #[derive(Clone, PartialEq, Debug)]

src/logger.rs

@@ -3,7 +3,7 @@ use tracing_subscriber;
 use tracing_subscriber::EnvFilter;
 
 pub fn init(args: &Args) {
-    // Iniitalize a default filter, and then override the builtin default "warning" with our
+    // Initialize a default filter, and then override the builtin default "warning" with our
     // commandline, (default: "info")
     let filter = EnvFilter::from_default_env().add_directive(args.log_level.into());
 

src/messages.rs

@@ -733,6 +733,10 @@ pub fn configure_socket(stream: &TcpStream) {
         }
         Err(err) => error!("Could not configure socket: {}", err),
     }
+    match sock_ref.set_nodelay(true) {
+        Ok(_) => (),
+        Err(err) => error!("Could not configure TCP_NODELAY for socket: {}", err),
+    }
 }
 
 pub trait BytesMutReader {
@@ -1109,7 +1113,7 @@ pub struct Describe {
 
     #[allow(dead_code)]
     len: i32,
-    target: char,
+    pub target: char,
     pub statement_name: String,
 }
 

src/mirrors.rs

@@ -85,8 +85,9 @@ impl MirroredClient {
                         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());
+                                server.mark_bad(
+                                    format!("Failed to send to mirror, Discarding message {:?}, {:?}", err, address.clone()).as_str()
+                                );
                             }
                         }
                     }
@@ -98,8 +99,9 @@ impl MirroredClient {
                                 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())
+                                        server.mark_bad(
+                                            format!("Failed to receive from mirror {:?} {:?}", err, address.clone()).as_str()
+                                        );
                                     }
                                 }
                             }

src/pool.rs

@@ -152,6 +152,14 @@ pub struct PoolSettings {
     /// Random or LeastOutstandingConnections.
     pub load_balancing_mode: LoadBalancingMode,
 
+    /// Maximum number of checkout failures a client is allowed before it
+    /// gets disconnected. This is needed to prevent persistent client/server
+    /// imbalance in high availability setups where multiple PgCat instances are placed
+    /// behind a single load balancer. If for any reason a client lands on a PgCat instance that has
+    /// a large number of connected clients, it might get stuck in perpetual checkout failure loop especially
+    /// in session mode
+    pub checkout_failure_limit: Option<u64>,
+
     // Number of shards.
     pub shards: usize,
 
@@ -174,6 +182,18 @@ pub struct PoolSettings {
     // Read from the primary as well or not.
     pub primary_reads_enabled: bool,
 
+    // Automatic primary/replica selection based on recent activity.
+    pub db_activity_based_routing: bool,
+
+    // DB activity init delay
+    pub db_activity_init_delay: u64,
+
+    // DB activity TTL
+    pub db_activity_ttl: u64,
+
+    // Table mutation cache TTL
+    pub table_mutation_cache_ms_ttl: u64,
+
     // Sharding function.
     pub sharding_function: ShardingFunction,
 
@@ -215,6 +235,7 @@ impl Default for PoolSettings {
         PoolSettings {
             pool_mode: PoolMode::Transaction,
             load_balancing_mode: LoadBalancingMode::Random,
+            checkout_failure_limit: None,
             shards: 1,
             user: User::default(),
             db: String::default(),
@@ -223,6 +244,10 @@ impl Default for PoolSettings {
             query_parser_max_length: None,
             query_parser_read_write_splitting: false,
             primary_reads_enabled: true,
+            db_activity_based_routing: false,
+            db_activity_init_delay: 100,
+            db_activity_ttl: 15 * 60,
+            table_mutation_cache_ms_ttl: 50,
             sharding_function: ShardingFunction::PgBigintHash,
             automatic_sharding_key: None,
             healthcheck_delay: General::default_healthcheck_delay(),
@@ -436,14 +461,20 @@ impl ConnectionPool {
                             pool_config.prepared_statements_cache_size,
                         );
 
-                        let connect_timeout = match pool_config.connect_timeout {
+                        let connect_timeout = match user.connect_timeout {
                             Some(connect_timeout) => connect_timeout,
-                            None => config.general.connect_timeout,
+                            None => match pool_config.connect_timeout {
+                                Some(connect_timeout) => connect_timeout,
+                                None => config.general.connect_timeout,
+                            },
                         };
 
-                        let idle_timeout = match pool_config.idle_timeout {
+                        let idle_timeout = match user.idle_timeout {
                             Some(idle_timeout) => idle_timeout,
-                            None => config.general.idle_timeout,
+                            None => match pool_config.idle_timeout {
+                                Some(idle_timeout) => idle_timeout,
+                                None => config.general.idle_timeout,
+                            },
                         };
 
                         let server_lifetime = match user.server_lifetime {
@@ -515,6 +546,7 @@ impl ConnectionPool {
                             None => pool_config.pool_mode,
                         },
                         load_balancing_mode: pool_config.load_balancing_mode,
+                        checkout_failure_limit: pool_config.checkout_failure_limit,
                         // shards: pool_config.shards.clone(),
                         shards: shard_ids.len(),
                         user: user.clone(),
@@ -531,6 +563,10 @@ impl ConnectionPool {
                             .query_parser_read_write_splitting,
                         primary_reads_enabled: pool_config.primary_reads_enabled,
                         sharding_function: pool_config.sharding_function,
+                        db_activity_based_routing: pool_config.db_activity_based_routing,
+                        db_activity_init_delay: pool_config.db_activity_init_delay,
+                        db_activity_ttl: pool_config.db_activity_ttl,
+                        table_mutation_cache_ms_ttl: pool_config.table_mutation_cache_ms_ttl,
                         automatic_sharding_key: pool_config.automatic_sharding_key.clone(),
                         healthcheck_delay: config.general.healthcheck_delay,
                         healthcheck_timeout: config.general.healthcheck_timeout,
@@ -763,7 +799,6 @@ impl ConnectionPool {
                     );
                     self.ban(address, BanReason::FailedCheckout, Some(client_stats));
                     address.stats.error();
-                    client_stats.idle();
                     client_stats.checkout_error();
                     continue;
                 }
@@ -782,7 +817,7 @@ impl ConnectionPool {
             // Health checks are pretty expensive.
             if !require_healthcheck {
                 let checkout_time = now.elapsed().as_micros() as u64;
-                client_stats.checkout_time(checkout_time);
+                client_stats.checkout_success();
                 server
                     .stats()
                     .checkout_time(checkout_time, client_stats.application_name());
@@ -796,7 +831,7 @@ impl ConnectionPool {
                 .await
             {
                 let checkout_time = now.elapsed().as_micros() as u64;
-                client_stats.checkout_time(checkout_time);
+                client_stats.checkout_success();
                 server
                     .stats()
                     .checkout_time(checkout_time, client_stats.application_name());
@@ -808,10 +843,7 @@ impl ConnectionPool {
             }
         }
 
-        client_stats.idle();
-
-        let checkout_time = now.elapsed().as_micros() as u64;
-        client_stats.checkout_time(checkout_time);
+        client_stats.checkout_error();
 
         Err(Error::AllServersDown)
     }
@@ -837,7 +869,7 @@ impl ConnectionPool {
             Ok(res) => match res {
                 Ok(_) => {
                     let checkout_time: u64 = start.elapsed().as_micros() as u64;
-                    client_info.checkout_time(checkout_time);
+                    client_info.checkout_success();
                     server
                         .stats()
                         .checkout_time(checkout_time, client_info.application_name());
@@ -865,7 +897,7 @@ impl ConnectionPool {
         }
 
         // Don't leave a bad connection in the pool.
-        server.mark_bad();
+        server.mark_bad("failed health check");
 
         self.ban(address, BanReason::FailedHealthCheck, Some(client_info));
         false

src/prometheus.rs

@@ -1,23 +1,41 @@
-use hyper::service::{make_service_fn, service_fn};
-use hyper::{Body, Method, Request, Response, Server, StatusCode};
+use http_body_util::Full;
+use hyper::body;
+use hyper::body::Bytes;
+
+use hyper::server::conn::http1;
+use hyper::service::service_fn;
+use hyper::{Method, Request, Response, StatusCode};
+use hyper_util::rt::TokioIo;
 use log::{debug, error, info};
 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 tokio::net::TcpListener;
 
 use crate::config::Address;
 use crate::pool::{get_all_pools, PoolIdentifier};
+use crate::stats::get_server_stats;
 use crate::stats::pool::PoolStats;
-use crate::stats::{get_server_stats, ServerStats};
 
 struct MetricHelpType {
     help: &'static str,
     ty: &'static str,
 }
 
+struct ServerPrometheusStats {
+    bytes_received: u64,
+    bytes_sent: u64,
+    transaction_count: u64,
+    query_count: u64,
+    error_count: u64,
+    active_count: u64,
+    idle_count: u64,
+    login_count: u64,
+    tested_count: u64,
+}
+
 // reference for metric types: https://prometheus.io/docs/concepts/metric_types/
 // counters only increase
 // gauges can arbitrarily increase or decrease
@@ -120,22 +138,46 @@ static METRIC_HELP_AND_TYPES_LOOKUP: phf::Map<&'static str, MetricHelpType> = ph
     },
     "servers_bytes_received" => MetricHelpType {
         help: "Volume in bytes of network traffic received by server",
-        ty: "gauge",
+        ty: "counter",
     },
     "servers_bytes_sent" => MetricHelpType {
         help: "Volume in bytes of network traffic sent by server",
-        ty: "gauge",
+        ty: "counter",
     },
     "servers_transaction_count" => MetricHelpType {
         help: "Number of transactions executed by server",
-        ty: "gauge",
+        ty: "counter",
     },
     "servers_query_count" => MetricHelpType {
         help: "Number of queries executed by server",
-        ty: "gauge",
+        ty: "counter",
     },
     "servers_error_count" => MetricHelpType {
         help: "Number of errors",
+        ty: "counter",
+    },
+    "servers_idle_count" => MetricHelpType {
+        help: "Number of server connection in idle state",
+        ty: "gauge",
+    },
+    "servers_active_count" => MetricHelpType {
+        help: "Number of server connection in active state",
+        ty: "gauge",
+    },
+    "servers_tested_count" => MetricHelpType {
+        help: "Number of server connection in tested state",
+        ty: "gauge",
+    },
+    "servers_login_count" => MetricHelpType {
+        help: "Number of server connection in login state",
+        ty: "gauge",
+    },
+    "servers_is_banned" => MetricHelpType {
+        help: "0 if server is not banned, 1 if server is banned",
+        ty: "gauge",
+    },
+    "servers_is_paused" => MetricHelpType {
+        help: "0 if server is not paused, 1 if server is paused",
         ty: "gauge",
     },
     "databases_pool_size" => MetricHelpType {
@@ -158,18 +200,17 @@ struct PrometheusMetric<Value: fmt::Display> {
 
 impl<Value: fmt::Display> fmt::Display for PrometheusMetric<Value> {
     fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
-        let formatted_labels = self
-            .labels
+        let mut sorted_labels: Vec<_> = self.labels.iter().collect();
+        sorted_labels.sort_by_key(|&(key, _)| key);
+        let formatted_labels = sorted_labels
             .iter()
             .map(|(key, value)| format!("{}=\"{}\"", key, value))
             .collect::<Vec<_>>()
             .join(",");
         write!(
             f,
-            "# HELP {name} {help}\n# TYPE {name} {ty}\n{name}{{{formatted_labels}}} {value}\n",
+            "{name}{{{formatted_labels}}} {value}",
             name = format_args!("pgcat_{}", self.name),
-            help = self.help,
-            ty = self.ty,
             formatted_labels = formatted_labels,
             value = self.value
         )
@@ -203,7 +244,9 @@ impl<Value: fmt::Display> PrometheusMetric<Value> {
         labels.insert("shard", address.shard.to_string());
         labels.insert("role", address.role.to_string());
         labels.insert("pool", address.pool_name.clone());
+        labels.insert("index", address.address_index.to_string());
         labels.insert("database", address.database.to_string());
+        labels.insert("username", address.username.clone());
 
         Self::from_name(&format!("databases_{}", name), value, labels)
     }
@@ -218,7 +261,9 @@ impl<Value: fmt::Display> PrometheusMetric<Value> {
         labels.insert("shard", address.shard.to_string());
         labels.insert("role", address.role.to_string());
         labels.insert("pool", address.pool_name.clone());
+        labels.insert("index", address.address_index.to_string());
         labels.insert("database", address.database.to_string());
+        labels.insert("username", address.username.clone());
 
