Compare commits

..

2 Commits

Author SHA1 Message Date
Lev Kokotov
5872354c3e remove debug 2022-08-28 17:29:13 -07:00
Lev Kokotov
48bb6ebeef Support settings custom search path 2022-08-28 17:23:28 -07:00
58 changed files with 2309 additions and 8118 deletions

View File

@@ -9,43 +9,20 @@ jobs:
# Specify the execution environment. You can specify an image from Dockerhub or use one of our Convenience Images from CircleCI's Developer Hub.
# See: https://circleci.com/docs/2.0/configuration-reference/#docker-machine-macos-windows-executor
docker:
- image: ghcr.io/levkk/pgcat-ci:1.67
- image: levkk/pgcat-ci:latest
environment:
RUST_LOG: info
LLVM_PROFILE_FILE: /tmp/pgcat-%m-%p.profraw
RUSTC_BOOTSTRAP: 1
CARGO_INCREMENTAL: 0
RUSTFLAGS: "-Zprofile -Ccodegen-units=1 -Copt-level=0 -Clink-dead-code -Coverflow-checks=off -Zpanic_abort_tests -Cpanic=abort -Cinstrument-coverage"
RUSTDOCFLAGS: "-Cpanic=abort"
RUSTFLAGS: "-C instrument-coverage"
LLVM_PROFILE_FILE: "pgcat-%m.profraw"
- image: postgres:14
command: ["postgres", "-p", "5432", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-c", "pg_stat_statements.max=100000"]
# auth:
# username: mydockerhub-user
# password: $DOCKERHUB_PASSWORD
environment:
POSTGRES_USER: postgres
POSTGRES_DB: postgres
POSTGRES_PASSWORD: postgres
POSTGRES_INITDB_ARGS: --auth-local=md5 --auth-host=md5 --auth=md5
- image: postgres:14
command: ["postgres", "-p", "7432", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-c", "pg_stat_statements.max=100000"]
environment:
POSTGRES_USER: postgres
POSTGRES_DB: postgres
POSTGRES_PASSWORD: postgres
POSTGRES_INITDB_ARGS: --auth-local=scram-sha-256 --auth-host=scram-sha-256 --auth=scram-sha-256
- image: postgres:14
command: ["postgres", "-p", "8432", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-c", "pg_stat_statements.max=100000"]
environment:
POSTGRES_USER: postgres
POSTGRES_DB: postgres
POSTGRES_PASSWORD: postgres
POSTGRES_INITDB_ARGS: --auth-local=scram-sha-256 --auth-host=scram-sha-256 --auth=scram-sha-256
- image: postgres:14
command: ["postgres", "-p", "9432", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-c", "pg_stat_statements.max=100000"]
environment:
POSTGRES_USER: postgres
POSTGRES_DB: postgres
POSTGRES_PASSWORD: postgres
POSTGRES_INITDB_ARGS: --auth-local=scram-sha-256 --auth-host=scram-sha-256 --auth=scram-sha-256
POSTGRES_HOST_AUTH_METHOD: scram-sha-256
# Add steps to the job
# See: https://circleci.com/docs/2.0/configuration-reference/#steps
steps:
@@ -55,9 +32,18 @@ jobs:
- run:
name: "Lint"
command: "cargo fmt --check"
- run:
name: "Install dependencies"
command: "sudo apt-get update && sudo apt-get install -y psmisc postgresql-contrib-12 postgresql-client-12 ruby ruby-dev libpq-dev python3 python3-pip lcov llvm-11 && sudo apt-get upgrade curl"
- run:
name: "Install rust tools"
command: "cargo install cargo-binutils rustfilt && rustup component add llvm-tools-preview"
- run:
name: "Build"
command: "cargo build"
- run:
name: "Tests"
command: "cargo clean && cargo build && cargo test && bash .circleci/run_tests.sh && .circleci/generate_coverage.sh"
command: "cargo test && bash .circleci/run_tests.sh && .circleci/generate_coverage.sh"
- store_artifacts:
path: /tmp/cov
destination: coverage-data

View File

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

View File

@@ -18,10 +18,10 @@ enable_prometheus_exporter = true
prometheus_exporter_port = 9930
# How long to wait before aborting a server connection (ms).
connect_timeout = 1000
connect_timeout = 100
# How much time to give the health check query to return with a result (ms).
healthcheck_timeout = 1000
healthcheck_timeout = 100
# How long to keep connection available for immediate re-use, without running a healthcheck query on it
healthcheck_delay = 30000
@@ -32,12 +32,6 @@ shutdown_timeout = 5000
# For how long to ban a server if it fails a health check (seconds).
ban_time = 60 # Seconds
# If we should log client connections
log_client_connections = false
# If we should log client disconnections
log_client_disconnections = false
# Reload config automatically if it changes.
autoreload = true
@@ -114,6 +108,7 @@ servers = [
]
# Database name (e.g. "postgres")
database = "shard0"
search_path = "\"$user\",public"
[pools.sharded_db.shards.1]
servers = [

View File

@@ -16,16 +16,17 @@ function start_pgcat() {
# Setup the database with shards and user
PGPASSWORD=postgres psql -e -h 127.0.0.1 -p 5432 -U postgres -f tests/sharding/query_routing_setup.sql
PGPASSWORD=postgres psql -e -h 127.0.0.1 -p 7432 -U postgres -f tests/sharding/query_routing_setup.sql
PGPASSWORD=postgres psql -e -h 127.0.0.1 -p 8432 -U postgres -f tests/sharding/query_routing_setup.sql
PGPASSWORD=postgres psql -e -h 127.0.0.1 -p 9432 -U postgres -f tests/sharding/query_routing_setup.sql
PGPASSWORD=sharding_user pgbench -h 127.0.0.1 -U sharding_user shard0 -i
PGPASSWORD=sharding_user pgbench -h 127.0.0.1 -U sharding_user shard1 -i
PGPASSWORD=sharding_user pgbench -h 127.0.0.1 -U sharding_user shard2 -i
# Install Toxiproxy to simulate a downed/slow database
wget -O toxiproxy-2.4.0.deb https://github.com/Shopify/toxiproxy/releases/download/v2.4.0/toxiproxy_2.4.0_linux_$(dpkg --print-architecture).deb
sudo dpkg -i toxiproxy-2.4.0.deb
# Start Toxiproxy
LOG_LEVEL=error toxiproxy-server &
toxiproxy-server &
sleep 1
# Create a database at port 5433, forward it to Postgres
@@ -52,17 +53,6 @@ psql -U sharding_user -h 127.0.0.1 -p 6432 -c 'COPY (SELECT * FROM pgbench_accou
sleep 1
killall psql -s SIGINT
# Pause/resume test.
# Running benches before, during, and after pause/resume.
pgbench -U sharding_user -t 500 -c 2 -h 127.0.0.1 -p 6432 --protocol extended &
BENCH_ONE=$!
PGPASSWORD=admin_pass psql -U admin_user -h 127.0.0.1 -p 6432 -d pgbouncer -c 'PAUSE sharded_db,sharding_user'
pgbench -U sharding_user -h 127.0.0.1 -p 6432 -t 500 -c 2 --protocol extended &
BENCH_TWO=$!
PGPASSWORD=admin_pass psql -U admin_user -h 127.0.0.1 -p 6432 -d pgbouncer -c 'RESUME sharded_db,sharding_user'
wait ${BENCH_ONE}
wait ${BENCH_TWO}
# Reload pool (closing unused server connections)
PGPASSWORD=admin_pass psql -U admin_user -h 127.0.0.1 -p 6432 -d pgbouncer -c 'RELOAD'
@@ -92,12 +82,12 @@ sed -i 's/statement_timeout = 100/statement_timeout = 0/' .circleci/pgcat.toml
kill -SIGHUP $(pgrep pgcat) # Reload config again
#
# Integration tests and ActiveRecord tests
# ActiveRecord tests
#
cd tests/ruby
sudo bundle install
bundle exec ruby tests.rb --format documentation || exit 1
bundle exec rspec *_spec.rb --format documentation || exit 1
sudo gem install bundler
bundle install
ruby tests.rb
cd ../..
#
@@ -105,7 +95,7 @@ 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
python3 tests/python/tests.py
start_pgcat "info"
@@ -115,9 +105,9 @@ psql -U admin_user -e -h 127.0.0.1 -p 6432 -d pgbouncer -c 'SHOW STATS' > /dev/n
psql -U admin_user -h 127.0.0.1 -p 6432 -d pgbouncer -c 'RELOAD' > /dev/null
psql -U admin_user -h 127.0.0.1 -p 6432 -d pgbouncer -c 'SHOW CONFIG' > /dev/null
psql -U admin_user -h 127.0.0.1 -p 6432 -d pgbouncer -c 'SHOW DATABASES' > /dev/null
psql -U admin_user -h 127.0.0.1 -p 6432 -d pgcat -c 'SHOW LISTS' > /dev/null
psql -U admin_user -h 127.0.0.1 -p 6432 -d pgcat -c 'SHOW POOLS' > /dev/null
psql -U admin_user -h 127.0.0.1 -p 6432 -d pgcat -c 'SHOW VERSION' > /dev/null
psql -U admin_user -h 127.0.0.1 -p 6432 -d pgbouncer -c 'SHOW LISTS' > /dev/null
psql -U admin_user -h 127.0.0.1 -p 6432 -d pgbouncer -c 'SHOW POOLS' > /dev/null
psql -U admin_user -h 127.0.0.1 -p 6432 -d pgbouncer -c 'SHOW VERSION' > /dev/null
psql -U admin_user -h 127.0.0.1 -p 6432 -d pgbouncer -c "SET client_encoding TO 'utf8'" > /dev/null # will ignore
(! psql -U admin_user -e -h 127.0.0.1 -p 6432 -d random_db -c 'SHOW STATS' > /dev/null)
export PGPASSWORD=sharding_user

View File

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

View File

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

View File

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

View File

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

10
.gitignore vendored
View File

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

340
CONFIG.md
View File

@@ -1,340 +0,0 @@
# PgCat Configurations
## `general` Section
### host
```
path: general.host
default: "0.0.0.0"
```
What IP to run on, 0.0.0.0 means accessible from everywhere.
### port
```
path: general.port
default: 6432
```
Port to run on, same as PgBouncer used in this example.
### enable_prometheus_exporter
```
path: general.enable_prometheus_exporter
default: true
```
Whether to enable prometheus exporter or not.
### prometheus_exporter_port
```
path: general.prometheus_exporter_port
default: 9930
```
Port at which prometheus exporter listens on.
### connect_timeout
```
path: general.connect_timeout
default: 5000 # milliseconds
```
How long to wait before aborting a server connection (ms).
### idle_timeout
```
path: general.idle_timeout
default: 30000 # milliseconds
```
How long an idle connection with a server is left open (ms).
### healthcheck_timeout
```
path: general.healthcheck_timeout
default: 1000 # milliseconds
```
How much time to give the health check query to return with a result (ms).
### healthcheck_delay
```
path: general.healthcheck_delay
default: 30000 # milliseconds
```
How long to keep connection available for immediate re-use, without running a healthcheck query on it
### shutdown_timeout
```
path: general.shutdown_timeout
default: 60000 # milliseconds
```
How much time to give clients during shutdown before forcibly killing client connections (ms).
### ban_time
```
path: general.ban_time
default: 60 # seconds
```
How long to ban a server if it fails a health check (seconds).
### log_client_connections
```
path: general.log_client_connections
default: false
```
If we should log client connections
### log_client_disconnections
```
path: general.log_client_disconnections
default: false
```
If we should log client disconnections
### autoreload
```
path: general.autoreload
default: false
```
When set to true, PgCat reloads configs if it detects a change in the config file.
### worker_threads
```
path: general.worker_threads
default: 5
```
Number of worker threads the Runtime will use (4 by default).
### tcp_keepalives_idle
```
path: general.tcp_keepalives_idle
default: 5
```
Number of seconds of connection idleness to wait before sending a keepalive packet to the server.
### tcp_keepalives_count
```
path: general.tcp_keepalives_count
default: 5
```
Number of unacknowledged keepalive packets allowed before giving up and closing the connection.
### tcp_keepalives_interval
```
path: general.tcp_keepalives_interval
default: 5
```
Number of seconds between keepalive packets.
### tls_certificate
```
path: general.tls_certificate
default: <UNSET>
example: "server.cert"
```
Path to TLS Certficate file to use for TLS connections
### tls_private_key
```
path: general.tls_private_key
default: <UNSET>
example: "server.key"
```
Path to TLS private key file to use for TLS connections
### admin_username
```
path: general.admin_username
default: "admin_user"
```
User name to access the virtual administrative database (pgbouncer or pgcat)
Connecting to that database allows running commands like `SHOW POOLS`, `SHOW DATABASES`, etc..
### admin_password
```
path: general.admin_password
default: "admin_pass"
```
Password to access the virtual administrative database
## `pools.<pool_name>` Section
### pool_mode
```
path: pools.<pool_name>.pool_mode
default: "transaction"
```
Pool mode (see PgBouncer docs for more).
`session` one server connection per connected client
`transaction` one server connection per client transaction
### load_balancing_mode
```
path: pools.<pool_name>.load_balancing_mode
default: "random"
```
Load balancing mode
`random` selects the server at random
`loc` selects the server with the least outstanding busy conncetions
### default_role
```
path: pools.<pool_name>.default_role
default: "any"
```
If the client doesn't specify, PgCat routes traffic to this role by default.
`any` round-robin between primary and replicas,
`replica` round-robin between replicas only without touching the primary,
`primary` all queries go to the primary unless otherwise specified.
### query_parser_enabled (experimental)
```
path: pools.<pool_name>.query_parser_enabled
default: true
```
If Query Parser is enabled, we'll attempt to parse
every incoming query to determine if it's a read or a write.
If it's a read query, we'll direct it to a replica. Otherwise, if it's a write,
we'll direct it to the primary.
### primary_reads_enabled
```
path: pools.<pool_name>.primary_reads_enabled
default: true
```
If the query parser is enabled and this setting is enabled, the primary will be part of the pool of databases used for
load balancing of read queries. Otherwise, the primary will only be used for write
queries. The primary can always be explicitly selected with our custom protocol.
### sharding_key_regex (experimental)
```
path: pools.<pool_name>.sharding_key_regex
default: <UNSET>
example: '/\* sharding_key: (\d+) \*/'
```
Allow sharding commands to be passed as statement comments instead of
separate commands. If these are unset this functionality is disabled.
### sharding_function
```
path: pools.<pool_name>.sharding_function
default: "pg_bigint_hash"
```
So what if you wanted to implement a different hashing function,
or you've already built one and you want this pooler to use it?
Current options:
`pg_bigint_hash`: PARTITION BY HASH (Postgres hashing function)
`sha1`: A hashing function based on SHA1
### automatic_sharding_key (experimental)
```
path: pools.<pool_name>.automatic_sharding_key
default: <UNSET>
example: "data.id"
```
Automatically parse this from queries and route queries to the right shard!
### idle_timeout
```
path: pools.<pool_name>.idle_timeout
default: 40000
```
Idle timeout can be overwritten in the pool
### connect_timeout
```
path: pools.<pool_name>.connect_timeout
default: 3000
```
Connect timeout can be overwritten in the pool
## `pools.<pool_name>.users.<user_index>` Section
### username
```
path: pools.<pool_name>.users.<user_index>.username
default: "sharding_user"
```
Postgresql username
### password
```
path: pools.<pool_name>.users.<user_index>.password
default: "sharding_user"
```
Postgresql password
### pool_size
```
path: pools.<pool_name>.users.<user_index>.pool_size
default: 9
```
Maximum number of server connections that can be established for this user
The maximum number of connection from a single Pgcat process to any database in the cluster
is the sum of pool_size across all users.
### statement_timeout
```
path: pools.<pool_name>.users.<user_index>.statement_timeout
default: 0
```
Maximum query duration. Dangerous, but protects against DBs that died in a non-obvious way.
0 means it is disabled.
## `pools.<pool_name>.shards.<shard_index>` Section
### servers
```
path: pools.<pool_name>.shards.<shard_index>.servers
default: [["127.0.0.1", 5432, "primary"], ["localhost", 5432, "replica"]]
```
Array of servers in the shard, each server entry is an array of `[host, port, role]`
### mirrors (experimental)
```
path: pools.<pool_name>.shards.<shard_index>.mirrors
default: <UNSET>
example: [["1.2.3.4", 5432, 0], ["1.2.3.4", 5432, 1]]
```
Array of mirrors for the shard, each mirror entry is an array of `[host, port, index of server in servers array]`
Traffic hitting the server identified by the index will be sent to the mirror.
### database
```
path: pools.<pool_name>.shards.<shard_index>.database
default: "shard0"
```
Database name (e.g. "postgres")

847
Cargo.lock generated

File diff suppressed because it is too large Load Diff

View File

@@ -9,34 +9,28 @@ edition = "2021"
tokio = { version = "1", features = ["full"] }
bytes = "1"
md-5 = "0.10"
bb8 = "0.8.0"
bb8 = "0.7"
async-trait = "0.1"
rand = "0.8"
chrono = "0.4"
sha-1 = "0.10"
toml = "0.7"
toml = "0.5"
serde = "1"
serde_derive = "1"
regex = "1"
num_cpus = "1"
once_cell = "1"
sqlparser = "0.32.0"
sqlparser = "0.14"
log = "0.4"
arc-swap = "1"
env_logger = "0.10"
parking_lot = "0.12.1"
env_logger = "0.9"
parking_lot = "0.11"
hmac = "0.12"
sha2 = "0.10"
base64 = "0.21"
base64 = "0.13"
stringprep = "0.1"
tokio-rustls = "0.23"
rustls-pemfile = "1"
hyper = { version = "0.14", features = ["full"] }
phf = { version = "0.11.1", features = ["macros"] }
phf = { version = "0.10", features = ["macros"] }
exitcode = "1.1.2"
futures = "0.3"
socket2 = { version = "0.4.7", features = ["all"] }
nix = "0.26.2"
[target.'cfg(not(target_env = "msvc"))'.dependencies]
jemallocator = "0.5.0"

View File

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

View File

@@ -1,9 +1,8 @@
##### PgCat: PostgreSQL at petabyte scale
# PgCat
[![CircleCI](https://circleci.com/gh/levkk/pgcat/tree/main.svg?style=svg)](https://circleci.com/gh/levkk/pgcat/tree/main)
<a href="https://discord.gg/DmyJP3qJ7U" target="_blank">
<img src="https://img.shields.io/discord/1013868243036930099" alt="Join our Discord!" />
</a>
![PgCat](./pgcat3.png)
PostgreSQL pooler (like PgBouncer) with sharding, load balancing and failover support.
@@ -34,12 +33,39 @@ For quick local example, use the Docker Compose environment provided:
docker-compose up
# In a new terminal:
PGPASSWORD=postgres psql -h 127.0.0.1 -p 6432 -U postgres -c 'SELECT 1'
psql -h 127.0.0.1 -p 6432 -c 'SELECT 1'
```
### Config
See [Configurations page](https://github.com/levkk/pgcat/blob/main/CONFIG.md)
| **Name** | **Description** | **Examples** |
|------------------------------|--------------------------------------------------------------------------------------------------------------------------------------------|----------------------------------|
| **`general`** | | |
| `host` | The pooler will run on this host, 0.0.0.0 means accessible from everywhere. | `0.0.0.0` |
| `port` | The pooler will run on this port. | `6432` |
| `enable_prometheus_exporter` | Enable prometheus exporter which will export metrics in prometheus exposition format. | `true` |
| `prometheus_exporter_port` | Port at which prometheus exporter listens on. | `9930` |
| `pool_size` | Maximum allowed server connections per pool. Pools are separated for each user/shard/server role. The connections are allocated as needed. | `15` |
| `pool_mode` | The pool mode to use, i.e. `session` or `transaction`. | `transaction` |
| `connect_timeout` | Maximum time to establish a connection to a server (milliseconds). If reached, the server is banned and the next target is attempted. | `5000` |
| `healthcheck_timeout` | Maximum time to pass a health check (`SELECT 1`, milliseconds). If reached, the server is banned and the next target is attempted. | `1000` |
| `shutdown_timeout` | Maximum time to give clients during shutdown before forcibly killing client connections (ms). | `60000` |
| `healthcheck_delay` | How long to keep connection available for immediate re-use, without running a healthcheck query on it | `30000` |
| `ban_time` | Ban time for a server (seconds). It won't be allowed to serve transactions until the ban expires; failover targets will be used instead. | `60` |
| `autoreload` | Enable auto-reload of config after fixed time-interval. | `false` |
| | | |
| **`user`** | | |
| `name` | The user name. | `sharding_user` |
| `password` | The user password in plaintext. | `hunter2` |
| | | |
| **`shards`** | Shards are numerically numbered starting from 0; the order in the config is preserved by the pooler to route queries accordingly. | `[shards.0]` |
| `servers` | List of servers to connect to and their roles. A server is: `[host, port, role]`, where `role` is either `primary` or `replica`. | `["127.0.0.1", 5432, "primary"]` |
| `database` | The name of the database to connect to. This is the same on all servers that are part of one shard. | |
| | | |
| **`query_router`** | | |
| `default_role` | Traffic is routed to this role by default (random), unless the client specifies otherwise. Default is `any`, for any role available. | `any`, `primary`, `replica` |
| `query_parser_enabled` | Enable the query parser which will inspect incoming queries and route them to a primary or replicas. | `false` |
| `primary_reads_enabled` | Enable this to allow read queries on the primary; otherwise read queries are routed to the replicas. | `true` |
## Local development
@@ -61,14 +87,6 @@ pgbench -t 1000 -p 6432 -h 127.0.0.1 --protocol extended
See [sharding README](./tests/sharding/README.md) for sharding logic testing.
Run `cargo test` to run Rust tests.
Run the following commands to run Integration tests locally.
```
cd tests/docker/
docker compose up --exit-code-from main # This will also produce coverage report under ./cov/
```
| **Feature** | **Tested in CI** | **Tested manually** | **Comments** |
|-----------------------|--------------------|---------------------|--------------------------------------------------------------------------------------------------------------------------|
| Transaction pooling | :white_check_mark: | :white_check_mark: | Used by default for all tests. |
@@ -81,17 +99,6 @@ docker compose up --exit-code-from main # This will also produce coverage report
| Statistics | :white_check_mark: | :white_check_mark: | Query the admin database with `psql -h 127.0.0.1 -p 6432 -d pgbouncer -c 'SHOW STATS'`. |
| Live config reloading | :white_check_mark: | :white_check_mark: | Run `kill -s SIGHUP $(pgrep pgcat)` and watch the config reload. |
### Dev
Also, you can open a 'dev' environment where you can debug tests easier by running the following command:
```
./dev/script/console
```
This will open a terminal in an environment similar to that used in tests. In there you can compile, run tests, do some debugging with the test environment, etc. Objects
compiled inside the contaner (and bundled gems) will be placed in `dev/cache` so they don't interfere with what you have in your host.
## Usage
### Session mode
@@ -171,7 +178,7 @@ We use the `PARTITION BY HASH` hashing function, the same as used by Postgres fo
To route queries to a particular shard, we use this custom SQL syntax:
```sql
-- To talk to a shard explicitly
-- To talk to a shard explicitely
SET SHARD TO '1';
-- To let the pooler choose based on a value
@@ -440,7 +447,7 @@ Always good to have a base line.
```
$ pgbench -t 1000 -c 16 -j 2 -p 5432 -h 127.0.0.1 -S --protocol extended shard0
Password:
Password:
starting vacuum...end.
transaction type: <builtin: select only>
scaling factor: 1
@@ -454,7 +461,7 @@ tps = 139443.955722 (including connections establishing)
tps = 142314.859075 (excluding connections establishing)
$ pgbench -t 1000 -c 32 -j 2 -p 5432 -h 127.0.0.1 -S --protocol extended shard0
Password:
Password:
starting vacuum...end.
transaction type: <builtin: select only>
scaling factor: 1
@@ -468,7 +475,7 @@ tps = 150644.840891 (including connections establishing)
tps = 152218.499430 (excluding connections establishing)
$ pgbench -t 1000 -c 64 -j 2 -p 5432 -h 127.0.0.1 -S --protocol extended shard0
Password:
Password:
starting vacuum...end.
transaction type: <builtin: select only>
scaling factor: 1
@@ -482,7 +489,7 @@ tps = 152517.663404 (including connections establishing)
tps = 153319.188482 (excluding connections establishing)
$ pgbench -t 1000 -c 128 -j 2 -p 5432 -h 127.0.0.1 -S --protocol extended shard0
Password:
Password:
starting vacuum...end.
transaction type: <builtin: select only>
scaling factor: 1

View File

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

View File

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

View File

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

View File

@@ -1,84 +0,0 @@
version: "3"
x-common-definition-pg:
&common-definition-pg
image: postgres:14
network_mode: "service:main"
healthcheck:
test: [ "CMD-SHELL", "pg_isready -U postgres -d postgres" ]
interval: 5s
timeout: 5s
retries: 5
volumes:
- type: bind
source: ../tests/sharding/query_routing_setup.sql
target: /docker-entrypoint-initdb.d/query_routing_setup.sql
- type: bind
source: ../tests/sharding/partition_hash_test_setup.sql
target: /docker-entrypoint-initdb.d/partition_hash_test_setup.sql
x-common-env-pg:
&common-env-pg
POSTGRES_USER: postgres
POSTGRES_DB: postgres
POSTGRES_PASSWORD: postgres
services:
main:
image: kubernetes/pause
pg1:
<<: *common-definition-pg
environment:
<<: *common-env-pg
POSTGRES_INITDB_ARGS: --auth-local=md5 --auth-host=md5 --auth=md5
PGPORT: 5432
command: ["postgres", "-p", "5432", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-c", "pg_stat_statements.max=100000"]
pg2:
<<: *common-definition-pg
environment:
<<: *common-env-pg
POSTGRES_INITDB_ARGS: --auth-local=scram-sha-256 --auth-host=scram-sha-256 --auth=scram-sha-256
PGPORT: 7432
command: ["postgres", "-p", "7432", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-c", "pg_stat_statements.max=100000"]
pg3:
<<: *common-definition-pg
environment:
<<: *common-env-pg
POSTGRES_INITDB_ARGS: --auth-local=scram-sha-256 --auth-host=scram-sha-256 --auth=scram-sha-256
PGPORT: 8432
command: ["postgres", "-p", "8432", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-c", "pg_stat_statements.max=100000"]
pg4:
<<: *common-definition-pg
environment:
<<: *common-env-pg
POSTGRES_INITDB_ARGS: --auth-local=scram-sha-256 --auth-host=scram-sha-256 --auth=scram-sha-256
PGPORT: 9432
command: ["postgres", "-p", "9432", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-c", "pg_stat_statements.max=100000"]
toxiproxy:
build: .
network_mode: "service:main"
container_name: toxiproxy
environment:
LOG_LEVEL: info
entrypoint: toxiproxy-server
depends_on:
- pg1
- pg2
- pg3
- pg4
pgcat-shell:
stdin_open: true
user: "${HOST_UID}:${HOST_GID}"
build: .
network_mode: "service:main"
depends_on:
- toxiproxy
volumes:
- ../:/app/
entrypoint:
- /bin/bash
- -i

View File

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

View File

@@ -32,12 +32,6 @@ shutdown_timeout = 60000
# 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 = false
@@ -54,7 +48,7 @@ admin_password = "postgres"
# configs are structured as pool.<pool_name>
# the pool_name is what clients use as database name when connecting
# For the example below a client can connect using "postgres://sharding_user:sharding_user@pgcat_host:pgcat_port/sharded"
[pools.postgres]
[pools.sharded]
# Pool mode (see PgBouncer docs for more).
# session: one server connection per connected client
# transaction: one server connection per client transaction
@@ -90,7 +84,7 @@ primary_reads_enabled = true
sharding_function = "pg_bigint_hash"
# Credentials for users that may connect to this cluster
[pools.postgres.users.0]
[pools.sharded.users.0]
username = "postgres"
password = "postgres"
# Maximum number of server connections that can be established for this user
@@ -101,8 +95,14 @@ pool_size = 9
# Maximum query duration. Dangerous, but protects against DBs that died in a non-obvious way.
statement_timeout = 0
[pools.sharded.users.1]
username = "postgres"
password = "postgres"
pool_size = 21
statement_timeout = 15000
# Shard 0
[pools.postgres.shards.0]
[pools.sharded.shards.0]
# [ host, port, role ]
servers = [
[ "postgres", 5432, "primary" ],
@@ -111,16 +111,37 @@ servers = [
# Database name (e.g. "postgres")
database = "postgres"
[pools.postgres.shards.1]
[pools.sharded.shards.1]
servers = [
[ "postgres", 5432, "primary" ],
[ "postgres", 5432, "replica" ],
]
database = "postgres"
[pools.postgres.shards.2]
[pools.sharded.shards.2]
servers = [
[ "postgres", 5432, "primary" ],
[ "postgres", 5432, "replica" ],
]
database = "postgres"
[pools.simple_db]
pool_mode = "session"
default_role = "primary"
query_parser_enabled = true
primary_reads_enabled = true
sharding_function = "pg_bigint_hash"
[pools.simple_db.users.0]
username = "postgres"
password = "postgres"
pool_size = 5
statement_timeout = 0
[pools.simple_db.shards.0]
servers = [
[ "postgres", 5432, "primary" ],
[ "postgres", 5432, "replica" ]
]
database = "postgres"

View File

@@ -18,75 +18,51 @@ enable_prometheus_exporter = true
prometheus_exporter_port = 9930
# How long to wait before aborting a server connection (ms).
connect_timeout = 5000 # milliseconds
# How long an idle connection with a server is left open (ms).
idle_timeout = 30000 # milliseconds
connect_timeout = 5000
# How much time to give the health check query to return with a result (ms).
healthcheck_timeout = 1000 # milliseconds
healthcheck_timeout = 1000
# How long to keep connection available for immediate re-use, without running a healthcheck query on it
healthcheck_delay = 30000 # milliseconds
healthcheck_delay = 30000
# How much time to give clients during shutdown before forcibly killing client connections (ms).
shutdown_timeout = 60000 # milliseconds
shutdown_timeout = 60000
# How long to ban a server if it fails a health check (seconds).
# 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
# When set to true, PgCat reloads configs if it detects a change in the config file.
# Reload config automatically if it changes.
autoreload = false
# Number of worker threads the Runtime will use (4 by default).
worker_threads = 5
# Number of seconds of connection idleness to wait before sending a keepalive packet to the server.
tcp_keepalives_idle = 5
# Number of unacknowledged keepalive packets allowed before giving up and closing the connection.
tcp_keepalives_count = 5
# Number of seconds between keepalive packets.
tcp_keepalives_interval = 5
# Path to TLS Certficate file to use for TLS connections
# TLS
# tls_certificate = "server.cert"
# Path to TLS private key file to use for TLS connections
# tls_private_key = "server.key"
# User name to access the virtual administrative database (pgbouncer or pgcat)
# 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"
# Password to access the virtual administrative database
admin_password = "admin_pass"
# pool configs are structured as pool.<pool_name>
# the pool_name is what clients use as database name when connecting.
# For a pool named `sharded_db`, clients access that pool using connection string like
# `postgres://sharding_user:sharding_user@pgcat_host:pgcat_port/sharded_db`
# pool
# configs are structured as pool.<pool_name>
# the pool_name is what clients use as database name when connecting
# For 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
# session: one server connection per connected client
# transaction: one server connection per client transaction
pool_mode = "transaction"
# Load balancing mode
# `random` selects the server at random
# `loc` selects the server with the least outstanding busy conncetions
load_balancing_mode = "random"
# If the client doesn't specify, PgCat routes traffic to this role by default.
# `any` round-robin between primary and replicas,
# `replica` round-robin between replicas only without touching the primary,
# `primary` all queries go to the primary unless otherwise specified.
# 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"
# If Query Parser is enabled, we'll attempt to parse
# 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.
@@ -97,34 +73,19 @@ query_parser_enabled = true
# queries. The primary can always be explicitly selected with our custom protocol.
primary_reads_enabled = true
# Allow sharding commands to be passed as statement comments instead of
# separate commands. If these are unset this functionality is disabled.
# sharding_key_regex = '/\* sharding_key: (\d+) \*/'
# shard_id_regex = '/\* shard_id: (\d+) \*/'
# regex_search_limit = 1000 # only look at the first 1000 characters of SQL statements
# So what if you wanted to implement a different hashing function,
# or you've already built one and you want this pooler to use it?
#
# Current options:
# `pg_bigint_hash`: PARTITION BY HASH (Postgres hashing function)
# `sha1`: A hashing function based on SHA1
#
# pg_bigint_hash: PARTITION BY HASH (Postgres hashing function)
# sha1: A hashing function based on SHA1
#
sharding_function = "pg_bigint_hash"
# Automatically parse this from queries and route queries to the right shard!
# automatic_sharding_key = "data.id"
# Idle timeout can be overwritten in the pool
idle_timeout = 40000
# Connect timeout can be overwritten in the pool
connect_timeout = 3000
# User configs are structured as pool.<pool_name>.users.<user_index>
# This secion holds the credentials for users that may connect to this cluster
# Credentials for users that may connect to this cluster
[pools.sharded_db.users.0]
# Postgresql username
username = "sharding_user"
# Postgresql password
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
@@ -132,7 +93,6 @@ password = "sharding_user"
pool_size = 9
# Maximum query duration. Dangerous, but protects against DBs that died in a non-obvious way.
# 0 means it is disabled.
statement_timeout = 0
[pools.sharded_db.users.1]
@@ -141,26 +101,28 @@ password = "other_user"
pool_size = 21
statement_timeout = 15000
# Shard configs are structured as pool.<pool_name>.shards.<shard_id>
# Each shard config contains a list of servers that make up the shard
# and the database name to use.
# Shard 0
[pools.sharded_db.shards.0]
# Array of servers in the shard, each server entry is an array of `[host, port, role]`
servers = [["127.0.0.1", 5432, "primary"], ["localhost", 5432, "replica"]]
# Array of mirrors for the shard, each mirror entry is an array of `[host, port, index of server in servers array]`
# Traffic hitting the server identified by the index will be sent to the mirror.
# mirrors = [["1.2.3.4", 5432, 0], ["1.2.3.4", 5432, 1]]
# [ 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"]]
servers = [
[ "127.0.0.1", 5432, "primary" ],
[ "localhost", 5432, "replica" ],
]
database = "shard1"
[pools.sharded_db.shards.2]
servers = [["127.0.0.1", 5432, "primary" ], ["localhost", 5432, "replica" ]]
servers = [
[ "127.0.0.1", 5432, "primary" ],
[ "localhost", 5432, "replica" ],
]
database = "shard2"

BIN
pgcat3.png Normal file

Binary file not shown.

After

Width:  |  Height:  |  Size: 44 KiB

View File

@@ -1,32 +1,25 @@
use crate::pool::BanReason;
/// Admin database.
use bytes::{Buf, BufMut, BytesMut};
use log::{error, info, trace};
use nix::sys::signal::{self, Signal};
use nix::unistd::Pid;
use log::{info, trace};
use std::collections::HashMap;
use std::time::{SystemTime, UNIX_EPOCH};
use tokio::time::Instant;
use crate::config::{get_config, reload_config, VERSION};
use crate::errors::Error;
use crate::messages::*;
use crate::pool::{get_all_pools, get_pool};
use crate::stats::{
get_address_stats, get_client_stats, get_pool_stats, get_server_stats, ClientState, ServerState,
};
use crate::pool::get_all_pools;
use crate::stats::get_stats;
use crate::ClientServerMap;
pub fn generate_server_info_for_admin() -> BytesMut {
let mut server_info = BytesMut::new();
server_info.put(server_parameter_message("application_name", ""));
server_info.put(server_parameter_message("client_encoding", "UTF8"));
server_info.put(server_parameter_message("server_encoding", "UTF8"));
server_info.put(server_parameter_message("server_version", VERSION));
server_info.put(server_parameter_message("DateStyle", "ISO, MDY"));
server_info.put(server_paramater_message("application_name", ""));
server_info.put(server_paramater_message("client_encoding", "UTF8"));
server_info.put(server_paramater_message("server_encoding", "UTF8"));
server_info.put(server_paramater_message("server_version", VERSION));
server_info.put(server_paramater_message("DateStyle", "ISO, MDY"));
server_info
return server_info;
}
/// Handle admin client.
@@ -41,28 +34,19 @@ where
let code = query.get_u8() as char;
if code != 'Q' {
return Err(Error::ProtocolSyncError(format!(
"Invalid code, expected 'Q' but got '{}'",
code
)));
return Err(Error::ProtocolSyncError);
}
let len = query.get_i32() as usize;
let query = String::from_utf8_lossy(&query[..len - 5]).to_string();
let query = String::from_utf8_lossy(&query[..len - 5])
.to_string()
.to_ascii_uppercase();
trace!("Admin query: {}", query);
let query_parts: Vec<&str> = query.trim_end_matches(';').split_whitespace().collect();
match query_parts[0].to_ascii_uppercase().as_str() {
"BAN" => {
trace!("BAN");
ban(stream, query_parts).await
}
"UNBAN" => {
trace!("UNBAN");
unban(stream, query_parts).await
}
match query_parts[0] {
"RELOAD" => {
trace!("RELOAD");
reload(stream, client_server_map).await
@@ -71,23 +55,7 @@ where
trace!("SET");
ignore_set(stream).await
}
"PAUSE" => {
trace!("PAUSE");
pause(stream, query_parts[1]).await
}
"RESUME" => {
trace!("RESUME");
resume(stream, query_parts[1]).await
}
"SHUTDOWN" => {
trace!("SHUTDOWN");
shutdown(stream).await
}
"SHOW" => match query_parts[1].to_ascii_uppercase().as_str() {
"BANS" => {
trace!("SHOW BANS");
show_bans(stream).await
}
"SHOW" => match query_parts[1] {
"CONFIG" => {
trace!("SHOW CONFIG");
show_config(stream).await
@@ -104,14 +72,6 @@ where
trace!("SHOW POOLS");
show_pools(stream).await
}
"CLIENTS" => {
trace!("SHOW CLIENTS");
show_clients(stream).await
}
"SERVERS" => {
trace!("SHOW SERVERS");
show_servers(stream).await
}
"STATS" => {
trace!("SHOW STATS");
show_stats(stream).await
@@ -120,10 +80,6 @@ where
trace!("SHOW VERSION");
show_version(stream).await
}
"USERS" => {
trace!("SHOW USERS");
show_users(stream).await
}
_ => error_response(stream, "Unsupported SHOW query against the admin database").await,