         Self::from_name(&format!("servers_{}", name), value, labels)
     }
@@ -229,7 +274,9 @@ impl<Value: fmt::Display> PrometheusMetric<Value> {
         labels.insert("shard", address.shard.to_string());
         labels.insert("pool", address.pool_name.clone());
         labels.insert("role", address.role.to_string());
+        labels.insert("index", address.address_index.to_string());
         labels.insert("database", address.database.to_string());
+        labels.insert("username", address.username.clone());
 
         Self::from_name(&format!("stats_{}", name), value, labels)
     }
@@ -241,9 +288,20 @@ impl<Value: fmt::Display> PrometheusMetric<Value> {
 
         Self::from_name(&format!("pools_{}", name), value, labels)
     }
+
+    fn get_header(&self) -> String {
+        format!(
+            "\n# HELP {name} {help}\n# TYPE {name} {ty}",
+            name = format_args!("pgcat_{}", self.name),
+            help = self.help,
+            ty = self.ty,
+        )
+    }
 }
 
-async fn prometheus_stats(request: Request<Body>) -> Result<Response<Body>, hyper::http::Error> {
+async fn prometheus_stats(
+    request: Request<body::Incoming>,
+) -> Result<Response<Full<Bytes>>, hyper::http::Error> {
     match (request.method(), request.uri().path()) {
         (&Method::GET, "/metrics") => {
             let mut lines = Vec::new();
@@ -251,6 +309,7 @@ async fn prometheus_stats(request: Request<Body>) -> Result<Response<Body>, hype
             push_pool_stats(&mut lines);
             push_server_stats(&mut lines);
             push_database_stats(&mut lines);
+            lines.push("".to_string()); // Ensure to end the stats with a line terminator as required by the specification.
 
             Response::builder()
                 .header("content-type", "text/plain; version=0.0.4")
@@ -264,6 +323,7 @@ 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>) {
+    let mut grouped_metrics: HashMap<String, Vec<PrometheusMetric<u64>>> = HashMap::new();
     for (_, pool) in get_all_pools() {
         for shard in 0..pool.shards() {
             for server in 0..pool.servers(shard) {
@@ -273,7 +333,10 @@ fn push_address_stats(lines: &mut Vec<String>) {
                     if let Some(prometheus_metric) =
                         PrometheusMetric::<u64>::from_address(address, &key, value)
                     {
-                        lines.push(prometheus_metric.to_string());
+                        grouped_metrics
+                            .entry(key)
+                            .or_default()
+                            .push(prometheus_metric);
                     } else {
                         debug!("Metric {} not implemented for {}", key, address.name());
                     }
@@ -281,33 +344,53 @@ fn push_address_stats(lines: &mut Vec<String>) {
             }
         }
     }
+    for (_key, metrics) in grouped_metrics {
+        if !metrics.is_empty() {
+            lines.push(metrics[0].get_header());
+            for metric in metrics {
+                lines.push(metric.to_string());
+            }
+        }
+    }
 }
 
 // Adds relevant metrics shown in a SHOW POOLS admin command.
 fn push_pool_stats(lines: &mut Vec<String>) {
+    let mut grouped_metrics: HashMap<String, Vec<PrometheusMetric<u64>>> = HashMap::new();
     let pool_stats = PoolStats::construct_pool_lookup();
     for (pool_id, stats) in pool_stats.iter() {
         for (name, value) in stats.clone() {
             if let Some(prometheus_metric) =
                 PrometheusMetric::<u64>::from_pool(pool_id.clone(), &name, value)
             {
-                lines.push(prometheus_metric.to_string());
+                grouped_metrics
+                    .entry(name)
+                    .or_default()
+                    .push(prometheus_metric);
             } else {
                 debug!("Metric {} not implemented for ({})", name, *pool_id);
             }
         }
     }
+    for (_key, metrics) in grouped_metrics {
+        if !metrics.is_empty() {
+            lines.push(metrics[0].get_header());
+            for metric in metrics {
+                lines.push(metric.to_string());
+            }
+        }
+    }
 }
 
 // Adds relevant metrics shown in a SHOW DATABASES admin command.
 fn push_database_stats(lines: &mut Vec<String>) {
+    let mut grouped_metrics: HashMap<String, Vec<PrometheusMetric<u32>>> = HashMap::new();
     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),
@@ -316,7 +399,10 @@ fn push_database_stats(lines: &mut Vec<String>) {
                     if let Some(prometheus_metric) =
                         PrometheusMetric::<u32>::from_database_info(address, key, value)
                     {
-                        lines.push(prometheus_metric.to_string());
+                        grouped_metrics
+                            .entry(key.to_string())
+                            .or_default()
+                            .push(prometheus_metric);
                     } else {
                         debug!("Metric {} not implemented for {}", key, address.name());
                     }
@@ -324,45 +410,73 @@ fn push_database_stats(lines: &mut Vec<String>) {
             }
         }
     }
+    for (_key, metrics) in grouped_metrics {
+        if !metrics.is_empty() {
+            lines.push(metrics[0].get_header());
+            for metric in metrics {
+                lines.push(metric.to_string());
+            }
+        }
+    }
 }
 
 // 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();
+    let mut prom_stats = HashMap::<String, ServerPrometheusStats>::new();
     for (_, stats) in server_stats {
-        server_stats_by_addresses.insert(stats.address_name(), stats);
+        let entry = prom_stats
+            .entry(stats.address_name())
+            .or_insert(ServerPrometheusStats {
+                bytes_received: 0,
+                bytes_sent: 0,
+                transaction_count: 0,
+                query_count: 0,
+                error_count: 0,
+                active_count: 0,
+                idle_count: 0,
+                login_count: 0,
+                tested_count: 0,
+            });
+        entry.bytes_received += stats.bytes_received.load(Ordering::Relaxed);
+        entry.bytes_sent += stats.bytes_sent.load(Ordering::Relaxed);
+        entry.transaction_count += stats.transaction_count.load(Ordering::Relaxed);
+        entry.query_count += stats.query_count.load(Ordering::Relaxed);
+        entry.error_count += stats.error_count.load(Ordering::Relaxed);
+        match stats.state.load(Ordering::Relaxed) {
+            crate::stats::ServerState::Login => entry.login_count += 1,
+            crate::stats::ServerState::Active => entry.active_count += 1,
+            crate::stats::ServerState::Tested => entry.tested_count += 1,
+            crate::stats::ServerState::Idle => entry.idle_count += 1,
+        }
     }
-
+    let mut grouped_metrics: HashMap<String, Vec<PrometheusMetric<u64>>> = HashMap::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(server_info) = server_stats_by_addresses.get(&address.name()) {
+                if let Some(server_info) = prom_stats.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),
-                        ),
+                        ("bytes_received", server_info.bytes_received),
+                        ("bytes_sent", server_info.bytes_sent),
+                        ("transaction_count", server_info.transaction_count),
+                        ("query_count", server_info.query_count),
+                        ("error_count", server_info.error_count),
+                        ("idle_count", server_info.idle_count),
+                        ("active_count", server_info.active_count),
+                        ("login_count", server_info.login_count),
+                        ("tested_count", server_info.tested_count),
+                        ("is_banned", if pool.is_banned(address) { 1 } else { 0 }),
+                        ("is_paused", if pool.paused() { 1 } else { 0 }),
                     ];
                     for (key, value) in metrics {
                         if let Some(prometheus_metric) =
                             PrometheusMetric::<u64>::from_server_info(address, key, value)
                         {
-                            lines.push(prometheus_metric.to_string());
+                            grouped_metrics
+                                .entry(key.to_string())
+                                .or_default()
+                                .push(prometheus_metric);
                         } else {
                             debug!("Metric {} not implemented for {}", key, address.name());
                         }
@@ -371,17 +485,46 @@ fn push_server_stats(lines: &mut Vec<String>) {
             }
         }
     }
+    for (_key, metrics) in grouped_metrics {
+        if !metrics.is_empty() {
+            lines.push(metrics[0].get_header());
+            for metric in metrics {
+                lines.push(metric.to_string());
+            }
+        }
+    }
 }
 
 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).serve(http_service_factory);
+    let listener = TcpListener::bind(http_addr);
+    let listener = match listener.await {
+        Ok(listener) => listener,
+        Err(e) => {
+            error!("Failed to bind prometheus server to HTTP address: {}.", e);
+            return;
+        }
+    };
     info!(
         "Exposing prometheus metrics on http://{}/metrics.",
         http_addr
     );
-    if let Err(e) = server.await {
-        error!("Failed to run HTTP server: {}.", e);
+    loop {
+        let stream = match listener.accept().await {
+            Ok((stream, _)) => stream,
+            Err(e) => {
+                error!("Error accepting connection: {}", e);
+                continue;
+            }
+        };
+        let io = TokioIo::new(stream);
+
+        tokio::task::spawn(async move {
+            if let Err(err) = http1::Builder::new()
+                .serve_connection(io, service_fn(prometheus_stats))
+                .await
+            {
+                eprintln!("Error serving HTTP connection for metrics: {:?}", err);
+            }
+        });
     }
 }

src/query_router.rs

@@ -2,6 +2,7 @@
 /// or implied query characteristics.
 use bytes::{Buf, BytesMut};
 use log::{debug, error};
+use mini_moka::sync::Cache;
 use once_cell::sync::OnceCell;
 use regex::{Regex, RegexSet};
 use sqlparser::ast::Statement::{Delete, Insert, Query, StartTransaction, Update};
@@ -11,6 +12,7 @@ use sqlparser::ast::{
 };
 use sqlparser::dialect::PostgreSqlDialect;
 use sqlparser::parser::Parser;
+use std::sync::OnceLock;
 
 use crate::config::Role;
 use crate::errors::Error;
@@ -21,6 +23,7 @@ use crate::sharding::Sharder;
 
 use std::collections::BTreeSet;
 use std::io::Cursor;
+use std::time::Duration;
 use std::{cmp, mem};
 
 /// Regexes used to parse custom commands.
@@ -66,6 +69,18 @@ static CUSTOM_SQL_REGEX_SET: OnceCell<RegexSet> = OnceCell::new();
 // Get the value inside the custom command.
 static CUSTOM_SQL_REGEX_LIST: OnceCell<Vec<Regex>> = OnceCell::new();
 
+#[derive(Debug, Clone, PartialEq)]
+enum DatabaseActivityState {
+    Active,
+    Initializing,
+}
+
+// A moka cache for the databases
+// the key is the database name and the value is the database activity state
+static DATABASE_ACTIVITY_CACHE: OnceLock<Cache<String, DatabaseActivityState>> = OnceLock::new();
+// A moka cache for the tables, the key is the db_table.
+static TABLE_MUTATIONS_CACHE: OnceLock<Cache<String, bool>> = OnceLock::new();
+
 /// The query router.
 pub struct QueryRouter {
     /// Which shard we should be talking to right now.
@@ -87,6 +102,12 @@ pub struct QueryRouter {
     placeholders: Vec<i16>,
 }
 
+struct ExtractedExprsAndTables<'a> {
+    exprs: Vec<Expr>,
+    table_names: Vec<Vec<Ident>>,
+    assignments_opt: Option<&'a Vec<Assignment>>,
+}
+
 impl QueryRouter {
     /// One-time initialization of regexes
     /// that parse our custom SQL protocol.
@@ -386,6 +407,53 @@ impl QueryRouter {
         }
     }
 