},
_ => error_response(stream, "Unsupported query against the admin database").await,
@@ -135,8 +91,7 @@ async fn show_lists<T>(stream: &mut T) -> Result<(), Error>
where
T: tokio::io::AsyncWrite + std::marker::Unpin,
{
let client_stats = get_client_stats();
let server_stats = get_server_stats();
let stats = get_stats();
let columns = vec![("list", DataType::Text), ("items", DataType::Int4)];
@@ -156,18 +111,18 @@ where
res.put(data_row(&vec!["pools".to_string(), databases.to_string()]));
res.put(data_row(&vec![
"free_clients".to_string(),
client_stats
stats
.keys()
.filter(|client_id| client_stats.get(client_id).unwrap().state == ClientState::Idle)
.count()
.map(|address_id| stats[&address_id]["cl_idle"])
.sum::<i64>()
.to_string(),
]));
res.put(data_row(&vec![
"used_clients".to_string(),
client_stats
stats
.keys()
.filter(|client_id| client_stats.get(client_id).unwrap().state == ClientState::Active)
.count()
.map(|address_id| stats[&address_id]["cl_active"])
.sum::<i64>()
.to_string(),
]));
res.put(data_row(&vec![
@@ -176,18 +131,18 @@ where
]));
res.put(data_row(&vec![
"free_servers".to_string(),
server_stats
stats
.keys()
.filter(|server_id| server_stats.get(server_id).unwrap().state == ServerState::Idle)
.count()
.map(|address_id| stats[&address_id]["sv_idle"])
.sum::<i64>()
.to_string(),
]));
res.put(data_row(&vec![
"used_servers".to_string(),
server_stats
stats
.keys()
.filter(|server_id| server_stats.get(server_id).unwrap().state == ServerState::Active)
.count()
.map(|address_id| stats[&address_id]["sv_active"])
.sum::<i64>()
.to_string(),
]));
res.put(data_row(&vec!["dns_names".to_string(), "0".to_string()]));
@@ -201,7 +156,7 @@ where
res.put_i32(5);
res.put_u8(b'I');
write_all_half(stream, &res).await
write_all_half(stream, res).await
}
/// Show PgCat version.
@@ -212,14 +167,14 @@ where
let mut res = BytesMut::new();
res.put(row_description(&vec![("version", DataType::Text)]));
res.put(data_row(&vec![format!("PgCat {}", VERSION)]));
res.put(data_row(&vec![format!("PgCat {}", VERSION).to_string()]));
res.put(command_complete("SHOW"));
res.put_u8(b'Z');
res.put_i32(5);
res.put_u8(b'I');
write_all_half(stream, &res).await
write_all_half(stream, res).await
}
/// Show utilization of connection pools for each shard and replicas.
@@ -227,12 +182,11 @@ async fn show_pools<T>(stream: &mut T) -> Result<(), Error>
where
T: tokio::io::AsyncWrite + std::marker::Unpin,
{
let all_pool_stats = get_pool_stats();
let stats = get_stats();
let columns = vec![
("database", DataType::Text),
("user", DataType::Text),
("pool_mode", DataType::Text),
("cl_idle", DataType::Numeric),
("cl_active", DataType::Numeric),
("cl_waiting", DataType::Numeric),
@@ -244,33 +198,32 @@ where
("sv_login", DataType::Numeric),
("maxwait", DataType::Numeric),
("maxwait_us", DataType::Numeric),
("pool_mode", DataType::Text),
];
let mut res = BytesMut::new();
res.put(row_description(&columns));
for (user_pool, pool) in get_all_pools() {
let def = HashMap::default();
let pool_stats = all_pool_stats
.get(&(user_pool.db.clone(), user_pool.user.clone()))
.unwrap_or(&def);
for (_, pool) in get_all_pools() {
let pool_config = &pool.settings;
let mut row = vec![
user_pool.db.clone(),
user_pool.user.clone(),
pool_config.pool_mode.to_string(),
];
for column in &columns[3..columns.len()] {
let value = match column.0 {
"maxwait" => (pool_stats.get("maxwait_us").unwrap_or(&0) / 1_000_000).to_string(),
"maxwait_us" => {
(pool_stats.get("maxwait_us").unwrap_or(&0) % 1_000_000).to_string()
for shard in 0..pool.shards() {
for server in 0..pool.servers(shard) {
let address = pool.address(shard, server);
let stats = match stats.get(&address.id) {
Some(stats) => stats.clone(),
None => HashMap::new(),
};
let mut row = vec![address.name(), pool_config.user.username.clone()];
for column in &columns[2..columns.len() - 1] {
let value = stats.get(column.0).unwrap_or(&0).to_string();
row.push(value);
}
_other_values => pool_stats.get(column.0).unwrap_or(&0).to_string(),
};
row.push(value);
row.push(pool_config.pool_mode.to_string());
res.put(data_row(&row));
}
}
res.put(data_row(&row));
}
res.put(command_complete("SHOW"));
@@ -280,7 +233,7 @@ where
res.put_i32(5);
res.put_u8(b'I');
write_all_half(stream, &res).await
write_all_half(stream, res).await
}
/// Show shards and replicas.
@@ -316,8 +269,7 @@ where
for server in 0..pool.servers(shard) {
let address = pool.address(shard, server);
let pool_state = pool.pool_state(shard, server);
let banned = pool.is_banned(address);
let paused = pool.paused();
let banned = pool.is_banned(address, Some(address.role));
res.put(data_row(&vec![
address.name(), // name
@@ -331,11 +283,7 @@ where
pool_config.pool_mode.to_string(), // pool_mode
pool_config.user.pool_size.to_string(), // max_connections
pool_state.connections.to_string(), // current_connections
match paused {
// paused
true => "1".to_string(),
false => "0".to_string(),
},
"0".to_string(), // paused
match banned {
// disabled
true => "1".to_string(),
@@ -352,7 +300,7 @@ where
res.put_i32(5);
res.put_u8(b'I');
write_all_half(stream, &res).await
write_all_half(stream, res).await
}
/// Ignore any SET commands the client sends.
@@ -364,163 +312,6 @@ where
custom_protocol_response_ok(stream, "SET").await
}
/// Bans a host from being used
async fn ban<T>(stream: &mut T, tokens: Vec<&str>) -> Result<(), Error>
where
T: tokio::io::AsyncWrite + std::marker::Unpin,
{
let host = match tokens.get(1) {
Some(host) => host,
None => return error_response(stream, "usage: BAN hostname duration_seconds").await,
};
let duration_seconds = match tokens.get(2) {
Some(duration_seconds) => match duration_seconds.parse::<i64>() {
Ok(duration_seconds) => duration_seconds,
Err(_) => {
return error_response(stream, "duration_seconds must be an integer").await;
}
},
None => return error_response(stream, "usage: BAN hostname duration_seconds").await,
};
if duration_seconds <= 0 {
return error_response(stream, "duration_seconds must be >= 0").await;
}
let columns = vec![
("db", DataType::Text),
("user", DataType::Text),
("role", DataType::Text),
("host", DataType::Text),
];
let mut res = BytesMut::new();
res.put(row_description(&columns));
for (id, pool) in get_all_pools().iter() {
for address in pool.get_addresses_from_host(host) {
if !pool.is_banned(&address) {
pool.ban(&address, BanReason::AdminBan(duration_seconds), -1);
res.put(data_row(&vec![
id.db.clone(),
id.user.clone(),
address.role.to_string(),
address.host,
]));
}
}
}
res.put(command_complete("BAN"));
// ReadyForQuery
res.put_u8(b'Z');
res.put_i32(5);
res.put_u8(b'I');
write_all_half(stream, &res).await
}
/// Clear a host for use
async fn unban<T>(stream: &mut T, tokens: Vec<&str>) -> Result<(), Error>
where
T: tokio::io::AsyncWrite + std::marker::Unpin,
{
let host = match tokens.get(1) {
Some(host) => host,
None => return error_response(stream, "UNBAN command requires a hostname to unban").await,
};
let columns = vec![
("db", DataType::Text),
("user", DataType::Text),
("role", DataType::Text),
("host", DataType::Text),
];
let mut res = BytesMut::new();
res.put(row_description(&columns));
for (id, pool) in get_all_pools().iter() {
for address in pool.get_addresses_from_host(host) {
if pool.is_banned(&address) {
pool.unban(&address);
res.put(data_row(&vec![
id.db.clone(),
id.user.clone(),
address.role.to_string(),
address.host,
]));
}
}
}
res.put(command_complete("UNBAN"));
// ReadyForQuery
res.put_u8(b'Z');
res.put_i32(5);
res.put_u8(b'I');
write_all_half(stream, &res).await
}
/// Shows all the bans
async fn show_bans<T>(stream: &mut T) -> Result<(), Error>
where
T: tokio::io::AsyncWrite + std::marker::Unpin,
{
let columns = vec![
("db", DataType::Text),
("user", DataType::Text),
("role", DataType::Text),
("host", DataType::Text),
("reason", DataType::Text),
("ban_time", DataType::Text),
("ban_duration_seconds", DataType::Text),
("ban_remaining_seconds", DataType::Text),
];
let mut res = BytesMut::new();
res.put(row_description(&columns));
// The block should be pretty quick so we cache the time outside
let now = SystemTime::now()
.duration_since(UNIX_EPOCH)
.expect("Time went backwards")
.as_secs() as i64;
for (id, pool) in get_all_pools().iter() {
for (address, (ban_reason, ban_time)) in pool.get_bans().iter() {
let ban_duration = match ban_reason {
BanReason::AdminBan(duration) => *duration,
_ => pool.settings.ban_time,
};
let remaining = ban_duration - (now - ban_time.timestamp());
if remaining <= 0 {
continue;
}
res.put(data_row(&vec![
id.db.clone(),
id.user.clone(),
address.role.to_string(),
address.host.clone(),
format!("{:?}", ban_reason),
ban_time.to_string(),
ban_duration.to_string(),
remaining.to_string(),
]));
}
}
res.put(command_complete("SHOW BANS"));
// ReadyForQuery
res.put_u8(b'Z');
res.put_i32(5);
res.put_u8(b'I');
write_all_half(stream, &res).await
}
/// Reload the configuration file without restarting the process.
async fn reload<T>(stream: &mut T, client_server_map: ClientServerMap) -> Result<(), Error>
where
@@ -541,7 +332,7 @@ where
res.put_i32(5);
res.put_u8(b'I');
write_all_half(stream, &res).await
write_all_half(stream, res).await
}
/// Shows current configuration.
@@ -587,7 +378,7 @@ where
res.put_i32(5);
res.put_u8(b'I');
write_all_half(stream, &res).await
write_all_half(stream, res).await
}
/// Show shard and replicas statistics.
@@ -596,7 +387,6 @@ where
T: tokio::io::AsyncWrite + std::marker::Unpin,
{
let columns = vec![
("instance", DataType::Text),
("database", DataType::Text),
("user", DataType::Text),
("total_xact_count", DataType::Numeric),
@@ -606,32 +396,32 @@ where
("total_xact_time", DataType::Numeric),
("total_query_time", DataType::Numeric),
("total_wait_time", DataType::Numeric),
("total_errors", DataType::Numeric),
("avg_xact_count", DataType::Numeric),
("avg_query_count", DataType::Numeric),
("avg_recv", DataType::Numeric),
("avg_sent", DataType::Numeric),
("avg_errors", DataType::Numeric),
("avg_xact_time", DataType::Numeric),
("avg_query_time", DataType::Numeric),
("avg_wait_time", DataType::Numeric),
];
let all_stats = get_address_stats();
let stats = get_stats();
let mut res = BytesMut::new();
res.put(row_description(&columns));
for (user_pool, pool) in get_all_pools() {
for ((_db_name, username), pool) in get_all_pools() {
for shard in 0..pool.shards() {
for server in 0..pool.servers(shard) {
let address = pool.address(shard, server);
let stats = match all_stats.get(&address.id) {
let stats = match stats.get(&address.id) {
Some(stats) => stats.clone(),
None => HashMap::new(),
};
let mut row = vec![address.name(), user_pool.db.clone(), user_pool.user.clone()];
for column in &columns[3..] {
let mut row = vec![address.name()];
row.push(username.clone());
for column in &columns[2..] {
row.push(stats.get(column.0).unwrap_or(&0).to_string());
}
@@ -647,260 +437,5 @@ where
res.put_i32(5);
res.put_u8(b'I');
write_all_half(stream, &res).await
}
/// Show currently connected clients
async fn show_clients<T>(stream: &mut T) -> Result<(), Error>
where
T: tokio::io::AsyncWrite + std::marker::Unpin,
{
let columns = vec![
("client_id", DataType::Text),
("database", DataType::Text),
("user", DataType::Text),
("application_name", DataType::Text),
("state", DataType::Text),
("transaction_count", DataType::Numeric),
("query_count", DataType::Numeric),
("error_count", DataType::Numeric),
("age_seconds", DataType::Numeric),
];
let new_map = get_client_stats();
let mut res = BytesMut::new();
res.put(row_description(&columns));
for (_, client) in new_map {
let row = vec![
format!("{:#010X}", client.client_id),
client.pool_name,
client.username,
client.application_name.clone(),
client.state.to_string(),
client.transaction_count.to_string(),
client.query_count.to_string(),
client.error_count.to_string(),
Instant::now()
.duration_since(client.connect_time)
.as_secs()
.to_string(),
];
res.put(data_row(&row));
}
res.put(command_complete("SHOW"));
// ReadyForQuery
res.put_u8(b'Z');
res.put_i32(5);
res.put_u8(b'I');
write_all_half(stream, &res).await
}
/// Show currently connected servers
async fn show_servers<T>(stream: &mut T) -> Result<(), Error>
where
T: tokio::io::AsyncWrite + std::marker::Unpin,
{
let columns = vec![
("server_id", DataType::Text),
("database_name", DataType::Text),
("user", DataType::Text),
("address_id", DataType::Text),
("application_name", DataType::Text),
("state", DataType::Text),
("transaction_count", DataType::Numeric),
("query_count", DataType::Numeric),
("bytes_sent", DataType::Numeric),
("bytes_received", DataType::Numeric),
("age_seconds", DataType::Numeric),
];
let new_map = get_server_stats();
let mut res = BytesMut::new();
res.put(row_description(&columns));
for (_, server) in new_map {
let row = vec![
format!("{:#010X}", server.server_id),
server.pool_name,
server.username,
server.address_name,
server.application_name,
server.state.to_string(),
server.transaction_count.to_string(),
server.query_count.to_string(),
server.bytes_sent.to_string(),
server.bytes_received.to_string(),
Instant::now()
.duration_since(server.connect_time)
.as_secs()
.to_string(),
];
res.put(data_row(&row));
}
res.put(command_complete("SHOW"));
// ReadyForQuery
res.put_u8(b'Z');
res.put_i32(5);
res.put_u8(b'I');
write_all_half(stream, &res).await
}
/// Pause a pool. It won't pass any more queries to the backends.
async fn pause<T>(stream: &mut T, query: &str) -> Result<(), Error>
where
T: tokio::io::AsyncWrite + std::marker::Unpin,
{
let parts: Vec<&str> = query.split(",").map(|part| part.trim()).collect();
if parts.len() != 2 {
error_response(
stream,
"PAUSE requires a database and a user, e.g. PAUSE my_db,my_user",
)
.await
} else {
let database = parts[0];
let user = parts[1];
match get_pool(database, user) {
Some(pool) => {
pool.pause();
let mut res = BytesMut::new();
res.put(command_complete(&format!("PAUSE {},{}", database, user)));
// ReadyForQuery
res.put_u8(b'Z');
res.put_i32(5);
res.put_u8(b'I');
write_all_half(stream, &res).await
}
None => {
error_response(
stream,
&format!(
"No pool configured for database: {}, user: {}",
database, user
),
)
.await
}
}
}
}
/// Resume a pool. Queries are allowed again.
async fn resume<T>(stream: &mut T, query: &str) -> Result<(), Error>
where
T: tokio::io::AsyncWrite + std::marker::Unpin,
{
let parts: Vec<&str> = query.split(",").map(|part| part.trim()).collect();
if parts.len() != 2 {
error_response(
stream,
"RESUME requires a database and a user, e.g. RESUME my_db,my_user",
)
.await
} else {
let database = parts[0];
let user = parts[1];
match get_pool(database, user) {
Some(pool) => {
pool.resume();
let mut res = BytesMut::new();
res.put(command_complete(&format!("RESUME {},{}", database, user)));
// ReadyForQuery
res.put_u8(b'Z');
res.put_i32(5);
res.put_u8(b'I');
write_all_half(stream, &res).await
}
None => {
error_response(
stream,
&format!(
"No pool configured for database: {}, user: {}",
database, user
),
)
.await
}
}
}
}
/// Send response packets for shutdown.
async fn shutdown<T>(stream: &mut T) -> Result<(), Error>
where
T: tokio::io::AsyncWrite + std::marker::Unpin,
{
let mut res = BytesMut::new();
res.put(row_description(&vec![("success", DataType::Text)]));
let mut shutdown_success = "t";
let pid = std::process::id();
if signal::kill(Pid::from_raw(pid.try_into().unwrap()), Signal::SIGINT).is_err() {
error!("Unable to send SIGINT to PID: {}", pid);
shutdown_success = "f";
}
res.put(data_row(&vec![shutdown_success.to_string()]));
res.put(command_complete("SHUTDOWN"));
res.put_u8(b'Z');
res.put_i32(5);
res.put_u8(b'I');
write_all_half(stream, &res).await
}
/// Show Users.
async fn show_users<T>(stream: &mut T) -> Result<(), Error>
where
T: tokio::io::AsyncWrite + std::marker::Unpin,
{
let mut res = BytesMut::new();
res.put(row_description(&vec![
("name", DataType::Text),
("pool_mode", DataType::Text),
]));
for (user_pool, pool) in get_all_pools() {
let pool_config = &pool.settings;
res.put(data_row(&vec![
user_pool.user.clone(),
pool_config.pool_mode.to_string(),
]));
}
res.put(command_complete("SHOW"));
res.put_u8(b'Z');
res.put_i32(5);
res.put_u8(b'I');
write_all_half(stream, &res).await
write_all_half(stream, res).await
}

File diff suppressed because it is too large Load Diff

View File

@@ -2,20 +2,18 @@
use arc_swap::ArcSwap;
use log::{error, info};
use once_cell::sync::Lazy;
use regex::Regex;
use serde_derive::{Deserialize, Serialize};
use std::collections::hash_map::DefaultHasher;
use std::collections::{BTreeMap, HashMap, HashSet};
use std::hash::{Hash, Hasher};
use std::collections::{HashMap, HashSet};
use std::hash::Hash;
use std::path::Path;
use std::sync::Arc;
use tokio::fs::File;
use tokio::io::AsyncReadExt;
use toml;
use crate::errors::Error;
use crate::pool::{ClientServerMap, ConnectionPool};
use crate::sharding::ShardingFunction;
use crate::tls::{load_certs, load_keys};
use crate::{ClientServerMap, ConnectionPool};
pub const VERSION: &str = env!("CARGO_PKG_VERSION");
@@ -25,12 +23,8 @@ static CONFIG: Lazy<ArcSwap<Config>> = Lazy::new(|| ArcSwap::from_pointee(Config
/// Server role: primary or replica.
#[derive(Clone, PartialEq, Serialize, Deserialize, Hash, std::cmp::Eq, Debug, Copy)]
pub enum Role {
#[serde(alias = "primary", alias = "Primary")]
Primary,
#[serde(alias = "replica", alias = "Replica")]
Replica,
#[serde(alias = "mirror", alias = "Mirror")]
Mirror,
}
impl ToString for Role {
@@ -38,7 +32,6 @@ impl ToString for Role {
match *self {
Role::Primary => "primary".to_string(),
Role::Replica => "replica".to_string(),
Role::Mirror => "mirror".to_string(),
}
}
}
@@ -79,6 +72,9 @@ pub struct Address {
/// The name of the Postgres database.
pub database: String,
/// Default search_path.
pub search_path: Option<String>,
/// Server role: replica, primary.
pub role: Role,
@@ -93,9 +89,6 @@ pub struct Address {
/// The name of this pool (i.e. database name visible to the client).
pub pool_name: String,
/// List of addresses to receive mirrored traffic.
pub mirrors: Vec<Address>,
}
impl Default for Address {
@@ -108,10 +101,10 @@ impl Default for Address {
address_index: 0,
replica_number: 0,
database: String::from("database"),
search_path: None,
role: Role::Replica,
username: String::from("username"),
pool_name: String::from("pool_name"),
mirrors: Vec::new(),
}
}
}
@@ -121,25 +114,21 @@ impl Address {
pub fn name(&self) -> String {
match self.role {
Role::Primary => format!("{}_shard_{}_primary", self.pool_name, self.shard),
Role::Replica => format!(
"{}_shard_{}_replica_{}",
self.pool_name, self.shard, self.replica_number
),
Role::Mirror => format!(
"{}_shard_{}_mirror_{}",
self.pool_name, self.shard, self.replica_number
),
}
}
}
/// PostgreSQL user.
#[derive(Clone, PartialEq, Hash, Eq, Serialize, Deserialize, Debug)]
#[derive(Clone, PartialEq, Hash, std::cmp::Eq, Serialize, Deserialize, Debug)]
pub struct User {
pub username: String,
pub password: String,
pub pool_size: u32,
#[serde(default)] // 0
pub statement_timeout: u64,
}
@@ -157,130 +146,34 @@ impl Default for User {
/// General configuration.
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq)]
pub struct General {
#[serde(default = "General::default_host")]
pub host: String,
#[serde(default = "General::default_port")]
pub port: i16,
pub enable_prometheus_exporter: Option<bool>,
pub prometheus_exporter_port: i16,
#[serde(default = "General::default_connect_timeout")]
pub connect_timeout: u64,
#[serde(default = "General::default_idle_timeout")]
pub idle_timeout: u64,
#[serde(default = "General::default_tcp_keepalives_idle")]
pub tcp_keepalives_idle: u64,
#[serde(default = "General::default_tcp_keepalives_count")]
pub tcp_keepalives_count: u32,
#[serde(default = "General::default_tcp_keepalives_interval")]
pub tcp_keepalives_interval: u64,
#[serde(default)] // False
pub log_client_connections: bool,
#[serde(default)] // False
pub log_client_disconnections: bool,
#[serde(default = "General::default_shutdown_timeout")]
pub shutdown_timeout: u64,
#[serde(default = "General::default_healthcheck_timeout")]
pub healthcheck_timeout: u64,
#[serde(default = "General::default_healthcheck_delay")]
pub shutdown_timeout: u64,
pub healthcheck_delay: u64,
#[serde(default = "General::default_ban_time")]
pub ban_time: i64,
#[serde(default = "General::default_worker_threads")]
pub worker_threads: usize,
#[serde(default)] // False
pub autoreload: bool,
pub tls_certificate: Option<String>,
pub tls_private_key: Option<String>,
pub admin_username: String,
pub admin_password: String,
}
impl General {
pub fn default_host() -> String {
"0.0.0.0".into()
}
pub fn default_port() -> i16 {
5432
}
pub fn default_connect_timeout() -> u64 {
1000
}
// These keepalive defaults should detect a dead connection within 30 seconds.
// Tokio defaults to disabling keepalives which keeps dead connections around indefinitely.
// This can lead to permenant server pool exhaustion
pub fn default_tcp_keepalives_idle() -> u64 {
5 // 5 seconds
}
pub fn default_tcp_keepalives_count() -> u32 {
5 // 5 time
}
pub fn default_tcp_keepalives_interval() -> u64 {
5 // 5 seconds
}
pub fn default_idle_timeout() -> u64 {
60000 // 10 minutes
}
pub fn default_shutdown_timeout() -> u64 {
60000
}
pub fn default_healthcheck_timeout() -> u64 {
1000
}
pub fn default_healthcheck_delay() -> u64 {
30000
}
pub fn default_ban_time() -> i64 {
60
}
pub fn default_worker_threads() -> usize {
4
}
}
impl Default for General {
fn default() -> General {
General {
host: Self::default_host(),
port: Self::default_port(),
host: String::from("localhost"),
port: 5432,
enable_prometheus_exporter: Some(false),
prometheus_exporter_port: 9930,
connect_timeout: General::default_connect_timeout(),
idle_timeout: General::default_idle_timeout(),
shutdown_timeout: Self::default_shutdown_timeout(),
healthcheck_timeout: Self::default_healthcheck_timeout(),
healthcheck_delay: Self::default_healthcheck_delay(),
ban_time: Self::default_ban_time(),
worker_threads: Self::default_worker_threads(),
tcp_keepalives_idle: Self::default_tcp_keepalives_idle(),
tcp_keepalives_count: Self::default_tcp_keepalives_count(),
tcp_keepalives_interval: Self::default_tcp_keepalives_interval(),
log_client_connections: false,
log_client_disconnections: false,
connect_timeout: 5000,
healthcheck_timeout: 1000,
shutdown_timeout: 60000,
healthcheck_delay: 30000,
ban_time: 60,
autoreload: false,
tls_certificate: None,
tls_private_key: None,
@@ -289,259 +182,52 @@ impl Default for General {
}
}
}
/// Pool mode:
/// - transaction: server serves one transaction,
/// - session: server is attached to the client.
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Eq, Copy, Hash)]
pub enum PoolMode {
#[serde(alias = "transaction", alias = "Transaction")]
Transaction,
#[serde(alias = "session", alias = "Session")]
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 LoadBalancingMode {
#[serde(alias = "random", alias = "Random")]
Random,
#[serde(alias = "loc", alias = "LOC", alias = "least_outstanding_connections")]
LeastOutstandingConnections,
}
impl ToString for LoadBalancingMode {
fn to_string(&self) -> String {
match *self {
LoadBalancingMode::Random => "random".to_string(),
LoadBalancingMode::LeastOutstandingConnections => {
"least_outstanding_connections".to_string()
}
}
}
}
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Eq, Hash)]
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq)]
pub struct Pool {
#[serde(default = "Pool::default_pool_mode")]
pub pool_mode: PoolMode,
#[serde(default = "Pool::default_load_balancing_mode")]
pub load_balancing_mode: LoadBalancingMode,
pub pool_mode: String,
pub default_role: String,
#[serde(default)] // False
pub query_parser_enabled: bool,
#[serde(default)] // False
pub primary_reads_enabled: bool,
pub connect_timeout: Option<u64>,
pub idle_timeout: Option<u64>,
pub sharding_function: ShardingFunction,
#[serde(default = "Pool::default_automatic_sharding_key")]
pub automatic_sharding_key: Option<String>,
pub sharding_key_regex: Option<String>,
pub shard_id_regex: Option<String>,
pub regex_search_limit: Option<usize>,
pub shards: BTreeMap<String, Shard>,
pub users: BTreeMap<String, User>,
// Note, don't put simple fields below these configs. There's a compatability issue with TOML that makes it
// incompatible to have simple fields in TOML after complex objects. See
// https://users.rust-lang.org/t/why-toml-to-string-get-error-valueaftertable/85903
pub sharding_function: String,
pub shards: HashMap<String, Shard>,
pub users: HashMap<String, User>,
}
impl Pool {
pub fn hash_value(&self) -> u64 {
let mut s = DefaultHasher::new();
self.hash(&mut s);
s.finish()
}
pub fn default_pool_mode() -> PoolMode {
PoolMode::Transaction
}
pub fn default_load_balancing_mode() -> LoadBalancingMode {
LoadBalancingMode::Random
}
pub fn default_automatic_sharding_key() -> Option<String> {
None
}
pub fn validate(&mut self) -> Result<(), Error> {
match self.default_role.as_ref() {
"any" => (),
"primary" => (),
"replica" => (),
other => {
error!(
"Query router default_role must be 'primary', 'replica', or 'any', got: '{}'",
other
);
return Err(Error::BadConfig);
}
};
for (shard_idx, shard) in &self.shards {
match shard_idx.parse::<usize>() {
Ok(_) => (),
Err(_) => {
error!(
"Shard '{}' is not a valid number, shards must be numbered starting at 0",
shard_idx
);
return Err(Error::BadConfig);
}
};
shard.validate()?;
}
for (option, name) in [
(&self.shard_id_regex, "shard_id_regex"),
(&self.sharding_key_regex, "sharding_key_regex"),
] {
if let Some(regex) = option {
if let Err(parse_err) = Regex::new(regex.as_str()) {
error!("{} is not a valid Regex: {}", name, parse_err);
return Err(Error::BadConfig);
}
}
}
self.automatic_sharding_key = match &self.automatic_sharding_key {
Some(key) => {
// No quotes in the key so we don't have to compare quoted
// to unquoted idents.
let key = key.replace("\"", "");
if key.split(".").count() != 2 {
error!(
"automatic_sharding_key '{}' must be fully qualified, e.g. t.{}`",
key, key
);
return Err(Error::BadConfig);
}
Some(key)
}
None => None,
};
Ok(())
}
}
impl Default for Pool {
fn default() -> Pool {
Pool {
pool_mode: Self::default_pool_mode(),
load_balancing_mode: Self::default_load_balancing_mode(),
shards: BTreeMap::from([(String::from("1"), Shard::default())]),
users: BTreeMap::default(),
pool_mode: String::from("transaction"),
shards: HashMap::from([(String::from("1"), Shard::default())]),
users: HashMap::default(),
default_role: String::from("any"),
query_parser_enabled: false,
primary_reads_enabled: false,
sharding_function: ShardingFunction::PgBigintHash,
automatic_sharding_key: None,
connect_timeout: None,
idle_timeout: None,
sharding_key_regex: None,
shard_id_regex: None,
regex_search_limit: Some(1000),
primary_reads_enabled: true,
sharding_function: "pg_bigint_hash".to_string(),
}
}
}
#[derive(Clone, PartialEq, Serialize, Deserialize, Debug, Hash, Eq)]
pub struct ServerConfig {
pub host: String,
pub port: u16,
pub role: Role,
}
#[derive(Clone, PartialEq, Serialize, Deserialize, Debug, Hash, Eq)]
pub struct MirrorServerConfig {
pub host: String,
pub port: u16,
pub mirroring_target_index: usize,
}
/// Shard configuration.
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Hash, Eq)]
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq)]
pub struct Shard {
pub database: String,
pub mirrors: Option<Vec<MirrorServerConfig>>,
pub servers: Vec<ServerConfig>,
}
impl Shard {
pub fn validate(&self) -> Result<(), Error> {
// We use addresses as unique identifiers,
// let's make sure they are unique in the config as well.
let mut dup_check = HashSet::new();
let mut primary_count = 0;
if self.servers.is_empty() {
error!("Shard {} has no servers configured", self.database);
return Err(Error::BadConfig);
}
for server in &self.servers {
dup_check.insert(server);
// Check that we define only zero or one primary.
if server.role == Role::Primary {
primary_count += 1
}
}
if primary_count > 1 {
error!(
"Shard {} has more than on primary configured",
self.database
);
return Err(Error::BadConfig);
}
if dup_check.len() != self.servers.len() {
error!("Shard {} contains duplicate server configs", self.database);
return Err(Error::BadConfig);
}
Ok(())
}
pub search_path: Option<String>,
pub servers: Vec<(String, u16, String)>,
}
impl Default for Shard {
fn default() -> Shard {
Shard {
servers: vec![ServerConfig {
host: String::from("localhost"),
port: 5432,
role: Role::Primary,
}],
mirrors: None,
servers: vec![(String::from("localhost"), 5432, String::from("primary"))],
search_path: None,
database: String::from("postgres"),
}
}
}
fn default_path() -> String {
String::from("pgcat.toml")
}
/// Configuration wrapper.
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq)]
pub struct Config {
@@ -549,30 +235,24 @@ pub struct Config {
// so we should always put simple fields before nested fields
// in all serializable structs to avoid ValueAfterTable errors
// These errors occur when the toml serializer is about to produce
// ambiguous toml structure like the one below
// ambigous toml structure like the one below
// [main]
// field1_under_main = 1
// field2_under_main = 2
// [main.subconf]
// field1_under_subconf = 1
// field3_under_main = 3 # This field will be interpreted as being under subconf and not under main
#[serde(default = "Config::default_path")]
#[serde(default = "default_path")]
pub path: String,
pub general: General,
pub pools: HashMap<String, Pool>,
}
impl Config {
pub fn default_path() -> String {
String::from("pgcat.toml")
}
}
impl Default for Config {
fn default() -> Config {
Config {
path: Self::default_path(),
path: String::from("pgcat.toml"),
general: General::default(),
pools: HashMap::default(),
}
@@ -588,11 +268,7 @@ impl From<&Config> for std::collections::HashMap<String, String> {
[
(
format!("pools.{}.pool_mode", pool_name),
pool.pool_mode.to_string(),
),
(
format!("pools.{}.load_balancing_mode", pool_name),
pool.load_balancing_mode.to_string(),
pool.pool_mode.clone(),
),
(
format!("pools.{}.primary_reads_enabled", pool_name),
@@ -608,7 +284,7 @@ impl From<&Config> for std::collections::HashMap<String, String> {
),
(
format!("pools.{}.sharding_function", pool_name),
pool.sharding_function.to_string(),
pool.sharding_function.clone(),
),
(
format!("pools.{:?}.shard_count", pool_name),
@@ -638,10 +314,6 @@ impl From<&Config> for std::collections::HashMap<String, String> {
"connect_timeout".to_string(),
config.general.connect_timeout.to_string(),
),
(
"idle_timeout".to_string(),
config.general.idle_timeout.to_string(),
),
(
"healthcheck_timeout".to_string(),
config.general.healthcheck_timeout.to_string(),
@@ -666,21 +338,11 @@ impl Config {
/// Print current configuration.
pub fn show(&self) {
info!("Ban time: {}s", self.general.ban_time);
info!("Worker threads: {}", self.general.worker_threads);
info!(
"Healthcheck timeout: {}ms",
self.general.healthcheck_timeout
);
info!("Connection timeout: {}ms", self.general.connect_timeout);
info!("Idle timeout: {}ms", self.general.idle_timeout);
info!(
"Log client connections: {}",
self.general.log_client_connections
);
info!(
"Log client disconnections: {}",
self.general.log_client_disconnections
);
info!("Shutdown timeout: {}ms", self.general.shutdown_timeout);
info!("Healthcheck delay: {}ms", self.general.healthcheck_delay);
match self.general.tls_certificate.clone() {
@@ -714,31 +376,10 @@ impl Config {
.sum::<u32>()
.to_string()
);
info!(
"[pool: {}] Pool mode: {:?}",
pool_name, pool_config.pool_mode
);
info!(
"[pool: {}] Load Balancing mode: {:?}",
pool_name, pool_config.load_balancing_mode
);
let connect_timeout = match pool_config.connect_timeout {
Some(connect_timeout) => connect_timeout,
None => self.general.connect_timeout,
};
info!(
"[pool: {}] Connection timeout: {}ms",
pool_name, connect_timeout
);
let idle_timeout = match pool_config.idle_timeout {
Some(idle_timeout) => idle_timeout,
None => self.general.idle_timeout,
};
info!("[pool: {}] Idle timeout: {}ms", pool_name, idle_timeout);
info!("[pool: {}] Pool mode: {}", pool_name, pool_config.pool_mode);
info!(
"[pool: {}] Sharding function: {}",
pool_name,
pool_config.sharding_function.to_string()
pool_name, pool_config.sharding_function
);
info!(
"[pool: {}] Primary reads: {}",
@@ -771,45 +412,6 @@ impl Config {
}
}
}
pub fn validate(&mut self) -> Result<(), Error> {
// Validate TLS!
match self.general.tls_certificate.clone() {
Some(tls_certificate) => {
match load_certs(Path::new(&tls_certificate)) {
Ok(_) => {
// Cert is okay, but what about the private key?
match self.general.tls_private_key.clone() {
Some(tls_private_key) => match load_keys(Path::new(&tls_private_key)) {
Ok(_) => (),
Err(err) => {
error!("tls_private_key is incorrectly configured: {:?}", err);
return Err(Error::BadConfig);
}
},
None => {
error!("tls_certificate is set, but the tls_private_key is not");
return Err(Error::BadConfig);
}
};
}
Err(err) => {
error!("tls_certificate is incorrectly configured: {:?}", err);
return Err(Error::BadConfig);
}
}
}
None => (),
};
for pool in self.pools.values_mut() {
pool.validate()?;
}
Ok(())
}
}
/// Get a read-only instance of the configuration
@@ -846,7 +448,132 @@ pub async fn parse(path: &str) -> Result<(), Error> {
}
};
config.validate()?;
// Validate TLS!
match config.general.tls_certificate.clone() {
Some(tls_certificate) => {
match load_certs(&Path::new(&tls_certificate)) {
Ok(_) => {
// Cert is okay, but what about the private key?
match config.general.tls_private_key.clone() {
Some(tls_private_key) => match load_keys(&Path::new(&tls_private_key)) {
Ok(_) => (),
Err(err) => {
error!("tls_private_key is incorrectly configured: {:?}", err);
return Err(Error::BadConfig);
}
},
None => {
error!("tls_certificate is set, but the tls_private_key is not");
return Err(Error::BadConfig);
}
};
}
Err(err) => {
error!("tls_certificate is incorrectly configured: {:?}", err);
return Err(Error::BadConfig);
}
}
}
None => (),
};
for (pool_name, pool) in &config.pools {
match pool.sharding_function.as_ref() {
"pg_bigint_hash" => (),
"sha1" => (),
_ => {
error!(
"Supported sharding functions are: 'pg_bigint_hash', 'sha1', got: '{}' in pool {} settings",
pool.sharding_function,
pool_name
);
return Err(Error::BadConfig);
}
};
match pool.default_role.as_ref() {
"any" => (),
"primary" => (),
"replica" => (),
other => {
error!(
"Query router default_role must be 'primary', 'replica', or 'any', got: '{}'",
other
);
return Err(Error::BadConfig);
}
};
match pool.pool_mode.as_ref() {
"transaction" => (),
"session" => (),
other => {
error!(
"pool_mode can be 'session' or 'transaction', got: '{}'",
other
);
return Err(Error::BadConfig);
}
};
for shard in &pool.shards {
// We use addresses as unique identifiers,
// let's make sure they are unique in the config as well.
let mut dup_check = HashSet::new();
let mut primary_count = 0;
match shard.0.parse::<usize>() {
Ok(_) => (),
Err(_) => {
error!(
"Shard '{}' is not a valid number, shards must be numbered starting at 0",
shard.0
);
return Err(Error::BadConfig);
}
};
if shard.1.servers.len() == 0 {
error!("Shard {} has no servers configured", shard.0);
return Err(Error::BadConfig);
}
for server in &shard.1.servers {
dup_check.insert(server);
// Check that we define only zero or one primary.
match server.2.as_ref() {
"primary" => primary_count += 1,
_ => (),
};
// Check role spelling.
match server.2.as_ref() {
"primary" => (),
"replica" => (),
_ => {
error!(
"Shard {} server role must be either 'primary' or 'replica', got: '{}'",
shard.0, server.2
);
return Err(Error::BadConfig);
}
};
}
if primary_count > 1 {
error!("Shard {} has more than on primary configured", &shard.0);
return Err(Error::BadConfig);
}
if dup_check.len() != shard.1.servers.len() {
error!("Shard {} contains duplicate server configs", &shard.0);
return Err(Error::BadConfig);
}
}
}
config.path = path.to_string();
@@ -868,7 +595,7 @@ pub async fn reload_config(client_server_map: ClientServerMap) -> Result<bool, E
let new_config = get_config();
if old_config.pools != new_config.pools {
info!("Pool configuration changed");
info!("Pool configuration changed, re-creating server pools");
ConnectionPool::from_config(client_server_map).await?;
Ok(true)
} else if old_config != new_config {
@@ -889,21 +616,19 @@ mod test {
assert_eq!(get_config().path, "pgcat.toml".to_string());
assert_eq!(get_config().general.ban_time, 60);
assert_eq!(get_config().general.idle_timeout, 30000);
assert_eq!(get_config().pools.len(), 2);
assert_eq!(get_config().pools["sharded_db"].shards.len(), 3);
assert_eq!(get_config().pools["sharded_db"].idle_timeout, Some(40000));
assert_eq!(get_config().pools["simple_db"].shards.len(), 1);
assert_eq!(get_config().pools["sharded_db"].users.len(), 2);
assert_eq!(get_config().pools["simple_db"].users.len(), 1);
assert_eq!(
get_config().pools["sharded_db"].shards["0"].servers[0].host,
get_config().pools["sharded_db"].shards["0"].servers[0].0,
"127.0.0.1"
);
assert_eq!(
get_config().pools["sharded_db"].shards["1"].servers[0].role,
Role::Primary
get_config().pools["sharded_db"].shards["1"].servers[0].2,
"primary"
);
assert_eq!(
get_config().pools["sharded_db"].shards["1"].database,
@@ -921,11 +646,11 @@ mod test {
assert_eq!(get_config().pools["sharded_db"].default_role, "any");
assert_eq!(
get_config().pools["simple_db"].shards["0"].servers[0].host,
get_config().pools["simple_db"].shards["0"].servers[0].0,
"127.0.0.1"
);
assert_eq!(
get_config().pools["simple_db"].shards["0"].servers[0].port,
get_config().pools["simple_db"].shards["0"].servers[0].1,
5432
);
assert_eq!(

View File

@@ -3,16 +3,14 @@
/// Various errors.
#[derive(Debug, PartialEq)]
pub enum Error {
SocketError(String),
SocketError,
ClientBadStartup,
ProtocolSyncError(String),
BadQuery(String),
ProtocolSyncError,
ServerError,
BadConfig,
AllServersDown,
ClientError(String),
ClientError,
TlsError,
StatementTimeout,
ShuttingDown,
ParseBytesError(String),
}

View File

@@ -1,34 +0,0 @@
pub mod config;
pub mod constants;
pub mod errors;
pub mod messages;
pub mod mirrors;
pub mod multi_logger;
pub mod pool;
pub mod scram;
pub mod server;
pub mod sharding;
pub mod stats;
pub mod tls;
/// Format chrono::Duration to be more human-friendly.
///
/// # Arguments
///
/// * `duration` - A duration of time
pub fn format_duration(duration: &chrono::Duration) -> String {
let milliseconds = format!("{:0>3}", duration.num_milliseconds() % 1000);
let seconds = format!("{:0>2}", duration.num_seconds() % 60);
let minutes = format!("{:0>2}", duration.num_minutes() % 60);
let hours = format!("{:0>2}", duration.num_hours() % 24);
let days = duration.num_days().to_string();
format!(
"{}d {}:{}:{}.{}",
days, hours, minutes, seconds, milliseconds
)
}

View File

@@ -37,22 +37,13 @@ extern crate tokio;
extern crate tokio_rustls;
extern crate toml;
#[cfg(not(target_env = "msvc"))]
use jemallocator::Jemalloc;
#[cfg(not(target_env = "msvc"))]
#[global_allocator]
static GLOBAL: Jemalloc = Jemalloc;
use log::{debug, error, info, warn};
use log::{debug, error, info};
use parking_lot::Mutex;
use pgcat::format_duration;
use tokio::net::TcpListener;
#[cfg(not(windows))]
use tokio::signal::unix::{signal as unix_signal, SignalKind};
#[cfg(windows)]
use tokio::signal::windows as win_signal;
use tokio::{runtime::Builder, sync::mpsc};
use tokio::{
signal::unix::{signal as unix_signal, SignalKind},
sync::mpsc,
};
use std::collections::HashMap;
use std::net::SocketAddr;
@@ -66,8 +57,6 @@ mod config;
mod constants;
mod errors;
mod messages;
mod mirrors;
mod multi_logger;
mod pool;
mod prometheus;
mod query_router;
@@ -78,13 +67,14 @@ mod stats;
mod tls;
use crate::config::{get_config, reload_config, VERSION};
use crate::errors::Error;
use crate::pool::{ClientServerMap, ConnectionPool};
use crate::prometheus::start_metric_server;
use crate::stats::{Collector, Reporter, REPORTER};
fn main() -> Result<(), Box<dyn std::error::Error>> {
multi_logger::MultiLogger::init().unwrap();
#[tokio::main(worker_threads = 4)]
async fn main() {
env_logger::init();
info!("Welcome to PgCat! Meow. (Version {})", VERSION);
if !query_router::QueryRouter::setup() {
@@ -100,249 +90,244 @@ fn main() -> Result<(), Box<dyn std::error::Error>> {
String::from("pgcat.toml")
};
// Create a transient runtime for loading the config for the first time.
{
let runtime = Builder::new_multi_thread().worker_threads(1).build()?;
runtime.block_on(async {
match config::parse(&config_file).await {
Ok(_) => (),
Err(err) => {
error!("Config parse error: {:?}", err);
std::process::exit(exitcode::CONFIG);
}
};
});
}
match config::parse(&config_file).await {
Ok(_) => (),
Err(err) => {
error!("Config parse error: {:?}", err);
std::process::exit(exitcode::CONFIG);
}
};
let config = get_config();
// Create the runtime now we know required worker_threads.
let runtime = Builder::new_multi_thread()
.worker_threads(config.general.worker_threads)
.enable_all()
.build()?;
runtime.block_on(async move {
if let Some(true) = config.general.enable_prometheus_exporter {
let http_addr_str = format!(
"{}:{}",
config.general.host, config.general.prometheus_exporter_port
);
let http_addr = match SocketAddr::from_str(&http_addr_str) {
Ok(addr) => addr,
Err(err) => {
error!("Invalid http address: {}", err);
std::process::exit(exitcode::CONFIG);
}
};
tokio::task::spawn(async move {
start_metric_server(http_addr).await;
});
}
let addr = format!("{}:{}", config.general.host, config.general.port);