+    /// Determines if a query is a mutation or not.
+    fn is_mutation_query(q: &sqlparser::ast::Query) -> bool {
+        use sqlparser::ast::*;
+
+        match q.body.as_ref() {
+            SetExpr::Insert(_) => true,
+            SetExpr::Update(_) => true,
+            SetExpr::Query(q) => Self::is_mutation_query(q),
+            _ => false,
+        }
+    }
+
+    fn database_activity_cache(&self) -> Cache<String, DatabaseActivityState> {
+        DATABASE_ACTIVITY_CACHE
+            .get_or_init(|| {
+                Cache::builder()
+                    .time_to_idle(Duration::from_secs(self.pool_settings.db_activity_ttl))
+                    .build()
+            })
+            .clone()
+    }
+
+    /// Check database activity state and reset it if necessary
+    fn database_activity_state(&self, db: &String) -> DatabaseActivityState {
+        let cache = self.database_activity_cache();
+
+        // Exists in cache
+        if cache.contains_key(db) {
+            return cache.get(db).unwrap();
+        }
+
+        // Not in cache
+        debug!("Adding database to cache: {}", db);
+
+        cache.insert(db.to_string(), DatabaseActivityState::Initializing);
+
+        // Set a timer to update the cache
+        let db = db.clone();
+        let db_activity_init_delay = self.pool_settings.db_activity_init_delay;
+        tokio::spawn(async move {
+            tokio::time::sleep(Duration::from_millis(db_activity_init_delay)).await;
+            cache.insert(db, DatabaseActivityState::Active);
+        });
+
+        DatabaseActivityState::Initializing
+    }
+
     /// Try to infer which server to connect to based on the contents of the query.
     pub fn infer(&mut self, ast: &Vec<sqlparser::ast::Statement>) -> Result<(), Error> {
         if !self.pool_settings.query_parser_read_write_splitting {
@@ -400,9 +468,23 @@ impl QueryRouter {
             return Err(Error::QueryRouterParserError("empty query".into()));
         }
 
+        let mut primary_set_based_on_activity = false;
         let mut visited_write_statement = false;
         let mut prev_inferred_shard = None;
 
+        if self.pool_settings.db_activity_based_routing {
+            let db = self.pool_settings.db.clone();
+            let state = self.database_activity_state(&db);
+            debug!("Database activity state: {:?}", state);
+
+            if let DatabaseActivityState::Initializing = state {
+                debug!("Database is initializing, going to primary");
+
+                self.active_role = Some(Role::Primary);
+                primary_set_based_on_activity = true;
+            }
+        }
+
         for q in ast {
             match q {
                 // All transactions go to the primary, probably a write.
@@ -413,6 +495,22 @@ impl QueryRouter {
 
                 // Likely a read-only query
                 Query(query) => {
+                    if primary_set_based_on_activity {
+                        // If we already set the role based on activity, we don't need to do it again
+                        continue;
+                    }
+
+                    if self.pool_settings.db_activity_based_routing {
+                        // Check if the tables in the query have been written to recently
+                        if self.query_handles_tables_in_mutation_cache(query) {
+                            debug!("Query handles tables in mutation cache, going to primary");
+
+                            self.active_role = Some(Role::Primary);
+                            primary_set_based_on_activity = true;
+                            continue;
+                        }
+                    }
+
                     match &self.pool_settings.automatic_sharding_key {
                         Some(_) => {
                             // TODO: if we have multiple queries in the same message,
@@ -427,8 +525,13 @@ impl QueryRouter {
                         None => (),
                     };
 
-                    // If we already visited a write statement, we should be going to the primary.
-                    if !visited_write_statement {
+                    let has_locks = !query.locks.is_empty();
+                    let has_mutation = Self::is_mutation_query(query);
+
+                    if has_locks || has_mutation {
+                        self.active_role = Some(Role::Primary);
+                    } else if !visited_write_statement {
+                        // If we already visited a write statement, we should be going to the primary.
                         self.active_role = match self.primary_reads_enabled() {
                             false => Some(Role::Replica), // If primary should not be receiving reads, use a replica.
                             true => None,                 // Any server role is fine in this case.
@@ -438,6 +541,13 @@ impl QueryRouter {
 
                 // Likely a write
                 _ => {
+                    debug!("Write statement found, going to primary");
+
+                    if self.pool_settings.db_activity_based_routing {
+                        // add all of the query tables to the mutation cache
+                        self.update_mutation_cache_on_write(q);
+                    }
+
                     match &self.pool_settings.automatic_sharding_key {
                         Some(_) => {
                             // TODO: similar to the above, if we have multiple queries in the
@@ -480,57 +590,69 @@ impl QueryRouter {
         Ok(())
     }
 
-    fn infer_shard_on_write(&mut self, q: &Statement) -> Result<Option<usize>, Error> {
-        let mut exprs = Vec::new();
+    fn table_mutations_cache(&self) -> Cache<String, bool> {
+        TABLE_MUTATIONS_CACHE
+            .get_or_init(|| {
+                Cache::builder()
+                    .time_to_live(Duration::from_millis(
+                        self.pool_settings.table_mutation_cache_ms_ttl,
+                    ))
+                    .build()
+            })
+            .clone()
+    }
 
-        // Collect all table names from the query.
+    fn query_handles_tables_in_mutation_cache(&self, query: &sqlparser::ast::Query) -> bool {
+        let table_mutations_cache = self.table_mutations_cache();
+        debug!("Checking if query handles tables in mutation cache");
+        debug!("Table mutations cache: {:?}", table_mutations_cache);
+
+        for tables in self.table_names(query) {
+            for table in tables {
+                if table_mutations_cache.contains_key(&self.table_mutation_cache_key(table)) {
+                    return true;
+                }
+            }
+        }
+
+        false
+    }
+    fn extract_exprs_and_table_names<'a>(
+        &'a self,
+        q: &'a Statement,
+    ) -> Option<ExtractedExprsAndTables<'a>> {
+        let mut exprs = Vec::new();
         let mut table_names = Vec::new();
+        let mut assignments_opt = None;
 
         match q {
-            Insert {
-                or,
-                into: _,
-                table_name,
-                columns,
-                overwrite: _,
-                source,
-                partitioned,
-                after_columns,
-                table: _,
-                on: _,
-                returning: _,
-            } => {
+            Insert(i) => {
                 // Not supported in postgres.
-                assert!(or.is_none());
-                assert!(partitioned.is_none());
-                assert!(after_columns.is_empty());
+                assert!(i.or.is_none());
+                assert!(i.partitioned.is_none());
+                assert!(i.after_columns.is_empty());
 
-                Self::process_table(table_name, &mut table_names);
-                Self::process_query(source, &mut exprs, &mut table_names, &Some(columns));
+                Self::process_table(&i.table_name, &mut table_names);
+                if let Some(source) = &i.source {
+                    Self::process_query(source, &mut exprs, &mut table_names, &Some(&i.columns));
+                }
             }
-            Delete {
-                tables,
-                from,
-                using,
-                selection,
-                returning: _,
-            } => {
-                if let Some(expr) = selection {
+            Delete(d) => {
+                if let Some(expr) = &d.selection {
                     exprs.push(expr.clone());
                 }
 
-                // Multi tables delete are not supported in postgres.
-                assert!(tables.is_empty());
+                // Multi-tables delete are not supported in postgres.
+                assert!(d.tables.is_empty());
 
-                Self::process_tables_with_join(from, &mut exprs, &mut table_names);
-                if let Some(using_tbl_with_join) = using {
+                if let Some(using_tbl_with_join) = &d.using {
                     Self::process_tables_with_join(
                         using_tbl_with_join,
                         &mut exprs,
                         &mut table_names,
                     );
                 }
-                Self::process_selection(selection, &mut exprs);
+                Self::process_selection(&d.selection, &mut exprs);
             }
             Update {
                 table,
@@ -544,14 +666,55 @@ impl QueryRouter {
                     Self::process_table_with_join(from_tbl, &mut exprs, &mut table_names);
                 }
                 Self::process_selection(selection, &mut exprs);
-                self.assignment_parser(assignments)?;
-            }
-            _ => {
-                return Ok(None);
+
+                assignments_opt = Some(assignments);
             }
+            _ => return None,
         };
 
-        Ok(self.infer_shard_from_exprs(exprs, table_names))
+        Some(ExtractedExprsAndTables {
+            exprs,
+            table_names,
+            assignments_opt,
+        })
+    }
+
+    fn infer_shard_on_write(&mut self, q: &Statement) -> Result<Option<usize>, Error> {
+        if let Some(extracted) = self.extract_exprs_and_table_names(q) {
+            let exprs = extracted.exprs;
+            let table_names = extracted.table_names;
+            let assignments_opt = extracted.assignments_opt;
+
+            if let Some(assignments) = assignments_opt {
+                self.assignment_parser(assignments)?;
+            }
+
+            Ok(self.infer_shard_from_exprs(exprs, table_names))
+        } else {
+            Ok(None)
+        }
+    }
+
+    fn update_mutation_cache_on_write(&self, q: &Statement) {
+        if let Some(extracted) = self.extract_exprs_and_table_names(q) {
+            debug!("Updating mutation cache on write");
+
+            let table_names = extracted.table_names;
+            debug!("Table names in mutation query: {:?}", table_names);
+            let table_mutations_cache = self.table_mutations_cache();
+            for tables in table_names {
+                for table in tables {
+                    table_mutations_cache.insert(self.table_mutation_cache_key(table), true);
+                }
+            }
+        }
+    }
+
+    // combines the database name and table name into a single string
+    // to be used as the key in the table mutation cache
+    // e.g. "mydb.mytable"
+    fn table_mutation_cache_key(&self, table: Ident) -> String {
+        format!("{}.{}", self.pool_settings.db, table.value)
     }
 
     fn process_query(
@@ -800,7 +963,13 @@ impl QueryRouter {
 
         for a in assignments {
             if sharding_key[0].value == "*"
-                && sharding_key[1].value == a.id.last().unwrap().value.to_lowercase()
+                && sharding_key[1].value
+                    == a.target
+                        .to_string()
+                        .split('.')
+                        .last()
+                        .unwrap()
+                        .to_lowercase()
             {
                 return Err(Error::QueryRouterParserError(
                     "Sharding key cannot be updated.".into(),
@@ -933,6 +1102,18 @@ impl QueryRouter {
         self.infer_shard_from_exprs(exprs, table_names)
     }
 
+    /// get table names from query
+    fn table_names(&self, query: &sqlparser::ast::Query) -> Vec<Vec<Ident>> {
+        let mut exprs = Vec::new();
+
+        let mut table_names = Vec::new();
+        Self::process_query(query, &mut exprs, &mut table_names, &None);
+
+        debug!("Table names in query: {:?}", table_names);
+
+        table_names
+    }
+
     fn infer_shard_from_exprs(
         &mut self,
         exprs: Vec<Expr>,
@@ -1039,6 +1220,11 @@ impl QueryRouter {
         self.active_shard
     }
 
+    /// Set active_role as the default_role specified in the pool.
+    pub fn set_default_role(&mut self) {
+        self.active_role = self.pool_settings.default_role;
+    }
+
     /// Get the current desired server role we should be talking to.
     pub fn role(&self) -> Option<Role> {
         self.active_role
@@ -1095,6 +1281,7 @@ mod test {
     use crate::messages::simple_query;
     use crate::sharding::ShardingFunction;
     use bytes::BufMut;
+    use serial_test::serial;
 
     #[test]
     fn test_defaults() {
@@ -1104,6 +1291,26 @@ mod test {
         assert_eq!(qr.role(), None);
     }
 
+    #[test]
+    fn test_split_cte_queries() {
+        QueryRouter::setup();
+        let mut qr = QueryRouter::new();
+        qr.pool_settings.query_parser_read_write_splitting = true;
+        qr.pool_settings.query_parser_enabled = true;
+
+        let query = simple_query(
+            "WITH t AS (
+                SELECT id FROM users WHERE name ILIKE '%ja%'
+            )
+            UPDATE user_languages
+               SET settings = '{}'
+              FROM t WHERE t.id = user_id;",
+        );
+        let ast = qr.parse(&query).unwrap();
+        assert!(qr.infer(&ast).is_ok());
+        assert_eq!(qr.role(), Some(Role::Primary));
+    }
+
     #[test]
     fn test_infer_replica() {
         QueryRouter::setup();
@@ -1153,6 +1360,29 @@ mod test {
         }
     }
 