let listener = match TcpListener::bind(&addr).await {
Ok(sock) => sock,
if let Some(true) = config.general.enable_prometheus_exporter {
let http_addr_str = format!(
"{}:{}",
config.general.host, config.general.prometheus_exporter_port
);
let http_addr = match SocketAddr::from_str(&http_addr_str) {
Ok(addr) => addr,
Err(err) => {
error!("Listener socket error: {:?}", err);
error!("Invalid http address: {}", err);
std::process::exit(exitcode::CONFIG);
}
};
info!("Running on {}", addr);
config.show();
// Tracks which client is connected to which server for query cancellation.
let client_server_map: ClientServerMap = Arc::new(Mutex::new(HashMap::new()));
// Statistics reporting.
let (stats_tx, stats_rx) = mpsc::channel(500_000);
REPORTER.store(Arc::new(Reporter::new(stats_tx.clone())));
// Connection pool that allows to query all shards and replicas.
match ConnectionPool::from_config(client_server_map.clone()).await {
Ok(_) => (),
Err(err) => {
error!("Pool error: {:?}", err);
std::process::exit(exitcode::CONFIG);
}
};
tokio::task::spawn(async move {
let mut stats_collector = Collector::new(stats_rx, stats_tx.clone());
stats_collector.collect().await;
start_metric_server(http_addr).await;
});
}
info!("Config autoreloader: {}", config.general.autoreload);
let addr = format!("{}:{}", config.general.host, config.general.port);
let mut autoreload_interval = tokio::time::interval(tokio::time::Duration::from_millis(15_000));
let autoreload_client_server_map = client_server_map.clone();
let listener = match TcpListener::bind(&addr).await {
Ok(sock) => sock,
Err(err) => {
error!("Listener socket error: {:?}", err);
std::process::exit(exitcode::CONFIG);
}
};
tokio::task::spawn(async move {
loop {
autoreload_interval.tick().await;
info!("Running on {}", addr);
config.show();
// Tracks which client is connected to which server for query cancellation.
let client_server_map: ClientServerMap = Arc::new(Mutex::new(HashMap::new()));
// Statistics reporting.
let (stats_tx, stats_rx) = mpsc::channel(100_000);
REPORTER.store(Arc::new(Reporter::new(stats_tx.clone())));
// Connection pool that allows to query all shards and replicas.
match ConnectionPool::from_config(client_server_map.clone()).await {
Ok(_) => (),
Err(err) => {
error!("Pool error: {:?}", err);
std::process::exit(exitcode::CONFIG);
}
};
tokio::task::spawn(async move {
let mut stats_collector = Collector::new(stats_rx, stats_tx.clone());
stats_collector.collect().await;
});
info!("Config autoreloader: {}", config.general.autoreload);
let mut term_signal = unix_signal(SignalKind::terminate()).unwrap();
let mut interrupt_signal = unix_signal(SignalKind::interrupt()).unwrap();
let mut sighup_signal = unix_signal(SignalKind::hangup()).unwrap();
let mut autoreload_interval = tokio::time::interval(tokio::time::Duration::from_millis(15_000));
let (shutdown_tx, _) = broadcast::channel::<()>(1);
let (drain_tx, mut drain_rx) = mpsc::channel::<i8>(2048);
let (exit_tx, mut exit_rx) = mpsc::channel::<()>(1);
info!("Waiting for clients");
let mut admin_only = false;
let mut total_clients = 0;
loop {
tokio::select! {
// Reload config:
// kill -SIGHUP $(pgrep pgcat)
_ = sighup_signal.recv() => {
info!("Reloading config");
match reload_config(client_server_map.clone()).await {
Ok(_) => (),
Err(_) => (),
};
get_config().show();
},
_ = autoreload_interval.tick() => {
if config.general.autoreload {
info!("Automatically reloading config");
if let Ok(changed) = reload_config(autoreload_client_server_map.clone()).await {
if changed {
get_config().show()
match reload_config(client_server_map.clone()).await {
Ok(changed) => {
if changed {
get_config().show()
}
}
Err(_) => (),
};
}
}
});
},
#[cfg(windows)]
let mut term_signal = win_signal::ctrl_close().unwrap();
#[cfg(windows)]
let mut interrupt_signal = win_signal::ctrl_c().unwrap();
#[cfg(windows)]
let mut sighup_signal = win_signal::ctrl_shutdown().unwrap();
// Initiate graceful shutdown sequence on sig int
_ = interrupt_signal.recv() => {
info!("Got SIGINT, waiting for client connection drain now");
admin_only = true;
#[cfg(not(windows))]
let mut term_signal = unix_signal(SignalKind::terminate()).unwrap();
#[cfg(not(windows))]
let mut interrupt_signal = unix_signal(SignalKind::interrupt()).unwrap();
#[cfg(not(windows))]
let mut sighup_signal = unix_signal(SignalKind::hangup()).unwrap();
let (shutdown_tx, _) = broadcast::channel::<()>(1);
let (drain_tx, mut drain_rx) = mpsc::channel::<i32>(2048);
let (exit_tx, mut exit_rx) = mpsc::channel::<()>(1);
let mut admin_only = false;
let mut total_clients = 0;
// Broadcast that client tasks need to finish
let _ = shutdown_tx.send(());
let exit_tx = exit_tx.clone();
let _ = drain_tx.send(0).await;
info!("Waiting for clients");
tokio::task::spawn(async move {
let mut interval = tokio::time::interval(tokio::time::Duration::from_millis(config.general.shutdown_timeout));
loop {
tokio::select! {
// Reload config:
// kill -SIGHUP $(pgrep pgcat)
_ = sighup_signal.recv() => {
info!("Reloading config");
// First tick fires immediately.
interval.tick().await;
_ = reload_config(client_server_map.clone()).await;
// Second one in the interval time.
interval.tick().await;
get_config().show();
},
// We're done waiting.
error!("Timed out waiting for clients");
// Initiate graceful shutdown sequence on sig int
_ = interrupt_signal.recv() => {
info!("Got SIGINT");
let _ = exit_tx.send(()).await;
});
},
// Don't want this to happen more than once
if admin_only {
_ = term_signal.recv() => break,
new_client = listener.accept() => {
let (socket, addr) = match new_client {
Ok((socket, addr)) => (socket, addr),
Err(err) => {
error!("{:?}", err);
continue;
}
};
admin_only = true;
let shutdown_rx = shutdown_tx.subscribe();
let drain_tx = drain_tx.clone();
let client_server_map = client_server_map.clone();
// Broadcast that client tasks need to finish
let _ = shutdown_tx.send(());
let exit_tx = exit_tx.clone();
let _ = drain_tx.send(0).await;
tokio::task::spawn(async move {
let start = chrono::offset::Utc::now().naive_utc();
tokio::task::spawn(async move {
let mut interval = tokio::time::interval(tokio::time::Duration::from_millis(config.general.shutdown_timeout));
match client::client_entrypoint(
socket,
client_server_map,
shutdown_rx,
drain_tx,
admin_only,
)
.await
{
Ok(()) => {
// First tick fires immediately.
interval.tick().await;
let duration = chrono::offset::Utc::now().naive_utc() - start;
// Second one in the interval time.
interval.tick().await;
info!(
"Client {:?} disconnected, session duration: {}",
addr,
format_duration(&duration)
);
}
// We're done waiting.
error!("Graceful shutdown timed out. {} active clients being closed", total_clients);
let _ = exit_tx.send(()).await;
});
},
_ = term_signal.recv() => {
info!("Got SIGTERM, closing with {} clients active", total_clients);
break;
},
new_client = listener.accept() => {
let (socket, addr) = match new_client {
Ok((socket, addr)) => (socket, addr),
Err(err) => {
error!("{:?}", err);
continue;
match err {
// Don't count the clients we rejected.
Error::ShuttingDown => (),
_ => {
// drain_tx.send(-1).await.unwrap();
}
}
debug!("Client disconnected with error {:?}", err);
}
};
});
}
let shutdown_rx = shutdown_tx.subscribe();
let drain_tx = drain_tx.clone();
let client_server_map = client_server_map.clone();
_ = exit_rx.recv() => {
break;
}
let tls_certificate = config.general.tls_certificate.clone();
client_ping = drain_rx.recv() => {
let client_ping = client_ping.unwrap();
total_clients += client_ping;
tokio::task::spawn(async move {
let start = chrono::offset::Utc::now().naive_utc();
match client::client_entrypoint(
socket,
client_server_map,
shutdown_rx,
drain_tx,
admin_only,
tls_certificate.clone(),
config.general.log_client_connections,
)
.await
{
Ok(()) => {
let duration = chrono::offset::Utc::now().naive_utc() - start;
if config.general.log_client_disconnections {
info!(
"Client {:?} disconnected, session duration: {}",
addr,
format_duration(&duration)
);
} else {
debug!(
"Client {:?} disconnected, session duration: {}",
addr,
format_duration(&duration)
);
}
}
Err(err) => {
match err {
errors::Error::ClientBadStartup => debug!("Client disconnected with error {:?}", err),
_ => warn!("Client disconnected with error {:?}", err),
}
}
};
});
}
_ = exit_rx.recv() => {
break;
}
client_ping = drain_rx.recv() => {
let client_ping = client_ping.unwrap();
total_clients += client_ping;
if total_clients == 0 && admin_only {
let _ = exit_tx.send(()).await;
}
if total_clients == 0 && admin_only {
let _ = exit_tx.send(()).await;
}
}
}
}
info!("Shutting down...");
});
Ok(())
}
/// Format chrono::Duration to be more human-friendly.
///
/// # Arguments
///
/// * `duration` - A duration of time
fn format_duration(duration: &chrono::Duration) -> String {
let seconds = {
let seconds = duration.num_seconds() % 60;
if seconds < 10 {
format!("0{}", seconds)
} else {
format!("{}", seconds)
}
};
let minutes = {
let minutes = duration.num_minutes() % 60;
if minutes < 10 {
format!("0{}", minutes)
} else {
format!("{}", minutes)
}
};
let hours = {
let hours = duration.num_hours() % 24;
if hours < 10 {
format!("0{}", hours)
} else {
format!("{}", hours)
}
};
let days = duration.num_days().to_string();
format!("{}d {}:{}:{}", days, hours, minutes, seconds)
}

View File

@@ -1,18 +1,13 @@
/// Helper functions to send one-off protocol messages
/// and handle TcpStream (TCP socket).
use bytes::{Buf, BufMut, BytesMut};
use log::error;
use md5::{Digest, Md5};
use socket2::{SockRef, TcpKeepalive};
use tokio::io::{AsyncReadExt, AsyncWriteExt};
use tokio::net::TcpStream;
use crate::config::get_config;
use crate::errors::Error;
use std::collections::HashMap;
use std::io::{BufRead, Cursor};
use std::mem;
use std::time::Duration;
/// Postgres data type mappings
/// used in RowDescription ('T') message.
@@ -43,7 +38,7 @@ where
auth_ok.put_i32(8);
auth_ok.put_i32(0);
write_all(stream, auth_ok).await
Ok(write_all(stream, auth_ok).await?)
}
/// Generate md5 password challenge.
@@ -84,7 +79,7 @@ where
key_data.put_i32(backend_id);
key_data.put_i32(secret_key);
write_all(stream, key_data).await
Ok(write_all(stream, key_data).await?)
}
/// Construct a `Q`: Query message.
@@ -93,7 +88,7 @@ pub fn simple_query(query: &str) -> BytesMut {
let query = format!("{}\0", query);
res.put_i32(query.len() as i32 + 4);
res.put_slice(query.as_bytes());
res.put_slice(&query.as_bytes());
res
}
@@ -111,25 +106,41 @@ where
bytes.put_i32(5);
bytes.put_u8(b'I'); // Idle
write_all(stream, bytes).await
Ok(write_all(stream, bytes).await?)
}
/// Send the startup packet the server. We're pretending we're a Pg client.
/// This tells the server which user we are and what database we want.
pub async fn startup(stream: &mut TcpStream, user: &str, database: &str) -> Result<(), Error> {
pub async fn startup(
stream: &mut TcpStream,
user: &str,
database: &str,
search_path: Option<&String>,
) -> Result<(), Error> {
let mut bytes = BytesMut::with_capacity(25);
bytes.put_i32(196608); // Protocol number
// User
bytes.put(&b"user\0"[..]);
bytes.put_slice(user.as_bytes());
bytes.put_slice(&user.as_bytes());
bytes.put_u8(0);
// Database
bytes.put(&b"database\0"[..]);
bytes.put_slice(database.as_bytes());
bytes.put_slice(&database.as_bytes());
bytes.put_u8(0);
// search_path
match search_path {
Some(search_path) => {
bytes.put(&b"options\0"[..]);
bytes.put_slice(&format!("-c search_path={}", search_path).as_bytes());
bytes.put_u8(0);
}
None => (),
};
bytes.put_u8(0); // Null terminator
let len = bytes.len() as i32 + 4i32;
@@ -141,12 +152,7 @@ pub async fn startup(stream: &mut TcpStream, user: &str, database: &str) -> Resu
match stream.write_all(&startup).await {
Ok(_) => Ok(()),
Err(err) => {
return Err(Error::SocketError(format!(
"Error writing startup to server socket - Error: {:?}",
err
)))
}
Err(_) => return Err(Error::SocketError),
}
}
@@ -165,7 +171,7 @@ pub fn parse_params(mut bytes: BytesMut) -> Result<HashMap<String, String>, Erro
c = bytes.get_u8();
}
if !tmp.is_empty() {
if tmp.len() > 0 {
buf.push(tmp.clone());
tmp.clear();
}
@@ -244,7 +250,7 @@ where
message.put_i32(password.len() as i32 + 4);
message.put_slice(&password[..]);
write_all(stream, message).await
Ok(write_all(stream, message).await?)
}
/// Implements a response to our custom `SET SHARDING KEY`
@@ -264,7 +270,7 @@ where
res.put_i32(len);
res.put_slice(&set_complete[..]);
write_all_half(stream, &res).await?;
write_all_half(stream, res).await?;
ready_for_query(stream).await
}
@@ -302,7 +308,7 @@ where
// The short error message.
error.put_u8(b'M');
error.put_slice(format!("{}\0", message).as_bytes());
error.put_slice(&format!("{}\0", message).as_bytes());
// No more fields follow.
error.put_u8(0);
@@ -314,7 +320,7 @@ where
res.put_i32(error.len() as i32 + 4);
res.put(error);
write_all_half(stream, &res).await
Ok(write_all_half(stream, res).await?)
}
pub async fn wrong_password<S>(stream: &mut S, user: &str) -> Result<(), Error>
@@ -337,7 +343,7 @@ where
// The short error message.
error.put_u8(b'M');
error.put_slice(format!("password authentication failed for user \"{}\"\0", user).as_bytes());
error.put_slice(&format!("password authentication failed for user \"{}\"\0", user).as_bytes());
// No more fields follow.
error.put_u8(0);
@@ -376,7 +382,7 @@ where
// CommandComplete
res.put(command_complete("SELECT 1"));
write_all_half(stream, &res).await?;
write_all_half(stream, res).await?;
ready_for_query(stream).await
}
@@ -389,7 +395,7 @@ pub fn row_description(columns: &Vec<(&str, DataType)>) -> BytesMut {
for (name, data_type) in columns {
// Column name
row_desc.put_slice(format!("{}\0", name).as_bytes());
row_desc.put_slice(&format!("{}\0", name).as_bytes());
// Doesn't belong to any table
row_desc.put_i32(0);
@@ -433,7 +439,7 @@ pub fn data_row(row: &Vec<String>) -> BytesMut {
for column in row {
let column = column.as_bytes();
data_row.put_i32(column.len() as i32);
data_row.put_slice(column);
data_row.put_slice(&column);
}
res.put_u8(b'D');
@@ -460,28 +466,18 @@ where
{
match stream.write_all(&buf).await {
Ok(_) => Ok(()),
Err(err) => {
return Err(Error::SocketError(format!(
"Error writing to socket - Error: {:?}",
err
)))
}
Err(_) => return Err(Error::SocketError),
}
}
/// Write all the data in the buffer to the TcpStream, write owned half (see mpsc).
pub async fn write_all_half<S>(stream: &mut S, buf: &BytesMut) -> Result<(), Error>
pub async fn write_all_half<S>(stream: &mut S, buf: BytesMut) -> Result<(), Error>
where
S: tokio::io::AsyncWrite + std::marker::Unpin,
{
match stream.write_all(buf).await {
match stream.write_all(&buf).await {
Ok(_) => Ok(()),
Err(err) => {
return Err(Error::SocketError(format!(
"Error writing to socket - Error: {:?}",
err
)))
}
Err(_) => return Err(Error::SocketError),
}
}
@@ -492,56 +488,31 @@ where
{
let code = match stream.read_u8().await {
Ok(code) => code,
Err(err) => {
return Err(Error::SocketError(format!(
"Error reading message code from socket - Error {:?}",
err
)))
}
Err(_) => return Err(Error::SocketError),
};
let len = match stream.read_i32().await {
Ok(len) => len,
Err(err) => {
return Err(Error::SocketError(format!(
"Error reading message len from socket - Code: {:?}, Error: {:?}",
code, err
)))
}
Err(_) => return Err(Error::SocketError),
};
let mut buf = vec![0u8; len as usize - 4];
match stream.read_exact(&mut buf).await {
Ok(_) => (),
Err(_) => return Err(Error::SocketError),
};
let mut bytes = BytesMut::with_capacity(len as usize + 1);
bytes.put_u8(code);
bytes.put_i32(len);
bytes.resize(bytes.len() + len as usize - mem::size_of::<i32>(), b'0');
let slice_start = mem::size_of::<u8>() + mem::size_of::<i32>();
let slice_end = slice_start + len as usize - mem::size_of::<i32>();
// Avoids a panic
if slice_end < slice_start {
return Err(Error::SocketError(format!(
"Error reading message from socket - Code: {:?} - Length {:?}, Error: {:?}",
code, len, "Unexpected length value for message"
)));
}
match stream.read_exact(&mut bytes[slice_start..slice_end]).await {
Ok(_) => (),
Err(err) => {
return Err(Error::SocketError(format!(
"Error reading message from socket - Code: {:?}, Error: {:?}",
code, err
)))
}
};
bytes.put_slice(&buf);
Ok(bytes)
}
pub fn server_parameter_message(key: &str, value: &str) -> BytesMut {
pub fn server_paramater_message(key: &str, value: &str) -> BytesMut {
let mut server_info = BytesMut::new();
let null_byte_size = 1;
@@ -555,41 +526,5 @@ pub fn server_parameter_message(key: &str, value: &str) -> BytesMut {
server_info.put_slice(value.as_bytes());
server_info.put_bytes(0, 1);
server_info
}
pub fn configure_socket(stream: &TcpStream) {
let sock_ref = SockRef::from(stream);
let conf = get_config();
match sock_ref.set_keepalive(true) {
Ok(_) => {
match sock_ref.set_tcp_keepalive(
&TcpKeepalive::new()
.with_interval(Duration::from_secs(conf.general.tcp_keepalives_interval))
.with_retries(conf.general.tcp_keepalives_count)
.with_time(Duration::from_secs(conf.general.tcp_keepalives_idle)),
) {
Ok(_) => (),
Err(err) => error!("Could not configure socket: {}", err),
}
}
Err(err) => error!("Could not configure socket: {}", err),
}
}
pub trait BytesMutReader {
fn read_string(&mut self) -> Result<String, Error>;
}
impl BytesMutReader for Cursor<&BytesMut> {
/// Should only be used when reading strings from the message protocol.
/// Can be used to read multiple strings from the same message which are separated by the null byte
fn read_string(&mut self) -> Result<String, Error> {
let mut buf = vec![];
match self.read_until(b'\0', &mut buf) {
Ok(_) => Ok(String::from_utf8_lossy(&buf[..buf.len() - 1]).to_string()),
Err(err) => return Err(Error::ParseBytesError(err.to_string())),
}
}
return server_info;
}

View File

@@ -1,179 +0,0 @@
/// A mirrored PostgreSQL client.
/// Packets arrive to us through a channel from the main client and we send them to the server.
use bb8::Pool;
use bytes::{Bytes, BytesMut};
use crate::config::{get_config, Address, Role, User};
use crate::pool::{ClientServerMap, ServerPool};
use crate::stats::get_reporter;
use log::{error, info, trace, warn};
use tokio::sync::mpsc::{channel, Receiver, Sender};
pub struct MirroredClient {
address: Address,
user: User,
database: String,
bytes_rx: Receiver<Bytes>,
disconnect_rx: Receiver<()>,
}
impl MirroredClient {
async fn create_pool(&self) -> Pool<ServerPool> {
let config = get_config();
let default = std::time::Duration::from_millis(10_000).as_millis() as u64;
let (connection_timeout, idle_timeout) = match config.pools.get(&self.address.pool_name) {
Some(cfg) => (
cfg.connect_timeout.unwrap_or(default),
cfg.idle_timeout.unwrap_or(default),
),
None => (default, default),
};
let manager = ServerPool::new(
self.address.clone(),
self.user.clone(),
self.database.as_str(),
ClientServerMap::default(),
get_reporter(),
);
Pool::builder()
.max_size(1)
.connection_timeout(std::time::Duration::from_millis(connection_timeout))
.idle_timeout(Some(std::time::Duration::from_millis(idle_timeout)))
.test_on_check_out(false)
.build(manager)
.await
.unwrap()
}
pub fn start(mut self) {
tokio::spawn(async move {
let pool = self.create_pool().await;
let address = self.address.clone();
loop {
let mut server = match pool.get().await {
Ok(server) => server,
Err(err) => {
error!(
"Failed to get connection from pool, Discarding message {:?}, {:?}",
err,
address.clone()
);
continue;
}
};
tokio::select! {
// Exit channel events
_ = self.disconnect_rx.recv() => {
info!("Got mirror exit signal, exiting {:?}", address.clone());
break;
}
// Incoming data from server (we read to clear the socket buffer and discard the data)
recv_result = server.recv() => {
match recv_result {
Ok(message) => trace!("Received from mirror: {} {:?}", String::from_utf8_lossy(&message[..]), address.clone()),
Err(err) => {
server.mark_bad();
error!("Failed to receive from mirror {:?} {:?}", err, address.clone());
}
}
}
// Messages to send to the server
message = self.bytes_rx.recv() => {
match message {
Some(bytes) => {
match server.send(&BytesMut::from(&bytes[..])).await {
Ok(_) => trace!("Sent to mirror: {} {:?}", String::from_utf8_lossy(&bytes[..]), address.clone()),
Err(err) => {
server.mark_bad();
error!("Failed to send to mirror, Discarding message {:?}, {:?}", err, address.clone())
}
}
}
None => {
info!("Mirror channel closed, exiting {:?}", address.clone());
break;
},
}
}
}
}
});
}
}
pub struct MirroringManager {
pub byte_senders: Vec<Sender<Bytes>>,
pub disconnect_senders: Vec<Sender<()>>,
}
impl MirroringManager {
pub fn from_addresses(
user: User,
database: String,
addresses: Vec<Address>,
) -> MirroringManager {
let mut byte_senders: Vec<Sender<Bytes>> = vec![];
let mut exit_senders: Vec<Sender<()>> = vec![];
addresses.iter().for_each(|mirror| {