+    #[test]
+    fn test_select_for_update() {
+        QueryRouter::setup();
+        let mut qr = QueryRouter::new();
+        qr.pool_settings.query_parser_read_write_splitting = true;
+
+        let queries_in_primary_role = vec![
+            simple_query("BEGIN"), // Transaction start
+            simple_query("SELECT * FROM items WHERE id = 5 FOR UPDATE"),
+            simple_query("UPDATE items SET name = 'pumpkin' WHERE id = 5"),
+        ];
+
+        for query in queries_in_primary_role {
+            assert!(qr.infer(&qr.parse(&query).unwrap()).is_ok());
+            assert_eq!(qr.role(), Some(Role::Primary));
+        }
+
+        // query without lock do not change role
+        let query = simple_query("SELECT * FROM items WHERE id = 5");
+        assert!(qr.infer(&qr.parse(&query).unwrap()).is_ok());
+        assert_eq!(qr.role(), None);
+    }
+
     #[test]
     fn test_infer_primary_reads_enabled() {
         QueryRouter::setup();
@@ -1367,6 +1597,19 @@ mod test {
         assert!(!qr.query_parser_enabled());
     }
 
+    #[test]
+    fn test_query_parser() {
+        QueryRouter::setup();
+        let mut qr = QueryRouter::new();
+        qr.pool_settings.query_parser_read_write_splitting = true;
+
+        let query = simple_query("SELECT req_tab_0.*  FROM validation req_tab_0  WHERE  array['http://www.w3.org/ns/shacl#ValidationResult'] && req_tab_0.type::text[] AND ( (  (req_tab_0.focusnode = 'DataSource_Credilogic_DataSourceAddress_144959227') )  )");
+        assert!(qr.infer(&qr.parse(&query).unwrap()).is_ok());
+
+        let query = simple_query("WITH EmployeeSalaries AS (SELECT Department, Salary FROM Employees) SELECT Department, AVG(Salary) AS AverageSalary FROM EmployeeSalaries GROUP BY Department;");
+        assert!(qr.infer(&qr.parse(&query).unwrap()).is_ok());
+    }
+
     #[test]
     fn test_update_from_pool_settings() {
         QueryRouter::setup();
@@ -1374,6 +1617,7 @@ mod test {
         let pool_settings = PoolSettings {
             pool_mode: PoolMode::Transaction,
             load_balancing_mode: crate::config::LoadBalancingMode::Random,
+            checkout_failure_limit: None,
             shards: 2,
             user: crate::config::User::default(),
             default_role: Some(Role::Replica),
@@ -1394,6 +1638,10 @@ mod test {
             auth_query_password: None,
             auth_query_user: None,
             db: "test".to_string(),
+            db_activity_based_routing: PoolSettings::default().db_activity_based_routing,
+            db_activity_init_delay: PoolSettings::default().db_activity_init_delay,
+            db_activity_ttl: PoolSettings::default().db_activity_ttl,
+            table_mutation_cache_ms_ttl: PoolSettings::default().table_mutation_cache_ms_ttl,
             plugins: None,
         };
         let mut qr = QueryRouter::new();
@@ -1452,6 +1700,7 @@ mod test {
         let pool_settings = PoolSettings {
             pool_mode: PoolMode::Transaction,
             load_balancing_mode: crate::config::LoadBalancingMode::Random,
+            checkout_failure_limit: Some(10),
             shards: 5,
             user: crate::config::User::default(),
             default_role: Some(Role::Replica),
@@ -1472,6 +1721,10 @@ mod test {
             auth_query_password: None,
             auth_query_user: None,
             db: "test".to_string(),
+            db_activity_based_routing: PoolSettings::default().db_activity_based_routing,
+            db_activity_init_delay: PoolSettings::default().db_activity_init_delay,
+            db_activity_ttl: PoolSettings::default().db_activity_ttl,
+            table_mutation_cache_ms_ttl: PoolSettings::default().table_mutation_cache_ms_ttl,
             plugins: None,
         };
 
@@ -1887,4 +2140,150 @@ mod test {
 
         assert_eq!(res, Ok(PluginOutput::Allow));
     }
+
+    #[tokio::test]
+    #[serial]
+    async fn test_db_activity_based_routing_initializing_state() {
+        QueryRouter::setup();
+        let mut qr = QueryRouter::new();
+        qr.pool_settings.db_activity_based_routing = true;
+        qr.pool_settings.query_parser_read_write_splitting = true;
+        qr.pool_settings.query_parser_enabled = true;
+        qr.pool_settings.db = "test_table_mutation_cache".to_string();
+
+        qr.database_activity_cache()
+            .invalidate(&qr.pool_settings.db.clone());
+
+        let query = simple_query("SELECT * FROM some_table");
+        let ast = qr.parse(&query).unwrap();
+
+        // Initially, the database activity should be in the "Initializing" state
+        let state = qr.database_activity_state(&qr.pool_settings.db.clone());
+        assert_eq!(state, DatabaseActivityState::Initializing);
+
+        // Check that the router chooses the primary role due to "Initializing" state
+        assert!(qr.infer(&ast).is_ok());
+        assert_eq!(qr.role(), Some(Role::Primary));
+    }
+
+    #[tokio::test]
+    #[serial]
+    async fn test_db_activity_based_routing_active_state() {
+        QueryRouter::setup();
+        let mut qr = QueryRouter::new();
+        qr.pool_settings.db_activity_based_routing = true;
+        qr.pool_settings.query_parser_read_write_splitting = true;
+        qr.pool_settings.query_parser_enabled = true;
+        qr.pool_settings.db = "test_table_mutation_cache".to_string();
+
+        let db_name = qr.pool_settings.db.clone();
+        let cache = qr.database_activity_cache();
+        cache.insert(db_name.clone(), DatabaseActivityState::Active);
+
+        let query = simple_query("SELECT * FROM some_table");
+        let ast = qr.parse(&query).unwrap();
+
+        // Check that the router can choose a replica role when in "Active" state
+        assert!(qr.infer(&ast).is_ok());
+        assert_eq!(qr.role(), None); // Default should allow replica due to active state
+    }
+
+    #[tokio::test]
+    #[serial]
+    async fn test_table_mutation_cache_on_write() {
+        QueryRouter::setup();
+        let mut qr = QueryRouter::new();
+        qr.pool_settings.db_activity_based_routing = true;
+        qr.pool_settings.table_mutation_cache_ms_ttl = 20_000; // 20 seconds in milliseconds
+        qr.pool_settings.query_parser_enabled = true;
+        qr.pool_settings.query_parser_read_write_splitting = true;
+        qr.pool_settings.db = "test_table_mutation_cache".to_string();
+
+        qr.database_activity_cache()
+            .invalidate(&qr.pool_settings.db.clone());
+
+        let query = simple_query("UPDATE some_table SET col1 = 'value' WHERE col2 = 1");
+        let ast = qr.parse(&query).unwrap();
+
+        // Simulate the mutation query which should populate the mutation cache
+        assert!(qr.infer(&ast).is_ok());
+        assert_eq!(qr.role(), Some(Role::Primary));
+
+        let table_cache_key = qr.table_mutation_cache_key(Ident::new("some_table"));
+        let cache = qr.table_mutations_cache();
+
+        // Ensure the table mutation cache contains the table with recent write
+        assert!(cache.contains_key(&table_cache_key));
+    }
+
+    #[tokio::test]
+    #[serial]
+    async fn test_db_activity_based_routing_multi_query() {
+        use super::*;
+        use crate::messages::simple_query;
+        use tokio::time::Duration;
+
+        QueryRouter::setup();
+        let mut qr = QueryRouter::new();
+
+        // Configure the pool settings for db_activity_based_routing
+        qr.pool_settings.query_parser_read_write_splitting = true;
+        qr.pool_settings.query_parser_enabled = true;
+        qr.pool_settings.db_activity_based_routing = true;
+        qr.pool_settings.db = "test_db_activity_routing".to_string();
+
+        qr.database_activity_cache()
+            .invalidate(&qr.pool_settings.db.clone());
+
+        // First query when database is initializing
+        let query = simple_query("SELECT * FROM test_table");
+        let ast = qr.parse(&query).unwrap();
+        assert!(qr.infer(&ast).is_ok());
+        // Should route to primary because database is initializing
+        assert_eq!(qr.role(), Some(Role::Primary));
+
+        // Wait for the initialization delay to pass
+        tokio::time::sleep(Duration::from_millis(
+            qr.pool_settings.db_activity_init_delay * 2,
+        ))
+        .await;
+
+        // Next query after database is active
+        let query = simple_query("SELECT * FROM test_table");
+        let ast = qr.parse(&query).unwrap();
+        qr.active_role = None; // Reset the active_role
+        assert!(qr.infer(&ast).is_ok());
+        // Should route to replica because database is active and no recent mutations
+        assert_eq!(qr.role(), None);
+
+        // Simulate a write query to update the mutation cache
+        let query = simple_query("INSERT INTO test_table (id, name) VALUES (1, 'test')");
+        let ast = qr.parse(&query).unwrap();
+        qr.active_role = None; // Reset the active_role
+        assert!(qr.infer(&ast).is_ok());
+        // Should route to primary because it's a write operation
+        assert_eq!(qr.role(), Some(Role::Primary));
+
+        // Immediately run a read query on the same table
+        let query = simple_query("SELECT * FROM test_table WHERE id = 1");
+        let ast = qr.parse(&query).unwrap();
+        qr.active_role = None; // Reset the active_role
+        assert!(qr.infer(&ast).is_ok());
+        // Should route to primary because the table was recently mutated
+        assert_eq!(qr.role(), Some(Role::Primary));
+
+        // Wait for the mutation cache TTL to expire
+        tokio::time::sleep(Duration::from_millis(
+            qr.pool_settings.table_mutation_cache_ms_ttl * 2,
+        ))
+        .await;
+
+        // Run the read query again after cache expiration
+        let query = simple_query("SELECT * FROM test_table WHERE id = 1");
+        let ast = qr.parse(&query).unwrap();
+        qr.active_role = None; // Reset the active_role
+        assert!(qr.infer(&ast).is_ok());
+        // Should route to replica because mutation cache has expired
+        assert_eq!(qr.role(), None);
+    }
 }

src/server.rs

@@ -7,7 +7,7 @@ use lru::LruCache;
 use once_cell::sync::Lazy;
 use parking_lot::{Mutex, RwLock};
 use postgres_protocol::message;
-use std::collections::{HashMap, HashSet};
+use std::collections::{HashMap, HashSet, VecDeque};
 use std::mem;
 use std::net::IpAddr;
 use std::num::NonZeroUsize;
@@ -325,6 +325,9 @@ pub struct Server {
 
     /// Prepared statements
     prepared_statement_cache: Option<LruCache<String, ()>>,
+
+    /// Prepared statement being currently registered on the server.
+    registering_prepared_statement: VecDeque<String>,
 }
 
 impl Server {
@@ -827,6 +830,7 @@ impl Server {
                                 NonZeroUsize::new(prepared_statement_cache_size).unwrap(),
                             )),
                         },
+                        registering_prepared_statement: VecDeque::new(),
                     };
 
                     return Ok(server);
@@ -956,7 +960,6 @@ impl Server {
 
                     // There is no more data available from the server.
                     self.data_available = false;
-
                     break;
                 }
 
@@ -966,6 +969,23 @@ impl Server {
                         self.in_copy_mode = false;
                     }
 
+                    // Remove the prepared statement from the cache, it has a syntax error or something else bad happened.
+                    if let Some(prepared_stmt_name) =
+                        self.registering_prepared_statement.pop_front()
+                    {
+                        if let Some(ref mut cache) = self.prepared_statement_cache {
+                            if let Some(_removed) = cache.pop(&prepared_stmt_name) {
+                                debug!(
+                                    "Removed {} from prepared statement cache",
+                                    prepared_stmt_name
+                                );
+                            } else {
+                                // Shouldn't happen.
+                                debug!("Prepared statement {} was not cached", prepared_stmt_name);
+                            }
+                        }
+                    }
+
                     if self.prepared_statement_cache.is_some() {
                         let error_message = PgErrorMsg::parse(&message)?;
                         if error_message.message == "cached plan must not change result type" {
@@ -1068,6 +1088,11 @@ impl Server {
                 // Buffer until ReadyForQuery shows up, so don't exit the loop yet.
                 'c' => (),
 
+                // Parse complete successfully
+                '1' => {
+                    self.registering_prepared_statement.pop_front();
+                }
+
                 // Anything else, e.g. errors, notices, etc.
                 // Keep buffering until ReadyForQuery shows up.
                 _ => (),
@@ -1107,7 +1132,7 @@ impl Server {
         has_it
     }
 