let (bytes_tx, bytes_rx) = channel::<Bytes>(10);
let (exit_tx, exit_rx) = channel::<()>(1);
let mut addr = mirror.clone();
addr.role = Role::Mirror;
let client = MirroredClient {
user: user.clone(),
database: database.to_owned(),
address: addr,
bytes_rx,
disconnect_rx: exit_rx,
};
exit_senders.push(exit_tx.clone());
byte_senders.push(bytes_tx.clone());
client.start();
});
Self {
byte_senders: byte_senders,
disconnect_senders: exit_senders,
}
}
pub fn send(self: &mut Self, bytes: &BytesMut) {
// We want to avoid performing an allocation if we won't be able to send the message
// There is a possibility of a race here where we check the capacity and then the channel is
// closed or the capacity is reduced to 0, but mirroring is best effort anyway
if self
.byte_senders
.iter()
.all(|sender| sender.capacity() == 0 || sender.is_closed())
{
return;
}
let immutable_bytes = bytes.clone().freeze();
self.byte_senders.iter_mut().for_each(|sender| {
match sender.try_send(immutable_bytes.clone()) {
Ok(_) => {}
Err(err) => {
warn!("Failed to send bytes to a mirror channel {}", err);
}
}
});
}
pub fn disconnect(self: &mut Self) {
self.disconnect_senders
.iter_mut()
.for_each(|sender| match sender.try_send(()) {
Ok(_) => {}
Err(err) => {
warn!(
"Failed to send disconnect signal to a mirror channel {}",
err
);
}
});
}
}

View File

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

File diff suppressed because it is too large Load Diff

View File

@@ -8,7 +8,7 @@ use std::net::SocketAddr;
use crate::config::Address;
use crate::pool::get_all_pools;
use crate::stats::{get_address_stats, get_pool_stats, get_server_stats, ServerInformation};
use crate::stats::get_stats;
struct MetricHelpType {
help: &'static str,
@@ -19,141 +19,109 @@ struct MetricHelpType {
// counters only increase
// gauges can arbitrarily increase or decrease
static METRIC_HELP_AND_TYPES_LOOKUP: phf::Map<&'static str, MetricHelpType> = phf_map! {
"stats_total_query_count" => MetricHelpType {
"total_query_count" => MetricHelpType {
help: "Number of queries sent by all clients",
ty: "counter",
},
"stats_total_query_time" => MetricHelpType {
"total_query_time" => MetricHelpType {
help: "Total amount of time for queries to execute",
ty: "counter",
},
"stats_total_received" => MetricHelpType {
"total_received" => MetricHelpType {
help: "Number of bytes received from the server",
ty: "counter",
},
"stats_total_sent" => MetricHelpType {
"total_sent" => MetricHelpType {
help: "Number of bytes sent to the server",
ty: "counter",
},
"stats_total_xact_count" => MetricHelpType {
"total_xact_count" => MetricHelpType {
help: "Total number of transactions started by the client",
ty: "counter",
},
"stats_total_xact_time" => MetricHelpType {
"total_xact_time" => MetricHelpType {
help: "Total amount of time for all transactions to execute",
ty: "counter",
},
"stats_total_wait_time" => MetricHelpType {
"total_wait_time" => MetricHelpType {
help: "Total time client waited for a server connection",
ty: "counter",
},
"stats_avg_query_count" => MetricHelpType {
"avg_query_count" => MetricHelpType {
help: "Average of total_query_count every 15 seconds",
ty: "gauge",
},
"stats_avg_query_time" => MetricHelpType {
"avg_query_time" => MetricHelpType {
help: "Average time taken for queries to execute every 15 seconds",
ty: "gauge",
},
"stats_avg_recv" => MetricHelpType {
"avg_recv" => MetricHelpType {
help: "Average of total_received bytes every 15 seconds",
ty: "gauge",
},
"stats_avg_sent" => MetricHelpType {
"avg_sent" => MetricHelpType {
help: "Average of total_sent bytes every 15 seconds",
ty: "gauge",
},
"stats_avg_errors" => MetricHelpType {
help: "Average number of errors every 15 seconds",
ty: "gauge",
},
"stats_avg_xact_count" => MetricHelpType {
"avg_xact_count" => MetricHelpType {
help: "Average of total_xact_count every 15 seconds",
ty: "gauge",
},
"stats_avg_xact_time" => MetricHelpType {
"avg_xact_time" => MetricHelpType {
help: "Average of total_xact_time every 15 seconds",
ty: "gauge",
},
"stats_avg_wait_time" => MetricHelpType {
"avg_wait_time" => MetricHelpType {
help: "Average of total_wait_time every 15 seconds",
ty: "gauge",
},
"pools_maxwait_us" => MetricHelpType {
"maxwait_us" => MetricHelpType {
help: "The time a client waited for a server connection in microseconds",
ty: "gauge",
},
"pools_maxwait" => MetricHelpType {
"maxwait" => MetricHelpType {
help: "The time a client waited for a server connection in seconds",
ty: "gauge",
},
"pools_cl_waiting" => MetricHelpType {
"cl_waiting" => MetricHelpType {
help: "How many clients are waiting for a connection from the pool",
ty: "gauge",
},
"pools_cl_active" => MetricHelpType {
"cl_active" => MetricHelpType {
help: "How many clients are actively communicating with a server",
ty: "gauge",
},
"pools_cl_idle" => MetricHelpType {
"cl_idle" => MetricHelpType {
help: "How many clients are idle",
ty: "gauge",
},
"pools_sv_idle" => MetricHelpType {
"sv_idle" => MetricHelpType {
help: "How many server connections are idle",
ty: "gauge",
},
"pools_sv_active" => MetricHelpType {
"sv_active" => MetricHelpType {
help: "How many server connections are actively communicating with a client",
ty: "gauge",
},
"pools_sv_login" => MetricHelpType {
"sv_login" => MetricHelpType {
help: "How many server connections are currently being created",
ty: "gauge",
},
"pools_sv_tested" => MetricHelpType {
"sv_tested" => MetricHelpType {
help: "How many server connections are currently waiting on a health check to succeed",
ty: "gauge",
},
"servers_bytes_received" => MetricHelpType {
help: "Volume in bytes of network traffic received by server",
ty: "gauge",
},
"servers_bytes_sent" => MetricHelpType {
help: "Volume in bytes of network traffic sent by server",
ty: "gauge",
},
"servers_transaction_count" => MetricHelpType {
help: "Number of transactions executed by server",
ty: "gauge",
},
"servers_query_count" => MetricHelpType {
help: "Number of queries executed by server",
ty: "gauge",
},
"servers_error_count" => MetricHelpType {
help: "Number of errors",
ty: "gauge",
},
"databases_pool_size" => MetricHelpType {
help: "Maximum number of server connections",
ty: "gauge",
},
"databases_current_connections" => MetricHelpType {
help: "Current number of connections for this database",
ty: "gauge",
},
};
struct PrometheusMetric<Value: fmt::Display> {
struct PrometheusMetric {
name: String,
help: String,
ty: String,
labels: HashMap<&'static str, String>,
value: Value,
value: i64,
}
impl<Value: fmt::Display> fmt::Display for PrometheusMetric<Value> {
impl fmt::Display for PrometheusMetric {
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
let formatted_labels = self
.labels
@@ -173,81 +141,50 @@ impl<Value: fmt::Display> fmt::Display for PrometheusMetric<Value> {
}
}
impl<Value: fmt::Display> PrometheusMetric<Value> {
fn from_name<V: fmt::Display>(
name: &str,
value: V,
labels: HashMap<&'static str, String>,
) -> Option<PrometheusMetric<V>> {
impl PrometheusMetric {
fn new(address: &Address, name: &str, value: i64) -> Option<PrometheusMetric> {
let mut labels = HashMap::new();
labels.insert("host", address.host.clone());
labels.insert("shard", address.shard.to_string());
labels.insert("role", address.role.to_string());
labels.insert("database", address.database.to_string());
METRIC_HELP_AND_TYPES_LOOKUP
.get(name)
.map(|metric| PrometheusMetric::<V> {
.map(|metric| PrometheusMetric {
name: name.to_owned(),
help: metric.help.to_owned(),
ty: metric.ty.to_owned(),
value,
labels,
value,
})
}
fn from_database_info(
address: &Address,
name: &str,
value: u32,
) -> Option<PrometheusMetric<u32>> {
let mut labels = HashMap::new();
labels.insert("host", address.host.clone());
labels.insert("shard", address.shard.to_string());
labels.insert("role", address.role.to_string());
labels.insert("pool", address.pool_name.clone());
labels.insert("database", address.database.to_string());
Self::from_name(&format!("databases_{}", name), value, labels)
}
fn from_server_info(
address: &Address,
name: &str,
value: u64,
) -> Option<PrometheusMetric<u64>> {
let mut labels = HashMap::new();
labels.insert("host", address.host.clone());
labels.insert("shard", address.shard.to_string());
labels.insert("role", address.role.to_string());
labels.insert("pool", address.pool_name.clone());
labels.insert("database", address.database.to_string());
Self::from_name(&format!("servers_{}", name), value, labels)
}
fn from_address(address: &Address, name: &str, value: i64) -> Option<PrometheusMetric<i64>> {
let mut labels = HashMap::new();
labels.insert("host", address.host.clone());
labels.insert("shard", address.shard.to_string());
labels.insert("pool", address.pool_name.clone());
labels.insert("role", address.role.to_string());
labels.insert("database", address.database.to_string());
Self::from_name(&format!("stats_{}", name), value, labels)
}
fn from_pool(pool: &(String, String), name: &str, value: i64) -> Option<PrometheusMetric<i64>> {
let mut labels = HashMap::new();
labels.insert("pool", pool.0.clone());
labels.insert("user", pool.1.clone());
Self::from_name(&format!("pools_{}", name), value, labels)
}
}
async fn prometheus_stats(request: Request<Body>) -> Result<Response<Body>, hyper::http::Error> {
match (request.method(), request.uri().path()) {
(&Method::GET, "/metrics") => {
let stats = get_stats();
let mut lines = Vec::new();
push_address_stats(&mut lines);
push_pool_stats(&mut lines);
push_server_stats(&mut lines);
push_database_stats(&mut lines);
for (_, pool) in get_all_pools() {
for shard in 0..pool.shards() {
for server in 0..pool.servers(shard) {
let address = pool.address(shard, server);
if let Some(address_stats) = stats.get(&address.id) {
for (key, value) in address_stats.iter() {
if let Some(prometheus_metric) =
PrometheusMetric::new(address, key, *value)
{
lines.push(prometheus_metric.to_string());
} else {
warn!("Metric {} not implemented for {}", key, address.name());
}
}
}
}
}
}
Response::builder()
.header("content-type", "text/plain; version=0.0.4")
@@ -259,113 +196,10 @@ 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 address_stats: HashMap<usize, HashMap<String, i64>> = get_address_stats();
for (_, pool) in get_all_pools() {
for shard in 0..pool.shards() {
for server in 0..pool.servers(shard) {
let address = pool.address(shard, server);
if let Some(address_stats) = address_stats.get(&address.id) {
for (key, value) in address_stats.iter() {
if let Some(prometheus_metric) =
PrometheusMetric::<i64>::from_address(address, key, *value)
{
lines.push(prometheus_metric.to_string());
} else {
warn!("Metric {} not implemented for {}", key, address.name());
}
}
}
}
}
}
}
// Adds relevant metrics shown in a SHOW POOLS admin command.
fn push_pool_stats(lines: &mut Vec<String>) {
let pool_stats = get_pool_stats();
for (pool, stats) in pool_stats.iter() {
for (name, value) in stats.iter() {
if let Some(prometheus_metric) = PrometheusMetric::<i64>::from_pool(pool, name, *value)
{
lines.push(prometheus_metric.to_string());
} else {
warn!(
"Metric {} not implemented for ({},{})",
name, pool.0, pool.1
);
}
}
}
}
// Adds relevant metrics shown in a SHOW DATABASES admin command.
fn push_database_stats(lines: &mut Vec<String>) {
for (_, pool) in get_all_pools() {
let pool_config = pool.settings.clone();
for shard in 0..pool.shards() {
for server in 0..pool.servers(shard) {
let address = pool.address(shard, server);
let pool_state = pool.pool_state(shard, server);
let metrics = vec![
("pool_size", pool_config.user.pool_size),
("current_connections", pool_state.connections),
];
for (key, value) in metrics {
if let Some(prometheus_metric) =
PrometheusMetric::<u32>::from_database_info(address, key, value)
{
lines.push(prometheus_metric.to_string());
} else {
warn!("Metric {} not implemented for {}", key, address.name());
}
}
}
}
}
}
// Adds relevant metrics shown in a SHOW SERVERS admin command.
fn push_server_stats(lines: &mut Vec<String>) {
let server_stats = get_server_stats();
let mut server_stats_by_addresses = HashMap::<String, ServerInformation>::new();
for (_, info) in server_stats {
server_stats_by_addresses.insert(info.address_name.clone(), info);
}
for (_, pool) in get_all_pools() {
for shard in 0..pool.shards() {
for server in 0..pool.servers(shard) {
let address = pool.address(shard, server);
if let Some(server_info) = server_stats_by_addresses.get(&address.name()) {
let metrics = [
("bytes_received", server_info.bytes_received),
("bytes_sent", server_info.bytes_sent),
("transaction_count", server_info.transaction_count),
("query_count", server_info.query_count),
("error_count", server_info.error_count),
];
for (key, value) in metrics {
if let Some(prometheus_metric) =
PrometheusMetric::<u64>::from_server_info(address, key, value)
{
lines.push(prometheus_metric.to_string());
} else {
warn!("Metric {} not implemented for {}", key, address.name());
}
}
}
}
}
}
}
pub async fn start_metric_server(http_addr: SocketAddr) {
let http_service_factory =
make_service_fn(|_conn| async { Ok::<_, hyper::Error>(service_fn(prometheus_stats)) });
let server = Server::bind(&http_addr).serve(http_service_factory);
let server = Server::bind(&http_addr.into()).serve(http_service_factory);
info!(
"Exposing prometheus metrics on http://{}/metrics.",
http_addr

File diff suppressed because it is too large Load Diff

View File

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

View File

@@ -1,8 +1,7 @@
/// Implementation of the PostgreSQL server (database) protocol.
/// Here we are pretending to the a Postgres client.
use bytes::{Buf, BufMut, BytesMut};
use log::{debug, error, info, trace, warn};
use std::io::Read;
use log::{debug, error, info, trace};
use std::time::SystemTime;
use tokio::io::{AsyncReadExt, BufReader};
use tokio::net::{
@@ -14,15 +13,12 @@ use crate::config::{Address, User};
use crate::constants::*;
use crate::errors::Error;
use crate::messages::*;
use crate::mirrors::MirroringManager;
use crate::pool::ClientServerMap;
use crate::scram::ScramSha256;
use crate::stats::Reporter;
use crate::ClientServerMap;
/// Server state.
pub struct Server {
server_id: i32,
/// Server host, e.g. localhost,
/// port, e.g. 5432, and role, e.g. primary or replica.
address: Address,
@@ -52,9 +48,6 @@ pub struct Server {
/// Is the server broken? We'll remote it from the pool if so.
bad: bool,
/// If server connection requires a DISCARD ALL before checkin
needs_cleanup: bool,
/// Mapping of clients and servers used for query cancellation.
client_server_map: ClientServerMap,
@@ -69,15 +62,12 @@ pub struct Server {
// Last time that a successful server send or response happened
last_activity: SystemTime,
mirror_manager: Option<MirroringManager>,
}
impl Server {
/// Pretend to be the Postgres client and connect to the server given host, port and credentials.
/// Perform the authentication and return the server in a ready for query state.
pub async fn startup(
server_id: i32,
address: &Address,
user: &User,
database: &str,
@@ -89,18 +79,20 @@ impl Server {
Ok(stream) => stream,
Err(err) => {
error!("Could not connect to server: {}", err);
return Err(Error::SocketError(format!(
"Could not connect to server: {}",
err
)));
return Err(Error::SocketError);
}
};
configure_socket(&stream);
trace!("Sending StartupMessage");
// StartupMessage
startup(&mut stream, &user.username, database).await?;
startup(
&mut stream,
&user.username,
database,
address.search_path.as_ref(),
)
.await?;
let mut server_info = BytesMut::new();
let mut process_id: i32 = 0;
@@ -113,12 +105,12 @@ impl Server {
loop {
let code = match stream.read_u8().await {
Ok(code) => code as char,
Err(_) => return Err(Error::SocketError(format!("Error reading message code on server startup {{ username: {:?}, database: {:?} }}", user.username, database))),
Err(_) => return Err(Error::SocketError),
};
let len = match stream.read_i32().await {
Ok(len) => len,
Err(_) => return Err(Error::SocketError(format!("Error reading message len on server startup {{ username: {:?}, database: {:?} }}", user.username, database))),
Err(_) => return Err(Error::SocketError),
};
trace!("Message: {}", code);
@@ -129,7 +121,7 @@ impl Server {
// Determine which kind of authentication is required, if any.
let auth_code = match stream.read_i32().await {
Ok(auth_code) => auth_code,
Err(_) => return Err(Error::SocketError(format!("Error reading auth code on server startup {{ username: {:?}, database: {:?} }}", user.username, database))),
Err(_) => return Err(Error::SocketError),
};
trace!("Auth: {}", auth_code);
@@ -142,7 +134,7 @@ impl Server {
match stream.read_exact(&mut salt).await {
Ok(_) => (),
Err(_) => return Err(Error::SocketError(format!("Error reading salt on server startup {{ username: {:?}, database: {:?} }}", user.username, database))),
Err(_) => return Err(Error::SocketError),
};
md5_password(&mut stream, &user.username, &user.password, &salt[..])
@@ -158,7 +150,7 @@ impl Server {
match stream.read_exact(&mut sasl_auth).await {
Ok(_) => (),
Err(_) => return Err(Error::SocketError(format!("Error reading sasl message on server startup {{ username: {:?}, database: {:?} }}", user.username, database))),
Err(_) => return Err(Error::SocketError),
};
let sasl_type = String::from_utf8_lossy(&sasl_auth[..sasl_len - 2]);
@@ -182,7 +174,7 @@ impl Server {
+ sasl_response.len() as i32, // length of SASL response
);
res.put_slice(format!("{}\0", SCRAM_SHA_256).as_bytes());
res.put_slice(&format!("{}\0", SCRAM_SHA_256).as_bytes()[..]);
res.put_i32(sasl_response.len() as i32);
res.put(sasl_response);
@@ -200,7 +192,7 @@ impl Server {
match stream.read_exact(&mut sasl_data).await {
Ok(_) => (),
Err(_) => return Err(Error::SocketError(format!("Error reading sasl cont message on server startup {{ username: {:?}, database: {:?} }}", user.username, database))),
Err(_) => return Err(Error::SocketError),
};
let msg = BytesMut::from(&sasl_data[..]);
@@ -221,7 +213,7 @@ impl Server {
let mut sasl_final = vec![0u8; len as usize - 8];
match stream.read_exact(&mut sasl_final).await {
Ok(_) => (),
Err(_) => return Err(Error::SocketError(format!("Error reading sasl final message on server startup {{ username: {:?}, database: {:?} }}", user.username, database))),
Err(_) => return Err(Error::SocketError),
};
match scram.finish(&BytesMut::from(&sasl_final[..])) {
@@ -247,7 +239,7 @@ impl Server {
'E' => {
let error_code = match stream.read_u8().await {
Ok(error_code) => error_code,
Err(_) => return Err(Error::SocketError(format!("Error reading error code message on server startup {{ username: {:?}, database: {:?} }}", user.username, database))),
Err(_) => return Err(Error::SocketError),
};
trace!("Error: {}", error_code);
@@ -263,7 +255,7 @@ impl Server {
match stream.read_exact(&mut error).await {
Ok(_) => (),
Err(_) => return Err(Error::SocketError(format!("Error reading error message on server startup {{ username: {:?}, database: {:?} }}", user.username, database))),
Err(_) => return Err(Error::SocketError),
};
// TODO: the error message contains multiple fields; we can decode them and
@@ -282,7 +274,7 @@ impl Server {
match stream.read_exact(&mut param).await {
Ok(_) => (),
Err(_) => return Err(Error::SocketError(format!("Error reading parameter status message on server startup {{ username: {:?}, database: {:?} }}", user.username, database))),
Err(_) => return Err(Error::SocketError),
};
// Save the parameter so we can pass it to the client later.
@@ -299,12 +291,12 @@ impl Server {
// See: <https://www.postgresql.org/docs/12/protocol-message-formats.html>.
process_id = match stream.read_i32().await {
Ok(id) => id,
Err(_) => return Err(Error::SocketError(format!("Error reading process id message on server startup {{ username: {:?}, database: {:?} }}", user.username, database))),
Err(_) => return Err(Error::SocketError),
};
secret_key = match stream.read_i32().await {
Ok(id) => id,
Err(_) => return Err(Error::SocketError(format!("Error reading secret key message on server startup {{ username: {:?}, database: {:?} }}", user.username, database))),
Err(_) => return Err(Error::SocketError),
};
}
@@ -314,7 +306,7 @@ impl Server {
match stream.read_exact(&mut idle).await {
Ok(_) => (),
Err(_) => return Err(Error::SocketError(format!("Error reading transaction status message on server startup {{ username: {:?}, database: {:?} }}", user.username, database))),
Err(_) => return Err(Error::SocketError),
};
let (read, write) = stream.into_split();
@@ -322,29 +314,19 @@ impl Server {
let mut server = Server {
address: address.clone(),
read: BufReader::new(read),
write,
write: write,
buffer: BytesMut::with_capacity(8196),
server_info,
server_id,
process_id,
secret_key,
server_info: server_info,