-    pub fn add_prepared_statement_to_cache(&mut self, name: &str) -> Option<String> {
+    fn add_prepared_statement_to_cache(&mut self, name: &str) -> Option<String> {
         let cache = match &mut self.prepared_statement_cache {
             Some(cache) => cache,
             None => return None,
@@ -1129,7 +1154,7 @@ impl Server {
         None
     }
 
-    pub fn remove_prepared_statement_from_cache(&mut self, name: &str) {
+    fn remove_prepared_statement_from_cache(&mut self, name: &str) {
         let cache = match &mut self.prepared_statement_cache {
             Some(cache) => cache,
             None => return,
@@ -1145,6 +1170,9 @@ impl Server {
         should_send_parse_to_server: bool,
     ) -> Result<(), Error> {
         if !self.has_prepared_statement(&parse.name) {
+            self.registering_prepared_statement
+                .push_back(parse.name.clone());
+
             let mut bytes = BytesMut::new();
 
             if should_send_parse_to_server {
@@ -1176,7 +1204,13 @@ impl Server {
             }
         };
 
-        Ok(())
+        // If it's not there, something went bad, I'm guessing bad syntax or permissions error
+        // on the server.
+        if !self.has_prepared_statement(&parse.name) {
+            Err(Error::PreparedStatementError)
+        } else {
+            Ok(())
+        }
     }
 
     /// If the server is still inside a transaction.
@@ -1186,6 +1220,7 @@ impl Server {
         self.in_transaction
     }
 
+    /// Currently copying data from client to server or vice-versa.
     pub fn in_copy_mode(&self) -> bool {
         self.in_copy_mode
     }
@@ -1244,8 +1279,8 @@ impl Server {
     }
 
     /// Indicate that this server connection cannot be re-used and must be discarded.
-    pub fn mark_bad(&mut self) {
-        error!("Server {:?} marked bad", self.address);
+    pub fn mark_bad(&mut self, reason: &str) {
+        error!("Server {:?} marked bad, reason: {}", self.address, reason);
         self.bad = true;
     }
 

src/sharding.rs

@@ -14,11 +14,11 @@ pub enum ShardingFunction {
     Sha1,
 }
 
-impl ToString for ShardingFunction {
-    fn to_string(&self) -> String {
-        match *self {
-            ShardingFunction::PgBigintHash => "pg_bigint_hash".to_string(),
-            ShardingFunction::Sha1 => "sha1".to_string(),
+impl std::fmt::Display for ShardingFunction {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        match self {
+            ShardingFunction::PgBigintHash => write!(f, "pg_bigint_hash"),
+            ShardingFunction::Sha1 => write!(f, "sha1"),
         }
     }
 }

src/stats/client.rs

@@ -41,6 +41,11 @@ pub struct ClientStats {
     /// Maximum time spent waiting for a connection from pool, measures in microseconds
     pub max_wait_time: Arc<AtomicU64>,
 
+    // Time when the client started waiting for a connection from pool, measures in microseconds
+    // We use connect_time as the reference point for this value
+    // U64 can represent ~5850 centuries in microseconds, so we should be fine
+    pub wait_start_us: Arc<AtomicU64>,
+
     /// Current state of the client
     pub state: Arc<AtomicClientState>,
 
@@ -64,6 +69,7 @@ impl Default for ClientStats {
             pool_name: String::new(),
             total_wait_time: Arc::new(AtomicU64::new(0)),
             max_wait_time: Arc::new(AtomicU64::new(0)),
+            wait_start_us: 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)),
@@ -111,6 +117,9 @@ impl ClientStats {
 
     /// Reports a client is waiting for a connection
     pub fn waiting(&self) {
+        let wait_start = self.connect_time.elapsed().as_micros() as u64;
+
+        self.wait_start_us.store(wait_start, Ordering::Relaxed);
         self.state.store(ClientState::Waiting, Ordering::Relaxed);
     }
 
@@ -122,6 +131,13 @@ impl ClientStats {
     /// 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);
+        self.update_wait_times();
+    }
+
+    /// Reports a client has succeeded in obtaining a connection from a connection pool
+    pub fn checkout_success(&self) {
+        self.state.store(ClientState::Active, Ordering::Relaxed);
+        self.update_wait_times();
     }
 
     /// Reports a client has had the server assigned to it be banned
@@ -130,12 +146,26 @@ impl ClientStats {
         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) {
+    fn update_wait_times(&self) {
+        if self.wait_start_us.load(Ordering::Relaxed) == 0 {
+            return;
+        }
+
+        let wait_time_us = self.get_current_wait_time_us();
         self.total_wait_time
-            .fetch_add(microseconds, Ordering::Relaxed);
+            .fetch_add(wait_time_us, Ordering::Relaxed);
         self.max_wait_time
-            .fetch_max(microseconds, Ordering::Relaxed);
+            .fetch_max(wait_time_us, Ordering::Relaxed);
+        self.wait_start_us.store(0, Ordering::Relaxed);
+    }
+
+    pub fn get_current_wait_time_us(&self) -> u64 {
+        let wait_start_us = self.wait_start_us.load(Ordering::Relaxed);
+        let microseconds_since_connection_epoch = self.connect_time.elapsed().as_micros() as u64;
+        if wait_start_us == 0 || microseconds_since_connection_epoch < wait_start_us {
+            return 0;
+        }
+        microseconds_since_connection_epoch - wait_start_us
     }
 
     /// Report a query executed by a client against a server

src/stats/pool.rs

@@ -64,8 +64,11 @@ impl PoolStats {
                         ClientState::Idle => pool_stats.cl_idle += 1,
                         ClientState::Waiting => pool_stats.cl_waiting += 1,
                     }
-                    let max_wait = client.max_wait_time.load(Ordering::Relaxed);
-                    pool_stats.maxwait = std::cmp::max(pool_stats.maxwait, max_wait);
+                    let wait_start_us = client.wait_start_us.load(Ordering::Relaxed);
+                    if wait_start_us > 0 {
+                        let wait_time_us = client.get_current_wait_time_us();
+                        pool_stats.maxwait = std::cmp::max(pool_stats.maxwait, wait_time_us);
+                    }
                 }
                 None => debug!("Client from an obselete pool"),
             }

start_test_env.sh

@@ -0,0 +1,34 @@
+GREEN="\033[0;32m"
+RED="\033[0;31m"
+BLUE="\033[0;34m"
+RESET="\033[0m"
+
+
+cd tests/docker/
+docker compose kill main || true
+docker compose build main
+docker compose down
+docker compose up -d
+# wait for the container to start
+while ! docker compose exec main ls; do
+    echo "Waiting for test environment to start"
+    sleep 1
+done
+echo "==================================="
+docker compose exec -e LOG_LEVEL=error -d main toxiproxy-server
+docker compose exec --workdir /app main cargo build
+docker compose exec -d --workdir /app main ./target/debug/pgcat ./.circleci/pgcat.toml
+docker compose exec --workdir /app/tests/ruby main bundle install
+docker compose exec --workdir /app/tests/python main pip3 install -r requirements.txt
+echo "Interactive test environment ready"
+echo "To run integration tests, you can use the following commands:"
+echo -e "   ${BLUE}Ruby:   ${RED}cd /app/tests/ruby && bundle exec ruby tests.rb --format documentation${RESET}"
+echo -e "   ${BLUE}Python: ${RED}cd /app/ && pytest ${RESET}"
+echo -e "   ${BLUE}Rust:   ${RED}cd /app/tests/rust && cargo run ${RESET}"
+echo -e "   ${BLUE}Go:     ${RED}cd /app/tests/go && /usr/local/go/bin/go test${RESET}"
+echo "the source code for tests are directly linked to the source code in the container so you can modify the code and run the tests again"
+echo "You can rebuild PgCat from within the container by running"
+echo -e "  ${GREEN}cargo build${RESET}"
+echo "and then run the tests again"
+echo "==================================="
+docker compose exec --workdir /app/tests main bash

tests/docker/Dockerfile

@@ -1,4 +1,4 @@
-FROM rust:bullseye
+FROM rust:1.81.0-slim-bookworm 
 
 COPY --from=sclevine/yj /bin/yj /bin/yj
 RUN /bin/yj -h
@@ -8,3 +8,6 @@ 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_$(dpkg --print-architecture).deb && \
 	sudo dpkg -i toxiproxy-2.4.0.deb
+RUN wget -O go1.21.3.linux-$(dpkg --print-architecture).tar.gz https://go.dev/dl/go1.21.3.linux-$(dpkg --print-architecture).tar.gz && \
+    sudo tar -C /usr/local -xzf go1.21.3.linux-$(dpkg --print-architecture).tar.gz && \
+    rm go1.21.3.linux-$(dpkg --print-architecture).tar.gz

tests/docker/docker-compose.yml

@@ -1,4 +1,3 @@
-version: "3"
 services:
   pg1:
     image: postgres:14
@@ -48,6 +47,8 @@ services:
   main:
     build: .
     command: ["bash", "/app/tests/docker/run.sh"]
+    environment:
+      - INTERACTIVE_TEST_ENVIRONMENT=true
     volumes:
       - ../../:/app/
       - /app/target/

tests/docker/run.sh

@@ -5,6 +5,38 @@ rm /app/*.profraw || true
 rm /app/pgcat.profdata || true
 rm -rf /app/cov || true
 
+# Prepares the interactive test environment
+# 
+if [ -n "$INTERACTIVE_TEST_ENVIRONMENT" ]; then
+    ports=(5432 7432 8432 9432 10432)
+    for port in "${ports[@]}"; do
+        is_it_up=0
+        attempts=0
+        while [ $is_it_up -eq 0 ]; do
+            PGPASSWORD=postgres psql -h 127.0.0.1 -p $port -U postgres -c '\q' > /dev/null 2>&1
+            if [ $? -eq 0 ]; then
+                echo "PostgreSQL on port $port is up."
+                is_it_up=1
+            else
+                attempts=$((attempts+1))
+                if [ $attempts -gt 10 ]; then
+                    echo "PostgreSQL on port $port is down, giving up."
+                    exit 1
+                fi
+                echo "PostgreSQL on port $port is down, waiting for it to start."
+                sleep 1
+            fi
+        done
+    done
+    PGPASSWORD=postgres psql -e -h 127.0.0.1 -p 5432  -U postgres -f /app/tests/sharding/query_routing_setup.sql
+    PGPASSWORD=postgres psql -e -h 127.0.0.1 -p 7432  -U postgres -f /app/tests/sharding/query_routing_setup.sql
+    PGPASSWORD=postgres psql -e -h 127.0.0.1 -p 8432  -U postgres -f /app/tests/sharding/query_routing_setup.sql
+    PGPASSWORD=postgres psql -e -h 127.0.0.1 -p 9432  -U postgres -f /app/tests/sharding/query_routing_setup.sql
+    PGPASSWORD=postgres psql -e -h 127.0.0.1 -p 10432 -U postgres -f /app/tests/sharding/query_routing_setup.sql
+    sleep 100000000000000000
+    exit 0
+fi
+
 export LLVM_PROFILE_FILE="/app/pgcat-%m-%p.profraw"
 export RUSTC_BOOTSTRAP=1
 export CARGO_INCREMENTAL=0

tests/go/go.mod

@@ -0,0 +1,5 @@
+module pgcat
+
+go 1.21
+
+require github.com/lib/pq v1.10.9

tests/go/go.sum

@@ -0,0 +1,2 @@
+github.com/lib/pq v1.10.9 h1:YXG7RB+JIjhP29X+OtkiDnYaXQwpS4JEWq7dtCCRUEw=
+github.com/lib/pq v1.10.9/go.mod h1:AlVN5x4E4T544tWzH6hKfbfQvm3HdbOxrmggDNAPY9o=

tests/go/pgcat.toml

@@ -0,0 +1,162 @@
+#
+# PgCat config example.
+#
+
+#
+# General pooler settings
+[general]
+# What IP to run on, 0.0.0.0 means accessible from everywhere.
+host = "0.0.0.0"
+
+# Port to run on, same as PgBouncer used in this example.
+port = "${PORT}"
+
+# Whether to enable prometheus exporter or not.
+enable_prometheus_exporter = true
+
+# Port at which prometheus exporter listens on.
+prometheus_exporter_port = 9930
+
+# How long to wait before aborting a server connection (ms).
+connect_timeout = 1000
+
+# How much time to give the health check query to return with a result (ms).
+healthcheck_timeout = 1000
+
+# How long to keep connection available for immediate re-use, without running a healthcheck query on it
+healthcheck_delay = 30000
+
+# How much time to give clients during shutdown before forcibly killing client connections (ms).
+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 = 15000
+
+server_round_robin = false
+
+# TLS
+tls_certificate = "../../.circleci/server.cert"
+tls_private_key = "../../.circleci/server.key"
+
+# Credentials 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"
+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"
+[pools.sharded_db]
+# Pool mode (see PgBouncer docs for more).
+# 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.
+default_role = "any"
+
+# Query parser. If 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.
+query_parser_enabled = true
+
+# If the query parser is enabled and this setting is enabled, we'll attempt to
+# infer the role from the query itself.
+query_parser_read_write_splitting = 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.
+primary_reads_enabled = true
+
+# 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
+#
+sharding_function = "pg_bigint_hash"
+
+# Prepared statements cache size.
+prepared_statements_cache_size = 500
+
+# Credentials for users that may connect to this cluster
+[pools.sharded_db.users.0]
+username = "sharding_user"
+password = "sharding_user"
+# 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 = 5
+statement_timeout = 0
+
+
+[pools.sharded_db.users.1]
+username = "other_user"
+password = "other_user"
+pool_size = 21
+statement_timeout = 30000
+
+# Shard 0
+[pools.sharded_db.shards.0]
+# [ host, port, role ]
+servers = [
+    [ "127.0.0.1", 5432, "primary" ],
+    [ "localhost", 5432, "replica" ]
+]
+# Database name (e.g. "postgres")
+database = "shard0"
+
+[pools.sharded_db.shards.1]
+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" ],
+]
+database = "shard2"
+
+
+[pools.simple_db]
+pool_mode = "session"
+default_role = "primary"
+query_parser_enabled = true
+query_parser_read_write_splitting = true
+primary_reads_enabled = true
+sharding_function = "pg_bigint_hash"
+
+[pools.simple_db.users.0]
+username = "simple_user"
+password = "simple_user"
+pool_size = 5
+statement_timeout = 30000
+
+[pools.simple_db.shards.0]
+servers = [
+    [ "127.0.0.1", 5432, "primary" ],
+    [ "localhost", 5432, "replica" ]
+]
+database = "some_db"

tests/go/prepared_test.go

@@ -0,0 +1,52 @@
+package pgcat
+
+import (
+	"context"
+	"database/sql"
+	"fmt"
+	_ "github.com/lib/pq"
+	"testing"
+)
+
+func Test(t *testing.T) {
+	t.Cleanup(setup(t))
+	t.Run("Named parameterized prepared statement works", namedParameterizedPreparedStatement)
+	t.Run("Unnamed parameterized prepared statement works", unnamedParameterizedPreparedStatement)
+}
+
+func namedParameterizedPreparedStatement(t *testing.T) {
+	db, err := sql.Open("postgres", fmt.Sprintf("host=localhost port=%d database=sharded_db user=sharding_user password=sharding_user sslmode=disable", port))
+	if err != nil {
+		t.Fatalf("could not open connection: %+v", err)
+	}
+
+	stmt, err := db.Prepare("SELECT $1")
+
+	if err != nil {
+		t.Fatalf("could not prepare: %+v", err)
+	}
+
+	for i := 0; i < 100; i++ {
+		rows, err := stmt.Query(1)
+		if err != nil {
+			t.Fatalf("could not query: %+v", err)
+		}
+		_ = rows.Close()
+	}
+}
+
+func unnamedParameterizedPreparedStatement(t *testing.T) {
+	db, err := sql.Open("postgres", fmt.Sprintf("host=localhost port=%d database=sharded_db user=sharding_user password=sharding_user sslmode=disable", port))
+	if err != nil {
+		t.Fatalf("could not open connection: %+v", err)
+	}
+
+	for i := 0; i < 100; i++ {
+		// Under the hood QueryContext generates an unnamed parameterized prepared statement
+		rows, err := db.QueryContext(context.Background(), "SELECT $1", 1)
+		if err != nil {
+			t.Fatalf("could not query: %+v", err)
+		}
+		_ = rows.Close()
+	}
+}

tests/go/setup.go

@@ -0,0 +1,81 @@
+package pgcat
+
+import (
+	"context"
+	"database/sql"
+	_ "embed"
+	"fmt"
+	"math/rand"
+	"os"
+	"os/exec"
+	"strings"
+	"testing"
+	"time"
+)
+
+//go:embed pgcat.toml
+var pgcatCfg string
+
+var port = rand.Intn(32760-20000) + 20000
+
+func setup(t *testing.T) func() {
+	cfg, err := os.CreateTemp("/tmp", "pgcat_cfg_*.toml")
+	if err != nil {
+		t.Fatalf("could not create temp file: %+v", err)
+	}
+
+	pgcatCfg = strings.Replace(pgcatCfg, "\"${PORT}\"", fmt.Sprintf("%d", port), 1)
+
+	_, err = cfg.Write([]byte(pgcatCfg))
+	if err != nil {
+		t.Fatalf("could not write temp file: %+v", err)
+	}
+
+	commandPath := "../../target/debug/pgcat"
+	if os.Getenv("CARGO_TARGET_DIR") != "" {
+		commandPath = os.Getenv("CARGO_TARGET_DIR") + "/debug/pgcat"
+	}
+
+	cmd := exec.Command(commandPath, cfg.Name())
+	cmd.Stdout = os.Stdout
+	cmd.Stderr = os.Stderr
+	go func() {
+		err = cmd.Run()
+		if err != nil {
+			t.Errorf("could not run pgcat: %+v", err)
+		}
+	}()
+
+	deadline, cancelFunc := context.WithDeadline(context.Background(), time.Now().Add(5*time.Second))
+	defer cancelFunc()
+	for {
+		select {
+		case <-deadline.Done():
+			break
+		case <-time.After(50 * time.Millisecond):
+			db, err := sql.Open("postgres", fmt.Sprintf("host=localhost port=%d database=pgcat user=admin_user password=admin_pass sslmode=disable", port))
+			if err != nil {
+				continue
+			}
+			rows, err := db.QueryContext(deadline, "SHOW STATS")
+			if err != nil {
+				continue
+			}
+			_ = rows.Close()
+			_ = db.Close()
+			break
+		}
+		break
+	}
+
+	return func() {
+		err := cmd.Process.Signal(os.Interrupt)
+		if err != nil {
+			t.Fatalf("could not interrupt pgcat: %+v", err)
+		}
+		err = os.Remove(cfg.Name())
+		if err != nil {
+			t.Fatalf("could not remove temp file: %+v", err)
+		}
+	}
+}

tests/python/requirements.txt

@@ -1,2 +1,3 @@
+pytest
 psycopg2==2.9.3
-psutil==5.9.1
+psutil==5.9.1

tests/python/test_auth.py

@@ -0,0 +1,71 @@
+import utils
+import signal
+
+class TestTrustAuth:
+    @classmethod
+    def setup_method(cls):
+        config= """
+        [general]
+        host = "0.0.0.0"
+        port = 6432
+        admin_username = "admin_user"
+        admin_password = ""
+        admin_auth_type = "trust"
+
+        [pools.sharded_db.users.0]
+        username = "sharding_user"
+        password = "sharding_user"
+        auth_type = "trust"
+        pool_size = 10
+        min_pool_size = 1
+        pool_mode = "transaction"
+
+        [pools.sharded_db.shards.0]
+        servers = [
+          [ "127.0.0.1", 5432, "primary" ],
+        ]
+        database = "shard0"
+        """
+        utils.pgcat_generic_start(config)
+
+    @classmethod
+    def teardown_method(self):
+        utils.pg_cat_send_signal(signal.SIGTERM)
+
+    def test_admin_trust_auth(self):
+        conn, cur = utils.connect_db_trust(admin=True)
+        cur.execute("SHOW POOLS")
+        res = cur.fetchall()
+        print(res)
+        utils.cleanup_conn(conn, cur)
+
+    def test_normal_trust_auth(self):
+        conn, cur = utils.connect_db_trust(autocommit=False)
+        cur.execute("SELECT 1")
+        res = cur.fetchall()
+        print(res)
+        utils.cleanup_conn(conn, cur)
+
+class TestMD5Auth:
+    @classmethod
+    def setup_method(cls):
+        utils.pgcat_start()
+
+    @classmethod
+    def teardown_method(self):
+        utils.pg_cat_send_signal(signal.SIGTERM)
+
+    def test_normal_db_access(self):
+        conn, cur = utils.connect_db(autocommit=False)
+        cur.execute("SELECT 1")
+        res = cur.fetchall()
+        print(res)
+        utils.cleanup_conn(conn, cur)
+
+    def test_admin_db_access(self):
+        conn, cur = utils.connect_db(admin=True)
+
+        cur.execute("SHOW POOLS")
+        res = cur.fetchall()
+        print(res)
+        utils.cleanup_conn(conn, cur)

tests/python/test_pgcat.py

@@ -1,84 +1,12 @@
-from typing import Tuple
-import psycopg2
-import psutil
-import os
+
 import signal
 import time
 
+import psycopg2
+import utils
+
 SHUTDOWN_TIMEOUT = 5
 
-PGCAT_HOST = "127.0.0.1"
-PGCAT_PORT = "6432"
-
-
-def pgcat_start():
-    pg_cat_send_signal(signal.SIGTERM)
-    os.system("./target/debug/pgcat .circleci/pgcat.toml &")
-    time.sleep(2)
-
-
-def pg_cat_send_signal(signal: signal.Signals):
-    try:
-        for proc in psutil.process_iter(["pid", "name"]):
-            if "pgcat" == proc.name():
-                os.kill(proc.pid, signal)
-    except Exception as e:
-        # The process can be gone when we send this signal
-        print(e)
-
-    if signal == signal.SIGTERM:
-        # Returns 0 if pgcat process exists
-        time.sleep(2)
-        if not os.system('pgrep pgcat'):
-            raise Exception("pgcat not closed after SIGTERM")
-
-
-def connect_db(
-    autocommit: bool = True,
-    admin: bool = False,
-) -> Tuple[psycopg2.extensions.connection, psycopg2.extensions.cursor]:
-
-    if admin:
-        user = "admin_user"
-        password = "admin_pass"
-        db = "pgcat"
-    else:
-        user = "sharding_user"
-        password = "sharding_user"
-        db = "sharded_db"
-
-    conn = psycopg2.connect(
-        f"postgres://{user}:{password}@{PGCAT_HOST}:{PGCAT_PORT}/{db}?application_name=testing_pgcat",
-        connect_timeout=2,
-    )
-    conn.autocommit = autocommit
-    cur = conn.cursor()
-
-    return (conn, cur)
-
-
-def cleanup_conn(conn: psycopg2.extensions.connection, cur: psycopg2.extensions.cursor):
-    cur.close()
-    conn.close()
-
-
-def test_normal_db_access():
-    pgcat_start()
-    conn, cur = connect_db(autocommit=False)
-    cur.execute("SELECT 1")
-    res = cur.fetchall()
-    print(res)
-    cleanup_conn(conn, cur)
-
-
-def test_admin_db_access():
-    conn, cur = connect_db(admin=True)
-
-    cur.execute("SHOW POOLS")
-    res = cur.fetchall()
-    print(res)
-    cleanup_conn(conn, cur)
-
 
 def test_shutdown_logic():
 
@@ -86,17 +14,17 @@ def test_shutdown_logic():
     # NO ACTIVE QUERIES SIGINT HANDLING
 
     # Start pgcat
-    pgcat_start()
+    utils.pgcat_start()
 
     # Create client connection and send query (not in transaction)
-    conn, cur = connect_db()
+    conn, cur = utils.connect_db()
 
     cur.execute("BEGIN;")
     cur.execute("SELECT 1;")
     cur.execute("COMMIT;")
 
     # Send sigint to pgcat
-    pg_cat_send_signal(signal.SIGINT)
+    utils.pg_cat_send_signal(signal.SIGINT)
     time.sleep(1)
 
     # Check that any new queries fail after sigint since server should close with no active transactions
@@ -108,18 +36,18 @@ def test_shutdown_logic():
         # Fail if query execution succeeded
         raise Exception("Server not closed after sigint")
 