process_id: process_id,
secret_key: secret_key,
in_transaction: false,
data_available: false,
bad: false,
needs_cleanup: false,
client_server_map,
client_server_map: client_server_map,
connected_at: chrono::offset::Utc::now().naive_utc(),
stats,
stats: stats,
application_name: String::new(),
last_activity: SystemTime::now(),
mirror_manager: match address.mirrors.len() {
0 => None,
_ => Some(MirroringManager::from_addresses(
user.clone(),
database.to_owned(),
address.mirrors.clone(),
)),
},
};
server.set_name("pgcat").await?;
@@ -356,10 +338,7 @@ impl Server {
// Means we implemented the protocol wrong or we're not talking to a Postgres server.
_ => {
error!("Unknown code: {}", code);
return Err(Error::ProtocolSyncError(format!(
"Unknown server code: {}",
code
)));
return Err(Error::ProtocolSyncError);
}
};
}
@@ -377,10 +356,9 @@ impl Server {
Ok(stream) => stream,
Err(err) => {
error!("Could not connect to server: {}", err);
return Err(Error::SocketError(format!("Error reading cancel message")));
return Err(Error::SocketError);
}
};
configure_socket(&stream);
debug!("Sending CancelRequest");
@@ -390,13 +368,13 @@ impl Server {
bytes.put_i32(process_id);
bytes.put_i32(secret_key);
write_all(&mut stream, bytes).await
Ok(write_all(&mut stream, bytes).await?)
}
/// Send messages to the server from the client.
pub async fn send(&mut self, messages: &BytesMut) -> Result<(), Error> {
self.mirror_send(messages);
self.stats.data_sent(messages.len(), self.server_id);
pub async fn send(&mut self, messages: BytesMut) -> Result<(), Error> {
self.stats
.data_sent(messages.len(), self.process_id, self.address.id);
match write_all_half(&mut self.write, messages).await {
Ok(_) => {
@@ -458,10 +436,7 @@ impl Server {
// Something totally unexpected, this is not a Postgres server we know.
_ => {
self.bad = true;
return Err(Error::ProtocolSyncError(format!(
"Unknown transaction state: {}",
transaction_state
)));
return Err(Error::ProtocolSyncError);
}
};
@@ -471,39 +446,6 @@ impl Server {
break;
}
// CommandComplete
'C' => {
let mut command_tag = String::new();
match message.reader().read_to_string(&mut command_tag) {
Ok(_) => {
// Non-exhaustive list of commands that are likely to change session variables/resources
// which can leak between clients. This is a best effort to block bad clients
// from poisoning a transaction-mode pool by setting inappropriate session variables
match command_tag.as_str() {
"SET\0" => {
// We don't detect set statements in transactions
// No great way to differentiate between set and set local
// As a result, we will miss cases when set statements are used in transactions
// This will reduce amount of discard statements sent
if !self.in_transaction {
debug!("Server connection marked for clean up");
self.needs_cleanup = true;
}
}
"PREPARE\0" => {
debug!("Server connection marked for clean up");
self.needs_cleanup = true;
}
_ => (),
}
}
Err(err) => {
warn!("Encountered an error while parsing CommandTag {}", err);
}
}
}
// DataRow
'D' => {
// More data is available after this message, this is not the end of the reply.
@@ -524,13 +466,9 @@ impl Server {
break;
}
// CopyData
'd' => {
// Don't flush yet, buffer until we reach limit
if self.buffer.len() >= 8196 {
break;
}
}
// CopyData: we are not buffering this one because there will be many more
// and we don't know how big this packet could be, best not to take a risk.
'd' => break,
// CopyDone
// Buffer until ReadyForQuery shows up, so don't exit the loop yet.
@@ -545,7 +483,8 @@ impl Server {
let bytes = self.buffer.clone();
// Keep track of how much data we got from the server for stats.
self.stats.data_received(bytes.len(), self.server_id);
self.stats
.data_received(bytes.len(), self.process_id, self.address.id);
// Clear the buffer for next query.
self.buffer.clear();
@@ -560,7 +499,6 @@ impl Server {
/// If the server is still inside a transaction.
/// If the client disconnects while the server is in a transaction, we will clean it up.
pub fn in_transaction(&self) -> bool {
debug!("Server in transaction: {}", self.in_transaction);
self.in_transaction
}
@@ -608,7 +546,7 @@ impl Server {
pub async fn query(&mut self, query: &str) -> Result<(), Error> {
let query = simple_query(query);
self.send(&query).await?;
self.send(query).await?;
loop {
let _ = self.recv().await?;
@@ -621,45 +559,14 @@ impl Server {
Ok(())
}
/// Perform any necessary cleanup before putting the server
/// connection back in the pool
pub async fn checkin_cleanup(&mut self) -> Result<(), Error> {
// Client disconnected with an open transaction on the server connection.
// Pgbouncer behavior is to close the server connection but that can cause
// server connection thrashing if clients repeatedly do this.
// Instead, we ROLLBACK that transaction before putting the connection back in the pool
if self.in_transaction() {
warn!("Server returned while still in transaction, rolling back transaction");
self.query("ROLLBACK").await?;
}
// Client disconnected but it performed session-altering operations such as
// SET statement_timeout to 1 or create a prepared statement. We clear that
// to avoid leaking state between clients. For performance reasons we only
// send `DISCARD ALL` if we think the session is altered instead of just sending
// it before each checkin.
if self.needs_cleanup {
warn!("Server returned with session state altered, discarding state");
self.query("DISCARD ALL").await?;
self.needs_cleanup = false;
}
Ok(())
}
/// A shorthand for `SET application_name = $1`.
#[allow(dead_code)]
pub async fn set_name(&mut self, name: &str) -> Result<(), Error> {
if self.application_name != name {
self.application_name = name.to_string();
// We don't want `SET application_name` to mark the server connection
// as needing cleanup
let needs_cleanup_before = self.needs_cleanup;
let result = Ok(self
Ok(self
.query(&format!("SET application_name = '{}'", name))
.await?);
self.needs_cleanup = needs_cleanup_before;
result
.await?)
} else {
Ok(())
}
@@ -671,35 +578,15 @@ impl Server {
self.address.clone()
}
/// Get the server connection identifier
/// Used to uniquely identify connection in statistics
pub fn server_id(&self) -> i32 {
self.server_id
/// Get the server's unique identifier.
pub fn process_id(&self) -> i32 {
self.process_id
}
// Get server's latest response timestamp
pub fn last_activity(&self) -> SystemTime {
self.last_activity
}
// Marks a connection as needing DISCARD ALL at checkin
pub fn mark_dirty(&mut self) {
self.needs_cleanup = true;
}
pub fn mirror_send(&mut self, bytes: &BytesMut) {
match self.mirror_manager.as_mut() {
Some(manager) => manager.send(bytes),
None => (),
}
}
pub fn mirror_disconnect(&mut self) {
match self.mirror_manager.as_mut() {
Some(manager) => manager.disconnect(),
None => (),
}
}
}
impl Drop for Server {
@@ -707,8 +594,8 @@ impl Drop for Server {
/// the socket is in non-blocking mode, so it may not be ready
/// for a write.
fn drop(&mut self) {
self.mirror_disconnect();
self.stats.server_disconnecting(self.server_id);
self.stats
.server_disconnecting(self.process_id(), self.address.id);
let mut bytes = BytesMut::with_capacity(4);
bytes.put_u8(b'X');
@@ -726,8 +613,7 @@ impl Drop for Server {
let duration = now - self.connected_at;
info!(
"Server connection closed {:?}, session duration: {}",
self.address,
"Server connection closed, session duration: {}",
crate::format_duration(&duration)
);
}

View File

@@ -1,4 +1,3 @@
use serde_derive::{Deserialize, Serialize};
/// Implements various sharding functions.
use sha1::{Digest, Sha1};
@@ -6,23 +5,12 @@ use sha1::{Digest, Sha1};
const PARTITION_HASH_SEED: u64 = 0x7A5B22367996DCFD;
/// The sharding functions we support.
#[derive(Debug, PartialEq, Copy, Clone, Serialize, Deserialize, Hash, std::cmp::Eq)]
#[derive(Debug, PartialEq, Copy, Clone)]
pub enum ShardingFunction {
#[serde(alias = "pg_bigint_hash", alias = "PgBigintHash")]
PgBigintHash,
#[serde(alias = "sha1", alias = "Sha1")]
Sha1,
}
impl ToString for ShardingFunction {
fn to_string(&self) -> String {
match *self {
ShardingFunction::PgBigintHash => "pg_bigint_hash".to_string(),
ShardingFunction::Sha1 => "sha1".to_string(),
}
}
}
/// The sharder.
pub struct Sharder {
/// Number of shards in the cluster.
@@ -133,7 +121,7 @@ impl Sharder {
#[inline]
fn combine(mut a: u64, b: u64) -> u64 {
a ^= b
.wrapping_add(0x49a0f4dd15e5a8e3_u64)
.wrapping_add(0x49a0f4dd15e5a8e3 as u64)
.wrapping_add(a << 54)
.wrapping_add(a >> 7);
a
@@ -141,7 +129,7 @@ impl Sharder {
#[inline]
fn pg_u32_hash(k: u32) -> u64 {
let mut a: u32 = 0x9e3779b9_u32 + std::mem::size_of::<u32>() as u32 + 3923095_u32;
let mut a: u32 = 0x9e3779b9 as u32 + std::mem::size_of::<u32>() as u32 + 3923095 as u32;
let mut b = a;
let c = a;

File diff suppressed because it is too large Load Diff

View File

@@ -1,7 +1,6 @@
// Stream wrapper.
use rustls_pemfile::{certs, read_one, Item};
use std::iter;
use rustls_pemfile::{certs, rsa_private_keys};
use std::path::Path;
use std::sync::Arc;
use tokio_rustls::rustls::{self, Certificate, PrivateKey};
@@ -18,17 +17,9 @@ pub fn load_certs(path: &Path) -> std::io::Result<Vec<Certificate>> {
}
pub fn load_keys(path: &Path) -> std::io::Result<Vec<PrivateKey>> {
let mut rd = std::io::BufReader::new(std::fs::File::open(path)?);
iter::from_fn(|| read_one(&mut rd).transpose())
.filter_map(|item| match item {
Err(err) => Some(Err(err)),
Ok(Item::RSAKey(key)) => Some(Ok(PrivateKey(key))),
Ok(Item::ECKey(key)) => Some(Ok(PrivateKey(key))),
Ok(Item::PKCS8Key(key)) => Some(Ok(PrivateKey(key))),
_ => None,
})
.collect()
rsa_private_keys(&mut std::io::BufReader::new(std::fs::File::open(path)?))
.map_err(|_| std::io::Error::new(std::io::ErrorKind::InvalidInput, "invalid key"))
.map(|mut keys| keys.drain(..).map(PrivateKey).collect())
}
pub struct Tls {
@@ -39,12 +30,12 @@ impl Tls {
pub fn new() -> Result<Self, Error> {
let config = get_config();
let certs = match load_certs(Path::new(&config.general.tls_certificate.unwrap())) {
let certs = match load_certs(&Path::new(&config.general.tls_certificate.unwrap())) {
Ok(certs) => certs,
Err(_) => return Err(Error::TlsError),
};
let mut keys = match load_keys(Path::new(&config.general.tls_private_key.unwrap())) {
let mut keys = match load_keys(&Path::new(&config.general.tls_private_key.unwrap())) {
Ok(keys) => keys,
Err(_) => return Err(Error::TlsError),
};

View File

@@ -1,8 +0,0 @@
FROM rust:bullseye
RUN apt-get update && apt-get install llvm-11 psmisc postgresql-contrib postgresql-client ruby ruby-dev libpq-dev python3 python3-pip lcov curl sudo iproute2 -y
RUN cargo install cargo-binutils rustfilt
RUN rustup component add llvm-tools-preview
RUN sudo gem install bundler
RUN wget -O toxiproxy-2.4.0.deb https://github.com/Shopify/toxiproxy/releases/download/v2.4.0/toxiproxy_2.4.0_linux_$(dpkg --print-architecture).deb && \
sudo dpkg -i toxiproxy-2.4.0.deb

View File

@@ -1,44 +0,0 @@
version: "3"
services:
pg1:
image: postgres:14
network_mode: "service:main"
environment:
POSTGRES_USER: postgres
POSTGRES_DB: postgres
POSTGRES_PASSWORD: postgres
POSTGRES_INITDB_ARGS: --auth-local=md5 --auth-host=md5 --auth=md5
command: ["postgres", "-p", "5432", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-c", "pg_stat_statements.max=100000"]
pg2:
image: postgres:14
network_mode: "service:main"
environment:
POSTGRES_USER: postgres
POSTGRES_DB: postgres
POSTGRES_PASSWORD: postgres
POSTGRES_INITDB_ARGS: --auth-local=scram-sha-256 --auth-host=scram-sha-256 --auth=scram-sha-256
command: ["postgres", "-p", "7432", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-c", "pg_stat_statements.max=100000"]
pg3:
image: postgres:14
network_mode: "service:main"
environment:
POSTGRES_USER: postgres
POSTGRES_DB: postgres
POSTGRES_PASSWORD: postgres
POSTGRES_INITDB_ARGS: --auth-local=scram-sha-256 --auth-host=scram-sha-256 --auth=scram-sha-256
command: ["postgres", "-p", "8432", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-c", "pg_stat_statements.max=100000"]
pg4:
image: postgres:14
network_mode: "service:main"
environment:
POSTGRES_USER: postgres
POSTGRES_DB: postgres
POSTGRES_PASSWORD: postgres
POSTGRES_INITDB_ARGS: --auth-local=scram-sha-256 --auth-host=scram-sha-256 --auth=scram-sha-256
command: ["postgres", "-p", "9432", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-c", "pg_stat_statements.max=100000"]
main:
build: .
command: ["bash", "/app/tests/docker/run.sh"]
volumes:
- ../../:/app/
- /app/target/

View File

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

View File

@@ -18,14 +18,9 @@ def pgcat_start():
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)
for proc in psutil.process_iter(["pid", "name"]):
if "pgcat" == proc.name():
os.kill(proc.pid, signal)
if signal == signal.SIGTERM:
# Returns 0 if pgcat process exists
time.sleep(2)
@@ -110,37 +105,6 @@ def test_shutdown_logic():
cleanup_conn(conn, cur)
pg_cat_send_signal(signal.SIGTERM)
# - - - - - - - - - - - - - - - - - -
# NO ACTIVE QUERIES ADMIN SHUTDOWN COMMAND
# Start pgcat
pgcat_start()
# Create client connection and begin transaction
conn, cur = connect_db()
admin_conn, admin_cur = connect_db(admin=True)
cur.execute("BEGIN;")
cur.execute("SELECT 1;")
cur.execute("COMMIT;")
# Send SHUTDOWN command pgcat while not in transaction
admin_cur.execute("SHUTDOWN;")
time.sleep(1)
# Check that any new queries fail after SHUTDOWN command since server should close with no active transactions
try:
cur.execute("SELECT 1;")
except psycopg2.OperationalError as e:
pass
else:
# Fail if query execution succeeded
raise Exception("Server not closed after sigint")
cleanup_conn(conn, cur)
cleanup_conn(admin_conn, admin_cur)
pg_cat_send_signal(signal.SIGTERM)
# - - - - - - - - - - - - - - - - - -
# HANDLE TRANSACTION WITH SIGINT
@@ -167,36 +131,6 @@ def test_shutdown_logic():
cleanup_conn(conn, cur)
pg_cat_send_signal(signal.SIGTERM)
# - - - - - - - - - - - - - - - - - -
# HANDLE TRANSACTION WITH ADMIN SHUTDOWN COMMAND
# Start pgcat
pgcat_start()
# Create client connection and begin transaction
conn, cur = connect_db()
admin_conn, admin_cur = connect_db(admin=True)
cur.execute("BEGIN;")
cur.execute("SELECT 1;")
# Send SHUTDOWN command pgcat while still in transaction
admin_cur.execute("SHUTDOWN;")
if admin_cur.fetchall()[0][0] != "t":
raise Exception("PgCat unable to send signal")
time.sleep(1)
# Check that any new queries succeed after SHUTDOWN command since server should still allow transaction to complete
try:
cur.execute("SELECT 1;")
except psycopg2.OperationalError as e:
# Fail if query fails since server closed
raise Exception("Server closed while in transaction", e.pgerror)
cleanup_conn(conn, cur)
cleanup_conn(admin_conn, admin_cur)
pg_cat_send_signal(signal.SIGTERM)
# - - - - - - - - - - - - - - - - - -
# NO NEW NON-ADMIN CONNECTIONS DURING SHUTDOWN
# Start pgcat

View File

@@ -1,8 +1,6 @@
source "https://rubygems.org"
gem "pg"
gem "toml"
gem "rspec"
gem "rubocop"
gem "toxiproxy"
gem "activerecord"
gem "rubocop"
gem "toml", "~> 0.3.0"

View File

@@ -1,22 +1,21 @@
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 (7.0.3.1)
activesupport (= 7.0.3.1)
activerecord (7.0.3.1)
activemodel (= 7.0.3.1)
activesupport (= 7.0.3.1)
activesupport (7.0.3.1)
concurrent-ruby (~> 1.0, >= 1.0.2)
i18n (>= 1.6, < 2)
minitest (>= 5.1)
tzinfo (~> 2.0)
ast (2.4.2)
concurrent-ruby (1.1.10)
diff-lcs (1.5.0)
i18n (1.12.0)
i18n (1.11.0)
concurrent-ruby (~> 1.0)
minitest (5.17.0)
minitest (5.16.2)
parallel (1.22.1)
parser (3.1.2.0)
ast (~> 2.4.1)
@@ -25,19 +24,6 @@ GEM
rainbow (3.1.1)
regexp_parser (2.3.1)
rexml (3.2.5)
rspec (3.11.0)
rspec-core (~> 3.11.0)
rspec-expectations (~> 3.11.0)
rspec-mocks (~> 3.11.0)
rspec-core (3.11.0)
rspec-support (~> 3.11.0)
rspec-expectations (3.11.0)
diff-lcs (>= 1.2.0, < 2.0)
rspec-support (~> 3.11.0)
rspec-mocks (3.11.1)
diff-lcs (>= 1.2.0, < 2.0)
rspec-support (~> 3.11.0)
rspec-support (3.11.0)
rubocop (1.29.0)
parallel (~> 1.10)
parser (>= 3.1.0.0)
@@ -52,23 +38,19 @@ GEM
ruby-progressbar (1.11.0)
toml (0.3.0)
parslet (>= 1.8.0, < 3.0.0)
toxiproxy (2.0.1)
tzinfo (2.0.5)
tzinfo (2.0.4)
concurrent-ruby (~> 1.0)
unicode-display_width (2.1.0)
PLATFORMS
aarch64-linux
arm64-darwin-21
x86_64-linux
DEPENDENCIES
activerecord
pg
rspec
rubocop
toml
toxiproxy
toml (~> 0.3.0)
BUNDLED WITH
2.3.21
2.3.7

View File

@@ -1,369 +0,0 @@
# frozen_string_literal: true
require 'uri'
require_relative 'spec_helper'
describe "Admin" do
let(:processes) { Helpers::Pgcat.single_instance_setup("sharded_db", 10) }
let(:pgcat_conn_str) { processes.pgcat.connection_string("sharded_db", "sharding_user") }
after do
processes.all_databases.map(&:reset)
processes.pgcat.shutdown
end
describe "SHOW STATS" do
context "clients connect and make one query" do
it "updates *_query_time and *_wait_time" do
connection = PG::connect("#{pgcat_conn_str}?application_name=one_query")
connection.async_exec("SELECT pg_sleep(0.25)")
connection.async_exec("SELECT pg_sleep(0.25)")
connection.async_exec("SELECT pg_sleep(0.25)")
connection.close
# wait for averages to be calculated, we shouldn't do this too often
sleep(15.5)
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
results = admin_conn.async_exec("SHOW STATS")[0]
admin_conn.close
expect(results["total_query_time"].to_i).to be_within(200).of(750)
expect(results["avg_query_time"].to_i).to_not eq(0)
expect(results["total_wait_time"].to_i).to_not eq(0)
expect(results["avg_wait_time"].to_i).to_not eq(0)
end
end
end
describe "SHOW POOLS" do
context "bad credentials" do
it "does not change any stats" do
bad_passsword_url = URI(pgcat_conn_str)
bad_passsword_url.password = "wrong"
expect { PG::connect("#{bad_passsword_url.to_s}?application_name=bad_password") }.to raise_error(PG::ConnectionBad)
sleep(1)
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
results = admin_conn.async_exec("SHOW POOLS")[0]
%w[cl_idle 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["sv_idle"]).to eq("1")
end
end
context "bad database name" do
it "does not change any stats" do
bad_db_url = URI(pgcat_conn_str)
bad_db_url.path = "/wrong_db"
expect { PG::connect("#{bad_db_url.to_s}?application_name=bad_db") }.to raise_error(PG::ConnectionBad)
sleep(1)
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
results = admin_conn.async_exec("SHOW POOLS")[0]
%w[cl_idle 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["sv_idle"]).to eq("1")
end
end
context "client connects but issues no queries" do
it "only affects cl_idle stats" do
connections = Array.new(20) { PG::connect(pgcat_conn_str) }
sleep(1)
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
results = admin_conn.async_exec("SHOW POOLS")[0]
%w[cl_active cl_waiting cl_cancel_req sv_active sv_used sv_tested sv_login maxwait].each do |s|
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("20")
expect(results["sv_idle"]).to eq("1")
connections.map(&:close)
sleep(1.1)
results = admin_conn.async_exec("SHOW POOLS")[0]
%w[cl_active cl_idle 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["sv_idle"]).to eq("1")
end
end
context "clients connect and make one query" do
it "only affects cl_idle, sv_idle stats" do
connections = Array.new(5) { PG::connect("#{pgcat_conn_str}?application_name=one_query") }
connections.each do |c|
Thread.new { c.async_exec("SELECT pg_sleep(2.5)") }
end
sleep(1.1)
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
results = admin_conn.async_exec("SHOW POOLS")[0]
%w[cl_idle cl_waiting cl_cancel_req sv_idle 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_active"]).to eq("5")
expect(results["sv_active"]).to eq("5")
sleep(3)
results = admin_conn.async_exec("SHOW POOLS")[0]
%w[cl_active cl_waiting cl_cancel_req sv_active sv_used sv_tested sv_login maxwait].each do |s|
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("5")
expect(results["sv_idle"]).to eq("5")
connections.map(&:close)
sleep(1)
results = admin_conn.async_exec("SHOW POOLS")[0]
%w[cl_idle 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["sv_idle"]).to eq("5")
end
end
context "client connects and opens a transaction and closes connection uncleanly" do
it "produces correct statistics" do
connections = Array.new(5) { PG::connect("#{pgcat_conn_str}?application_name=one_query") }
connections.each do |c|
Thread.new do
c.async_exec("BEGIN")
c.async_exec("SELECT pg_sleep(0.01)")
c.close
end
end
sleep(1.1)
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
results = admin_conn.async_exec("SHOW POOLS")[0]
%w[cl_idle 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["sv_idle"]).to eq("5")
end
end
context "client fail to checkout connection from the pool" do
it "counts clients as idle" do
new_configs = processes.pgcat.current_config
new_configs["general"]["connect_timeout"] = 500