-    cleanup_conn(conn, cur)
-    pg_cat_send_signal(signal.SIGTERM)
+    utils.cleanup_conn(conn, cur)
+    utils.pg_cat_send_signal(signal.SIGTERM)
 
     # - - - - - - - - - - - - - - - - - -
     # NO ACTIVE QUERIES ADMIN SHUTDOWN COMMAND
 
     # Start pgcat
-    pgcat_start()
+    utils.pgcat_start()
 
     # Create client connection and begin transaction
-    conn, cur = connect_db()
-    admin_conn, admin_cur = connect_db(admin=True)
+    conn, cur = utils.connect_db()
+    admin_conn, admin_cur = utils.connect_db(admin=True)
 
     cur.execute("BEGIN;")
     cur.execute("SELECT 1;")
@@ -138,24 +66,24 @@ def test_shutdown_logic():
         # 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)
+    utils.cleanup_conn(conn, cur)
+    utils.cleanup_conn(admin_conn, admin_cur)
+    utils.pg_cat_send_signal(signal.SIGTERM)
 
     # - - - - - - - - - - - - - - - - - -
     # HANDLE TRANSACTION WITH SIGINT
 
     # Start pgcat
-    pgcat_start()
+    utils.pgcat_start()
 
     # Create client connection and begin transaction
-    conn, cur = connect_db()
+    conn, cur = utils.connect_db()
 
     cur.execute("BEGIN;")
     cur.execute("SELECT 1;")
 
     # Send sigint to pgcat while still in transaction
-    pg_cat_send_signal(signal.SIGINT)
+    utils.pg_cat_send_signal(signal.SIGINT)
     time.sleep(1)
 
     # Check that any new queries succeed after sigint since server should still allow transaction to complete
@@ -165,18 +93,18 @@ def test_shutdown_logic():
         # Fail if query fails since server closed
         raise Exception("Server closed while in transaction", e.pgerror)
 
-    cleanup_conn(conn, cur)
-    pg_cat_send_signal(signal.SIGTERM)
+    utils.cleanup_conn(conn, cur)
+    utils.pg_cat_send_signal(signal.SIGTERM)
 
     # - - - - - - - - - - - - - - - - - -
     # HANDLE TRANSACTION WITH ADMIN SHUTDOWN COMMAND
 
     # Start pgcat
-    pgcat_start()
+    utils.pgcat_start()
 
     # Create client connection and begin transaction
-    conn, cur = connect_db()
-    admin_conn, admin_cur = connect_db(admin=True)
+    conn, cur = utils.connect_db()
+    admin_conn, admin_cur = utils.connect_db(admin=True)
 
     cur.execute("BEGIN;")
     cur.execute("SELECT 1;")
@@ -194,30 +122,30 @@ def test_shutdown_logic():
         # 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)
+    utils.cleanup_conn(conn, cur)
+    utils.cleanup_conn(admin_conn, admin_cur)
+    utils.pg_cat_send_signal(signal.SIGTERM)
 
     # - - - - - - - - - - - - - - - - - -
     # NO NEW NON-ADMIN CONNECTIONS DURING SHUTDOWN
     # Start pgcat
-    pgcat_start()
+    utils.pgcat_start()
 
     # Create client connection and begin transaction
-    transaction_conn, transaction_cur = connect_db()
+    transaction_conn, transaction_cur = utils.connect_db()
 
     transaction_cur.execute("BEGIN;")
     transaction_cur.execute("SELECT 1;")
 
     # Send sigint to pgcat while still in transaction
-    pg_cat_send_signal(signal.SIGINT)
+    utils.pg_cat_send_signal(signal.SIGINT)
     time.sleep(1)
 
     start = time.perf_counter()
     try:
-        conn, cur = connect_db()
+        conn, cur = utils.connect_db()
         cur.execute("SELECT 1;")
-        cleanup_conn(conn, cur)
+        utils.cleanup_conn(conn, cur)
     except psycopg2.OperationalError as e:
         time_taken = time.perf_counter() - start
         if time_taken > 0.1:
@@ -227,49 +155,49 @@ def test_shutdown_logic():
     else:
         raise Exception("Able connect to database during shutdown")
 
-    cleanup_conn(transaction_conn, transaction_cur)
-    pg_cat_send_signal(signal.SIGTERM)
+    utils.cleanup_conn(transaction_conn, transaction_cur)
+    utils.pg_cat_send_signal(signal.SIGTERM)
 
     # - - - - - - - - - - - - - - - - - -
     # ALLOW NEW ADMIN CONNECTIONS DURING SHUTDOWN
     # Start pgcat
-    pgcat_start()
+    utils.pgcat_start()
 
     # Create client connection and begin transaction
-    transaction_conn, transaction_cur = connect_db()
+    transaction_conn, transaction_cur = utils.connect_db()
 
     transaction_cur.execute("BEGIN;")
     transaction_cur.execute("SELECT 1;")
 
     # Send sigint to pgcat while still in transaction
-    pg_cat_send_signal(signal.SIGINT)
+    utils.pg_cat_send_signal(signal.SIGINT)
     time.sleep(1)
 
     try:
-        conn, cur = connect_db(admin=True)
+        conn, cur = utils.connect_db(admin=True)
         cur.execute("SHOW DATABASES;")
-        cleanup_conn(conn, cur)
+        utils.cleanup_conn(conn, cur)
     except psycopg2.OperationalError as e:
         raise Exception(e)
 
-    cleanup_conn(transaction_conn, transaction_cur)
-    pg_cat_send_signal(signal.SIGTERM)
+    utils.cleanup_conn(transaction_conn, transaction_cur)
+    utils.pg_cat_send_signal(signal.SIGTERM)
 
     # - - - - - - - - - - - - - - - - - -
     # ADMIN CONNECTIONS CONTINUING TO WORK AFTER SHUTDOWN
     # Start pgcat
-    pgcat_start()
+    utils.pgcat_start()
 
     # Create client connection and begin transaction
-    transaction_conn, transaction_cur = connect_db()
+    transaction_conn, transaction_cur = utils.connect_db()
     transaction_cur.execute("BEGIN;")
     transaction_cur.execute("SELECT 1;")
 
-    admin_conn, admin_cur = connect_db(admin=True)
+    admin_conn, admin_cur = utils.connect_db(admin=True)
     admin_cur.execute("SHOW DATABASES;")
 
     # Send sigint to pgcat while still in transaction
-    pg_cat_send_signal(signal.SIGINT)
+    utils.pg_cat_send_signal(signal.SIGINT)
     time.sleep(1)
 
     try:
@@ -277,24 +205,24 @@ def test_shutdown_logic():
     except psycopg2.OperationalError as e:
         raise Exception("Could not execute admin command:", e)
 
-    cleanup_conn(transaction_conn, transaction_cur)
-    cleanup_conn(admin_conn, admin_cur)
-    pg_cat_send_signal(signal.SIGTERM)
+    utils.cleanup_conn(transaction_conn, transaction_cur)
+    utils.cleanup_conn(admin_conn, admin_cur)
+    utils.pg_cat_send_signal(signal.SIGTERM)
 
     # - - - - - - - - - - - - - - - - - -
     # HANDLE SHUTDOWN TIMEOUT WITH SIGINT
 
     # Start pgcat
-    pgcat_start()
+    utils.pgcat_start()
 
     # Create client connection and begin transaction, which should prevent server shutdown unless shutdown timeout is reached
-    conn, cur = connect_db()
+    conn, cur = utils.connect_db()
 
     cur.execute("BEGIN;")
     cur.execute("SELECT 1;")
 
     # Send sigint to pgcat while still in transaction
-    pg_cat_send_signal(signal.SIGINT)
+    utils.pg_cat_send_signal(signal.SIGINT)
 
     # pgcat shutdown timeout is set to SHUTDOWN_TIMEOUT seconds, so we sleep for SHUTDOWN_TIMEOUT + 1 seconds
     time.sleep(SHUTDOWN_TIMEOUT + 1)
@@ -308,12 +236,7 @@ def test_shutdown_logic():
         # Fail if query execution succeeded
         raise Exception("Server not closed after sigint and expected timeout")
 
-    cleanup_conn(conn, cur)
-    pg_cat_send_signal(signal.SIGTERM)
+    utils.cleanup_conn(conn, cur)
+    utils.pg_cat_send_signal(signal.SIGTERM)
 
     # - - - - - - - - - - - - - - - - - -
-
-
-test_normal_db_access()
-test_admin_db_access()
-test_shutdown_logic()

tests/python/utils.py

@@ -0,0 +1,110 @@
+import os
+import signal
+import time
+from typing import Tuple
+import tempfile
+
+import psutil
+import psycopg2
+
+PGCAT_HOST = "127.0.0.1"
+PGCAT_PORT = "6432"
+
+
+def _pgcat_start(config_path: str):
+    pg_cat_send_signal(signal.SIGTERM)
+    os.system(f"./target/debug/pgcat {config_path} &")
+    time.sleep(2)
+
+
+def pgcat_start():
+    _pgcat_start(config_path='.circleci/pgcat.toml')
+
+
+def pgcat_generic_start(config: str):
+    tmp = tempfile.NamedTemporaryFile()
+    with open(tmp.name, 'w') as f:
+        f.write(config)
+    _pgcat_start(config_path=tmp.name)
+
+
+def glauth_send_signal(signal: signal.Signals):
+    try:
+        for proc in psutil.process_iter(["pid", "name"]):
+            if proc.name() == "glauth":
+                os.kill(proc.pid, signal)
+    except Exception as e:
+        # The process can be gone when we send this signal
+        print(e)
+
+    if signal == signal.SIGTERM:
+        # Returns 0 if pgcat process exists
+        time.sleep(2)
+        if not os.system('pgrep glauth'):
+            raise Exception("glauth not closed after SIGTERM")
+
+
+def pg_cat_send_signal(signal: signal.Signals):
+    try:
+        for proc in psutil.process_iter(["pid", "name"]):
+            if "pgcat" == proc.name():
+                os.kill(proc.pid, signal)
+    except Exception as e:
+        # The process can be gone when we send this signal
+        print(e)
+
+    if signal == signal.SIGTERM:
+        # Returns 0 if pgcat process exists
+        time.sleep(2)
+        if not os.system('pgrep pgcat'):
+            raise Exception("pgcat not closed after SIGTERM")
+
+
+def connect_db(
+    autocommit: bool = True,
+    admin: bool = False,
+) -> Tuple[psycopg2.extensions.connection, psycopg2.extensions.cursor]:
+
+    if admin:
+        user = "admin_user"
+        password = "admin_pass"
+        db = "pgcat"
+    else:
+        user = "sharding_user"
+        password = "sharding_user"
+        db = "sharded_db"
+
+    conn = psycopg2.connect(
+        f"postgres://{user}:{password}@{PGCAT_HOST}:{PGCAT_PORT}/{db}?application_name=testing_pgcat",
+        connect_timeout=2,
+    )
+    conn.autocommit = autocommit
+    cur = conn.cursor()
+
+    return (conn, cur)
+
+def connect_db_trust(
+    autocommit: bool = True,
+    admin: bool = False,
+) -> Tuple[psycopg2.extensions.connection, psycopg2.extensions.cursor]:
+
+    if admin:
+        user = "admin_user"
+        db = "pgcat"
+    else:
+        user = "sharding_user"
+        db = "sharded_db"
+
+    conn = psycopg2.connect(
+        f"postgres://{user}@{PGCAT_HOST}:{PGCAT_PORT}/{db}?application_name=testing_pgcat",
+        connect_timeout=2,
+    )
+    conn.autocommit = autocommit
+    cur = conn.cursor()
+
+    return (conn, cur)
+
+
+def cleanup_conn(conn: psycopg2.extensions.connection, cur: psycopg2.extensions.cursor):
+    cur.close()
+    conn.close()

tests/ruby/Gemfile.lock

@@ -1,22 +1,33 @@
 GEM
   remote: https://rubygems.org/
   specs:
-    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)
+    activemodel (7.1.4)
+      activesupport (= 7.1.4)
+    activerecord (7.1.4)
+      activemodel (= 7.1.4)
+      activesupport (= 7.1.4)
+      timeout (>= 0.4.0)
+    activesupport (7.1.4)
+      base64
+      bigdecimal
       concurrent-ruby (~> 1.0, >= 1.0.2)
+      connection_pool (>= 2.2.5)
+      drb
       i18n (>= 1.6, < 2)
       minitest (>= 5.1)
+      mutex_m
       tzinfo (~> 2.0)
     ast (2.4.2)
-    concurrent-ruby (1.1.10)
+    base64 (0.2.0)
+    bigdecimal (3.1.8)
+    concurrent-ruby (1.3.4)
+    connection_pool (2.4.1)
     diff-lcs (1.5.0)
-    i18n (1.12.0)
+    drb (2.2.1)
+    i18n (1.14.5)
       concurrent-ruby (~> 1.0)
-    minitest (5.17.0)
+    minitest (5.25.1)
+    mutex_m (0.2.0)
     parallel (1.22.1)
     parser (3.1.2.0)
       ast (~> 2.4.1)
@@ -24,7 +35,8 @@ GEM
     pg (1.3.2)
     rainbow (3.1.1)
     regexp_parser (2.3.1)
-    rexml (3.2.5)
+    rexml (3.3.6)
+      strscan
     rspec (3.11.0)
       rspec-core (~> 3.11.0)
       rspec-expectations (~> 3.11.0)
@@ -50,10 +62,12 @@ GEM
     rubocop-ast (1.17.0)
       parser (>= 3.1.1.0)
     ruby-progressbar (1.11.0)
+    strscan (3.1.0)
+    timeout (0.4.1)
     toml (0.3.0)
       parslet (>= 1.8.0, < 3.0.0)
     toxiproxy (2.0.1)
-    tzinfo (2.0.5)
+    tzinfo (2.0.6)
       concurrent-ruby (~> 1.0)
     unicode-display_width (2.1.0)
 

tests/ruby/admin_spec.rb

@@ -91,6 +91,27 @@ describe "Admin" do
     end
   end
 
+  [
+    "SHOW ME THE MONEY", 
+    "SHOW ME THE WAY", 
+    "SHOW UP", 
+    "SHOWTIME",
+    "HAMMER TIME",
+    "SHOWN TO BE TRUE",
+    "SHOW ",
+    "SHOW     ",
+    "SHOW 1",
+    ";;;;;"
+  ].each do |cmd|
+    describe "Bad command #{cmd}" do
+      it "does not panic and responds with PG::SystemError" do
+        admin_conn = PG::connect(processes.pgcat.admin_connection_string)
+        expect { admin_conn.async_exec(cmd) }.to raise_error(PG::SystemError).with_message(/Unsupported/)
+        admin_conn.close
+      end
+    end
+  end
+
   describe "PAUSE" do
     it "pauses all pools" do
       admin_conn = PG::connect(processes.pgcat.admin_connection_string)

tests/ruby/load_balancing_spec.rb

@@ -56,6 +56,41 @@ describe "Random Load Balancing" do
       end
     end
   end
+
+  context "when all replicas are down " do
+    let(:processes) { Helpers::Pgcat.single_shard_setup("sharded_db", 5, "transaction", "random", "debug", {"default_role" => "replica"}) }
+
+    it "unbans them automatically to prevent false positives in health checks that could make all replicas unavailable" do
+      conn = PG.connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
+      failed_count = 0
+      number_of_replicas = processes[:replicas].length
+
+      # Take down all replicas
+      processes[:replicas].each(&:take_down)
+
+      (number_of_replicas + 1).times do |n|
+        conn.async_exec("SELECT 1 + 2")
+      rescue
+        conn = PG.connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
+        failed_count += 1
+      end
+
+      expect(failed_count).to eq(number_of_replicas + 1)
+      failed_count = 0
+
+      # Ban_time is configured to 60 so this reset will only work
+      # if the replicas are unbanned automatically
+      processes[:replicas].each(&:reset)
+
+      number_of_replicas.times do
+        conn.async_exec("SELECT 1 + 2")
+      rescue
+        conn = PG.connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
+        failed_count += 1
+      end
+      expect(failed_count).to eq(0)
+    end
+  end
 end
 
 describe "Least Outstanding Queries Load Balancing" do
@@ -161,4 +196,3 @@ describe "Least Outstanding Queries Load Balancing" do
     end
   end
 end
-

tests/ruby/misc_spec.rb

@@ -188,6 +188,102 @@ describe "Miscellaneous" do
     end
   end
 
+  describe "Checkout failure limit" do
+    context "when no checkout failure limit is set" do
+      before do 
+        new_configs = processes.pgcat.current_config
+        new_configs["general"]["connect_timeout"] = 200
+        new_configs["pools"]["sharded_db"]["users"]["0"]["pool_size"] = 1
+        processes.pgcat.update_config(new_configs)
+        processes.pgcat.reload_config
+        sleep 0.5
+      end
+  
+      it "does not disconnect client" do
+        Array.new(5) do
+          Thread.new do
+            conn = PG::connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
+            for i in 0..4
+              begin
+                conn.async_exec("SELECT pg_sleep(0.5);")
+                expect(conn.status).to eq(PG::CONNECTION_OK)
+              rescue PG::SystemError
+                expect(conn.status).to eq(PG::CONNECTION_OK)
+              end
+            end
+            conn.close
+          end
+        end.each(&:join)
+      end
+    end
+
+    context "when checkout failure limit is set high" do
+      before do 
+        new_configs = processes.pgcat.current_config
+        new_configs["general"]["connect_timeout"] = 200
+        new_configs["pools"]["sharded_db"]["users"]["0"]["pool_size"] = 1
+        new_configs["pools"]["sharded_db"]["checkout_failure_limit"] = 10000
+        processes.pgcat.update_config(new_configs)
+        processes.pgcat.reload_config
+        sleep 0.5
+      end
+  
+      it "does not disconnect client" do
+        Array.new(5) do
+          Thread.new do
+            conn = PG::connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
+            for i in 0..4
+              begin
+                conn.async_exec("SELECT pg_sleep(0.5);")
+                expect(conn.status).to eq(PG::CONNECTION_OK)
+              rescue PG::SystemError
+                expect(conn.status).to eq(PG::CONNECTION_OK)
+              end
+            end
+            conn.close
+          end
+        end.each(&:join)
+      end
+    end
+
+    context "when checkout failure limit is set low" do
+      before do 
+        new_configs = processes.pgcat.current_config
+        new_configs["general"]["connect_timeout"] = 200
+        new_configs["pools"]["sharded_db"]["users"]["0"]["pool_size"] = 1
+        new_configs["pools"]["sharded_db"]["checkout_failure_limit"] = 2
+        processes.pgcat.update_config(new_configs)
+        processes.pgcat.reload_config
+        sleep 0.5
+      end
+  
+      it "disconnects client after reaching limit" do
+        Array.new(5) do
+          Thread.new do
+            conn = PG::connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
+            checkout_failure_count = 0
+            for i in 0..4
+              begin
+                conn.async_exec("SELECT pg_sleep(1);")
+                expect(conn.status).to eq(PG::CONNECTION_OK)
+              rescue PG::SystemError
+                checkout_failure_count += 1
+                expect(conn.status).to eq(PG::CONNECTION_OK)
+              rescue PG::ConnectionBad
+                expect(checkout_failure_count).to eq(2)
+                expect(conn.status).to eq(PG::CONNECTION_BAD)
+                break
+              end
+            end
+            conn.close
+          end
+        end.each(&:join)
+        puts processes.pgcat.logs
+
+      end
+    end
+  end
+  
   describe "Server version reporting" do
     it "reports correct version for normal and admin databases" do
       server_conn = PG::connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))

tests/ruby/stats_spec.rb

@@ -233,17 +233,19 @@ describe "Stats" do
         sleep(1.1) # Allow time for stats to update
         admin_conn = PG::connect(processes.pgcat.admin_connection_string)
         results = admin_conn.async_exec("SHOW POOLS")[0]
-        %w[cl_idle cl_cancel_req sv_idle sv_used sv_tested sv_login maxwait].each do |s|
+
+        %w[cl_idle cl_cancel_req sv_idle 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["maxwait"]).to eq("1")
         expect(results["cl_waiting"]).to eq("2")
         expect(results["cl_active"]).to eq("2")
         expect(results["sv_active"]).to eq("2")
 
         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].each do |s|
+        %w[cl_active cl_waiting cl_cancel_req sv_active sv_used sv_tested sv_login maxwait].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")
@@ -255,22 +257,23 @@ describe "Stats" do
 
       it "show correct max_wait" do
         threads = []
+        admin_conn = PG::connect(processes.pgcat.admin_connection_string)
         connections = Array.new(4) { PG::connect("#{pgcat_conn_str}?application_name=one_query") }
         connections.each do |c|
           threads << Thread.new { c.async_exec("SELECT pg_sleep(1.5)") rescue nil }
         end
+        sleep(1.1)
+        results = admin_conn.async_exec("SHOW POOLS")[0]
+        # Value is only reported when there are clients waiting
+        expect(results["maxwait"]).to eq("1")
+        expect(results["maxwait_us"].to_i).to be_within(20_000).of(100_000)
 
         sleep(2.5) # Allow time for stats to update
-        admin_conn = PG::connect(processes.pgcat.admin_connection_string)
-        results = admin_conn.async_exec("SHOW POOLS")[0]
-
-        expect(results["maxwait"]).to eq("1")
-        expect(results["maxwait_us"].to_i).to be_within(200_000).of(500_000)
-        connections.map(&:close)
-
-        sleep(4.5) # Allow time for stats to update
         results = admin_conn.async_exec("SHOW POOLS")[0]
+        # no clients are waiting so value is 0
         expect(results["maxwait"]).to eq("0")
+        expect(results["maxwait_us"]).to eq("0")
+        connections.map(&:close)
 
         threads.map(&:join)
       end
@@ -329,6 +332,40 @@ describe "Stats" do
       admin_conn.close
       connections.map(&:close)
     end
+
+    context "when client has waited for a server" do
+      let(:processes) { Helpers::Pgcat.single_instance_setup("sharded_db", 2) }
+
+      it "shows correct maxwait" do
+        threads = []
+        connections = Array.new(3) { |i| PG::connect("#{pgcat_conn_str}?application_name=app#{i}") }
+        connections.each do |c|
+          threads << Thread.new { c.async_exec("SELECT pg_sleep(1.5)") rescue nil }
+        end
+
+        sleep(2.5) # Allow time for stats to update
+        admin_conn = PG::connect(processes.pgcat.admin_connection_string)
+        results = admin_conn.async_exec("SHOW CLIENTS")
+
+        normal_client_results = results.reject { |r| r["database"] == "pgcat" }
+
+        non_waiting_clients = normal_client_results.select { |c| c["maxwait"] == "0" }
+        waiting_clients = normal_client_results.select { |c| c["maxwait"].to_i > 0 }
+
+        expect(non_waiting_clients.count).to eq(2)
+        non_waiting_clients.each do |client|
+          expect(client["maxwait_us"].to_i).to be_between(0, 50_000)
+        end
+
+        expect(waiting_clients.count).to eq(1)
+        waiting_clients.each do |client|
+          expect(client["maxwait_us"].to_i).to be_within(200_000).of(500_000)
+        end
+
+        admin_conn.close
+        connections.map(&:close)
+      end
+    end
   end
 
 

tests/rust/src/main.rs

@@ -15,8 +15,13 @@ async fn test_prepared_statements() {
     for _ in 0..5 {
         let pool = pool.clone();
         let handle = tokio::task::spawn(async move {
-            for _ in 0..1000 {
-                sqlx::query("SELECT 1").fetch_all(&pool).await.unwrap();
+            for i in 0..1000 {
+                match sqlx::query(&format!("SELECT {:?}", i % 5)).fetch_all(&pool).await {
+                    Ok(_) => (),
+                    Err(err) => {
+                        panic!("prepared statement error: {}", err);
+                    }
+                }
             }
         });
 

utilities/deb.sh

@@ -22,7 +22,7 @@ mkdir -p "$deb_dir/etc/systemd/system"
 cp target/release/pgcat "$deb_dir/usr/bin/pgcat"
 chmod +x "$deb_dir/usr/bin/pgcat"
 
-cp pgcat.toml "$deb_dir/etc/pgcat.toml"
+cp pgcat.toml "$deb_dir/etc/pgcat.example.toml"
 cp pgcat.service "$deb_dir/etc/systemd/system/pgcat.service"
 
 (cat control | envsubst) > "$deb_dir/DEBIAN/control"