new_configs["general"]["ban_time"] = 1
new_configs["general"]["shutdown_timeout"] = 1
new_configs["pools"]["sharded_db"]["users"]["0"]["pool_size"] = 1
processes.pgcat.update_config(new_configs)
processes.pgcat.reload_config
threads = []
connections = Array.new(5) { PG::connect("#{pgcat_conn_str}?application_name=one_query") }
connections.each do |c|
threads << Thread.new { c.async_exec("SELECT pg_sleep(1)") rescue PG::SystemError }
end
sleep(2)
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
results = admin_conn.async_exec("SHOW POOLS")[0]
%w[cl_active cl_waiting cl_cancel_req sv_active sv_used sv_tested sv_login maxwait].each do |s|
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("5")
expect(results["sv_idle"]).to eq("1")
threads.map(&:join)
connections.map(&:close)
end
end
context "clients overwhelm server pools" do
let(:processes) { Helpers::Pgcat.single_instance_setup("sharded_db", 2) }
it "cl_waiting is updated to show it" do
threads = []
connections = Array.new(4) { PG::connect("#{pgcat_conn_str}?application_name=one_query") }
connections.each do |c|
threads << Thread.new { c.async_exec("SELECT pg_sleep(1.5)") }
end
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|
raise StandardError, "Field #{s} was expected to be 0 but found to be #{results[s]}" if results[s] != "0"
end
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 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")
expect(results["sv_idle"]).to eq("2")
threads.map(&:join)
connections.map(&:close)
end
it "show correct max_wait" do
threads = []
connections = Array.new(4) { PG::connect("#{pgcat_conn_str}?application_name=one_query") }
connections.each do |c|
threads << Thread.new { c.async_exec("SELECT pg_sleep(1.5)") }
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 POOLS")[0]
expect(results["maxwait"]).to eq("1")
expect(results["maxwait_us"].to_i).to be_within(200_000).of(500_000)
sleep(4.5) # Allow time for stats to update
results = admin_conn.async_exec("SHOW POOLS")[0]
expect(results["maxwait"]).to eq("0")
threads.map(&:join)
connections.map(&:close)
end
end
end
describe "SHOW CLIENTS" do
it "reports correct number and application names" do
conn_str = processes.pgcat.connection_string("sharded_db", "sharding_user")
connections = Array.new(20) { |i| PG::connect("#{conn_str}?application_name=app#{i % 5}") }
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
sleep(1) # Wait for stats to be updated
results = admin_conn.async_exec("SHOW CLIENTS")
expect(results.count).to eq(21) # count admin clients
expect(results.select { |c| c["application_name"] == "app3" || c["application_name"] == "app4" }.count).to eq(8)
expect(results.select { |c| c["database"] == "pgcat" }.count).to eq(1)
connections[0..5].map(&:close)
sleep(1) # Wait for stats to be updated
results = admin_conn.async_exec("SHOW CLIENTS")
expect(results.count).to eq(15)
connections[6..].map(&:close)
sleep(1) # Wait for stats to be updated
expect(admin_conn.async_exec("SHOW CLIENTS").count).to eq(1)
admin_conn.close
end
it "reports correct number of queries and transactions" do
conn_str = processes.pgcat.connection_string("sharded_db", "sharding_user")
connections = Array.new(2) { |i| PG::connect("#{conn_str}?application_name=app#{i}") }
connections.each do |c|
c.async_exec("SELECT 1")
c.async_exec("SELECT 2")
c.async_exec("SELECT 3")
c.async_exec("BEGIN")
c.async_exec("SELECT 4")
c.async_exec("SELECT 5")
c.async_exec("COMMIT")
end
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
sleep(1) # Wait for stats to be updated
results = admin_conn.async_exec("SHOW CLIENTS")
expect(results.count).to eq(3)
normal_client_results = results.reject { |r| r["database"] == "pgcat" }
expect(normal_client_results[0]["transaction_count"]).to eq("4")
expect(normal_client_results[1]["transaction_count"]).to eq("4")
expect(normal_client_results[0]["query_count"]).to eq("7")
expect(normal_client_results[1]["query_count"]).to eq("7")
admin_conn.close
connections.map(&:close)
end
end
describe "Manual Banning" do
let(:processes) { Helpers::Pgcat.single_shard_setup("sharded_db", 10) }
before do
new_configs = processes.pgcat.current_config
# Prevent immediate unbanning when we ban localhost
new_configs["pools"]["sharded_db"]["shards"]["0"]["servers"][0][0] = "127.0.0.1"
new_configs["pools"]["sharded_db"]["shards"]["0"]["servers"][1][0] = "127.0.0.1"
processes.pgcat.update_config(new_configs)
processes.pgcat.reload_config
end
describe "BAN/UNBAN and SHOW BANS" do
it "bans/unbans hosts" do
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
# Returns a list of the banned addresses
results = admin_conn.async_exec("BAN localhost 10").to_a
expect(results.count).to eq(2)
expect(results.map{ |r| r["host"] }.uniq).to eq(["localhost"])
# Subsequent calls should yield no results
results = admin_conn.async_exec("BAN localhost 10").to_a
expect(results.count).to eq(0)
results = admin_conn.async_exec("SHOW BANS").to_a
expect(results.count).to eq(2)
expect(results.map{ |r| r["host"] }.uniq).to eq(["localhost"])
# Returns a list of the unbanned addresses
results = admin_conn.async_exec("UNBAN localhost").to_a
expect(results.count).to eq(2)
expect(results.map{ |r| r["host"] }.uniq).to eq(["localhost"])
# Subsequent calls should yield no results
results = admin_conn.async_exec("UNBAN localhost").to_a
expect(results.count).to eq(0)
results = admin_conn.async_exec("SHOW BANS").to_a
expect(results.count).to eq(0)
end
it "honors ban duration" do
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
# Returns a list of the banned addresses
results = admin_conn.async_exec("BAN localhost 1").to_a
expect(results.count).to eq(2)
expect(results.map{ |r| r["host"] }.uniq).to eq(["localhost"])
sleep(2)
# After 2 seconds the ban should be lifted
results = admin_conn.async_exec("SHOW BANS").to_a
expect(results.count).to eq(0)
end
it "can handle bad input" do
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
expect { admin_conn.async_exec("BAN").to_a }.to raise_error(PG::SystemError)
expect { admin_conn.async_exec("BAN a").to_a }.to raise_error(PG::SystemError)
expect { admin_conn.async_exec("BAN a a").to_a }.to raise_error(PG::SystemError)
expect { admin_conn.async_exec("BAN a -5").to_a }.to raise_error(PG::SystemError)
expect { admin_conn.async_exec("BAN a 0").to_a }.to raise_error(PG::SystemError)
expect { admin_conn.async_exec("BAN a a a").to_a }.to raise_error(PG::SystemError)
expect { admin_conn.async_exec("UNBAN").to_a }.to raise_error(PG::SystemError)
end
end
end
describe "SHOW users" do
it "returns the right users" do
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
results = admin_conn.async_exec("SHOW USERS")[0]
admin_conn.close
expect(results["name"]).to eq("sharding_user")
expect(results["pool_mode"]).to eq("transaction")
end
end
end

View File

@@ -1,94 +0,0 @@
require 'pg'
require 'toxiproxy'
class PgInstance
attr_reader :port
attr_reader :username
attr_reader :password
attr_reader :database_name
def initialize(port, username, password, database_name)
@original_port = port
@toxiproxy_port = 10000 + port.to_i
@port = @toxiproxy_port
@username = username
@password = password
@database_name = database_name
@toxiproxy_name = "database_#{@original_port}"
Toxiproxy.populate([{
name: @toxiproxy_name,
listen: "0.0.0.0:#{@toxiproxy_port}",
upstream: "localhost:#{@original_port}",
}])
# Toxiproxy server will outlive our PgInstance objects
# so we want to destroy our proxies before exiting
# Ruby finalizer is ideal for doing this
ObjectSpace.define_finalizer(@toxiproxy_name, proc { Toxiproxy[@toxiproxy_name].destroy })
end
def with_connection
conn = PG.connect("postgres://#{@username}:#{@password}@localhost:#{port}/#{database_name}")
yield conn
ensure
conn&.close
end
def reset
reset_toxics
reset_stats
drop_connections
sleep 0.1
end
def toxiproxy
Toxiproxy[@toxiproxy_name]
end
def take_down
if block_given?
Toxiproxy[@toxiproxy_name].toxic(:limit_data, bytes: 5).apply { yield }
else
Toxiproxy[@toxiproxy_name].toxic(:limit_data, bytes: 5).toxics.each(&:save)
end
end
def add_latency(latency)
if block_given?
Toxiproxy[@toxiproxy_name].toxic(:latency, latency: latency).apply { yield }
else
Toxiproxy[@toxiproxy_name].toxic(:latency, latency: latency).toxics.each(&:save)
end
end
def delete_proxy
Toxiproxy[@toxiproxy_name].delete
end
def reset_toxics
Toxiproxy[@toxiproxy_name].toxics.each(&:destroy)
sleep 0.1
end
def reset_stats
with_connection { |c| c.async_exec("SELECT pg_stat_statements_reset()") }
end
def drop_connections
username = with_connection { |c| c.async_exec("SELECT current_user")[0]["current_user"] }
with_connection { |c| c.async_exec("SELECT pg_terminate_backend(pid) FROM pg_stat_activity WHERE pid <> pg_backend_pid() AND usename='#{username}'") }
end
def count_connections
with_connection { |c| c.async_exec("SELECT COUNT(*) as count FROM pg_stat_activity")[0]["count"].to_i }
end
def count_query(query)
with_connection { |c| c.async_exec("SELECT SUM(calls) FROM pg_stat_statements WHERE query = '#{query}'")[0]["sum"].to_i }
end
def count_select_1_plus_2
with_connection { |c| c.async_exec("SELECT SUM(calls) FROM pg_stat_statements WHERE query = 'SELECT $1 + $2'")[0]["sum"].to_i }
end
end

View File

@@ -1,148 +0,0 @@
require 'json'
require 'ostruct'
require_relative 'pgcat_process'
require_relative 'pg_instance'
module Helpers
module Pgcat
def self.three_shard_setup(pool_name, pool_size, pool_mode="transaction", lb_mode="random", log_level="info")
user = {
"password" => "sharding_user",
"pool_size" => pool_size,
"statement_timeout" => 0,
"username" => "sharding_user"
}
pgcat = PgcatProcess.new(log_level)
primary0 = PgInstance.new(5432, user["username"], user["password"], "shard0")
primary1 = PgInstance.new(7432, user["username"], user["password"], "shard1")
primary2 = PgInstance.new(8432, user["username"], user["password"], "shard2")
pgcat_cfg = pgcat.current_config
pgcat_cfg["pools"] = {
"#{pool_name}" => {
"default_role" => "any",
"pool_mode" => pool_mode,
"load_balancing_mode" => lb_mode,
"primary_reads_enabled" => true,
"query_parser_enabled" => true,
"automatic_sharding_key" => "data.id",
"sharding_function" => "pg_bigint_hash",
"shards" => {
"0" => { "database" => "shard0", "servers" => [["localhost", primary0.port.to_s, "primary"]] },
"1" => { "database" => "shard1", "servers" => [["localhost", primary1.port.to_s, "primary"]] },
"2" => { "database" => "shard2", "servers" => [["localhost", primary2.port.to_s, "primary"]] },
},
"users" => { "0" => user }
}
}
pgcat.update_config(pgcat_cfg)
pgcat.start
pgcat.wait_until_ready
OpenStruct.new.tap do |struct|
struct.pgcat = pgcat
struct.shards = [primary0, primary1, primary2]
struct.all_databases = [primary0, primary1, primary2]
end
end
def self.single_instance_setup(pool_name, pool_size, pool_mode="transaction", lb_mode="random", log_level="trace")
user = {
"password" => "sharding_user",
"pool_size" => pool_size,
"statement_timeout" => 0,
"username" => "sharding_user"
}
pgcat = PgcatProcess.new(log_level)
pgcat_cfg = pgcat.current_config
primary = PgInstance.new(5432, user["username"], user["password"], "shard0")
# Main proxy configs
pgcat_cfg["pools"] = {
"#{pool_name}" => {
"default_role" => "primary",
"pool_mode" => pool_mode,
"load_balancing_mode" => lb_mode,
"primary_reads_enabled" => false,
"query_parser_enabled" => false,
"sharding_function" => "pg_bigint_hash",
"shards" => {
"0" => {
"database" => "shard0",
"servers" => [
["localhost", primary.port.to_s, "primary"]
]
},
},
"users" => { "0" => user }
}
}
pgcat_cfg["general"]["port"] = pgcat.port
pgcat.update_config(pgcat_cfg)
pgcat.start
pgcat.wait_until_ready
OpenStruct.new.tap do |struct|
struct.pgcat = pgcat
struct.primary = primary
struct.all_databases = [primary]
end
end
def self.single_shard_setup(pool_name, pool_size, pool_mode="transaction", lb_mode="random", log_level="info")
user = {
"password" => "sharding_user",
"pool_size" => pool_size,
"statement_timeout" => 0,
"username" => "sharding_user"
}
pgcat = PgcatProcess.new(log_level)
pgcat_cfg = pgcat.current_config
primary = PgInstance.new(5432, user["username"], user["password"], "shard0")
replica0 = PgInstance.new(7432, user["username"], user["password"], "shard0")
replica1 = PgInstance.new(8432, user["username"], user["password"], "shard0")
replica2 = PgInstance.new(9432, user["username"], user["password"], "shard0")
# Main proxy configs
pgcat_cfg["pools"] = {
"#{pool_name}" => {
"default_role" => "any",
"pool_mode" => pool_mode,
"load_balancing_mode" => lb_mode,
"primary_reads_enabled" => false,
"query_parser_enabled" => false,
"sharding_function" => "pg_bigint_hash",
"shards" => {
"0" => {
"database" => "shard0",
"servers" => [
["localhost", primary.port.to_s, "primary"],
["localhost", replica0.port.to_s, "replica"],
["localhost", replica1.port.to_s, "replica"],
["localhost", replica2.port.to_s, "replica"]
]
},
},
"users" => { "0" => user }
}
}
pgcat_cfg["general"]["port"] = pgcat.port
pgcat.update_config(pgcat_cfg)
pgcat.start
pgcat.wait_until_ready
OpenStruct.new.tap do |struct|
struct.pgcat = pgcat
struct.primary = primary
struct.replicas = [replica0, replica1, replica2]
struct.all_databases = [primary, replica0, replica1, replica2]
end
end
end
end

View File

@@ -1,131 +0,0 @@
require 'pg'
require 'toml'
require 'fileutils'
require 'securerandom'
class PgcatProcess
attr_reader :port
attr_reader :pid
def self.finalize(pid, log_filename, config_filename)
if pid
Process.kill("TERM", pid)
Process.wait(pid)
end
File.delete(config_filename) if File.exist?(config_filename)
File.delete(log_filename) if File.exist?(log_filename)
end
def initialize(log_level)
@env = {"RUST_LOG" => log_level}
@port = rand(20000..32760)
@log_level = log_level
@log_filename = "/tmp/pgcat_log_#{SecureRandom.urlsafe_base64}.log"
@config_filename = "/tmp/pgcat_cfg_#{SecureRandom.urlsafe_base64}.toml"
command_path = if ENV['CARGO_TARGET_DIR'] then
"#{ENV['CARGO_TARGET_DIR']}/debug/pgcat"
else
'../../target/debug/pgcat'
end
@command = "#{command_path} #{@config_filename}"
FileUtils.cp("../../pgcat.toml", @config_filename)
cfg = current_config
cfg["general"]["port"] = @port.to_i
cfg["general"]["enable_prometheus_exporter"] = false
update_config(cfg)
end
def logs
File.read(@log_filename)
end
def update_config(config_hash)
@original_config = current_config
output_to_write = TOML::Generator.new(config_hash).body
output_to_write = output_to_write.gsub(/,\s*["|'](\d+)["|']\s*,/, ',\1,')
output_to_write = output_to_write.gsub(/,\s*["|'](\d+)["|']\s*\]/, ',\1]')
File.write(@config_filename, output_to_write)
end
def current_config
loadable_string = File.read(@config_filename)
loadable_string = loadable_string.gsub(/,\s*(\d+)\s*,/, ', "\1",')
loadable_string = loadable_string.gsub(/,\s*(\d+)\s*\]/, ', "\1"]')
TOML.load(loadable_string)
end
def reload_config
`kill -s HUP #{@pid}`
sleep 0.5
end
def start
raise StandardError, "Process is already started" unless @pid.nil?
@pid = Process.spawn(@env, @command, err: @log_filename, out: @log_filename)
ObjectSpace.define_finalizer(@log_filename, proc { PgcatProcess.finalize(@pid, @log_filename, @config_filename) })
return self
end
def wait_until_ready
exc = nil
10.times do
PG::connect(example_connection_string).close
return self
rescue => e
exc = e
sleep(0.5)
end
puts exc
raise StandardError, "Process #{@pid} never became ready. Logs #{logs}"
end
def stop
return unless @pid
Process.kill("TERM", @pid)
Process.wait(@pid)
@pid = nil
end
def shutdown
stop
File.delete(@config_filename) if File.exist?(@config_filename)
File.delete(@log_filename) if File.exist?(@log_filename)
end
def admin_connection_string
cfg = current_config
username = cfg["general"]["admin_username"]
password = cfg["general"]["admin_password"]
"postgresql://#{username}:#{password}@0.0.0.0:#{@port}/pgcat"
end
def connection_string(pool_name, username)
cfg = current_config
user_idx, user_obj = cfg["pools"][pool_name]["users"].detect { |k, user| user["username"] == username }
password = user_obj["password"]
"postgresql://#{username}:#{password}@0.0.0.0:#{@port}/#{pool_name}"
end
def example_connection_string
cfg = current_config
first_pool_name = cfg["pools"].keys[0]
db_name = first_pool_name
username = cfg["pools"][first_pool_name]["users"]["0"]["username"]
password = cfg["pools"][first_pool_name]["users"]["0"]["password"]
"postgresql://#{username}:#{password}@0.0.0.0:#{@port}/#{db_name}?application_name=example_app"
end
end

View File

@@ -1,164 +0,0 @@
# frozen_string_literal: true
require_relative 'spec_helper'
describe "Random Load Balancing" do
let(:processes) { Helpers::Pgcat.single_shard_setup("sharded_db", 5) }
after do
processes.all_databases.map(&:reset)
processes.pgcat.shutdown
end
context "under regular circumstances" do
it "balances query volume between all instances" do
conn = PG.connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
query_count = QUERY_COUNT
expected_share = query_count / processes.all_databases.count
failed_count = 0
query_count.times do
conn.async_exec("SELECT 1 + 2")
rescue
failed_count += 1
end
expect(failed_count).to eq(0)
processes.all_databases.map(&:count_select_1_plus_2).each do |instance_share|
expect(instance_share).to be_within(expected_share * MARGIN_OF_ERROR).of(expected_share)
end
end
end
context "when some replicas are down" do
it "balances query volume between working instances" do
conn = PG.connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
expected_share = QUERY_COUNT / (processes.all_databases.count - 2)
failed_count = 0
processes[:replicas][0].take_down do
processes[:replicas][1].take_down do
QUERY_COUNT.times do
conn.async_exec("SELECT 1 + 2")
rescue
conn = PG.connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
failed_count += 1
end
end
end
processes.all_databases.each do |instance|
queries_routed = instance.count_select_1_plus_2
if processes.replicas[0..1].include?(instance)
expect(queries_routed).to eq(0)
else
expect(queries_routed).to be_within(expected_share * MARGIN_OF_ERROR).of(expected_share)
end
end
end
end
end
describe "Least Outstanding Queries Load Balancing" do
let(:processes) { Helpers::Pgcat.single_shard_setup("sharded_db", 1, "transaction", "loc") }
after do
processes.all_databases.map(&:reset)
processes.pgcat.shutdown
end
context "under homogenous load" do
it "balances query volume between all instances" do
conn = PG.connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
query_count = QUERY_COUNT
expected_share = query_count / processes.all_databases.count
failed_count = 0
query_count.times do
conn.async_exec("SELECT 1 + 2")
rescue
failed_count += 1
end
expect(failed_count).to eq(0)
processes.all_databases.map(&:count_select_1_plus_2).each do |instance_share|
expect(instance_share).to be_within(expected_share * MARGIN_OF_ERROR).of(expected_share)
end
end
end
context "under heterogeneous load" do
xit "balances query volume between all instances based on how busy they are" do
slow_query_count = 2
threads = Array.new(slow_query_count) do
Thread.new do
conn = PG.connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
conn.async_exec("BEGIN")
end
end
conn = PG.connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
query_count = QUERY_COUNT
expected_share = query_count / (processes.all_databases.count - slow_query_count)
failed_count = 0
query_count.times do
conn.async_exec("SELECT 1 + 2")
rescue
failed_count += 1
end
expect(failed_count).to eq(0)
# Under LOQ, we expect replicas running the slow pg_sleep
# to get no selects
expect(
processes.
all_databases.
map(&:count_select_1_plus_2).
count { |instance_share| instance_share == 0 }
).to eq(slow_query_count)
# We also expect the quick queries to be spread across
# the idle servers only
processes.
all_databases.
map(&:count_select_1_plus_2).
reject { |instance_share| instance_share == 0 }.
each do |instance_share|
expect(instance_share).to be_within(expected_share * MARGIN_OF_ERROR).of(expected_share)
end
threads.map(&:join)
end
end
context "when some replicas are down" do
it "balances query volume between working instances" do
conn = PG.connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
expected_share = QUERY_COUNT / (processes.all_databases.count - 2)
failed_count = 0
processes[:replicas][0].take_down do
processes[:replicas][1].take_down do
QUERY_COUNT.times do
conn.async_exec("SELECT 1 + 2")
rescue
conn = PG.connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
failed_count += 1
end
end
end
expect(failed_count).to be <= 2
processes.all_databases.each do |instance|
queries_routed = instance.count_select_1_plus_2
if processes.replicas[0..1].include?(instance)
expect(queries_routed).to eq(0)
else
expect(queries_routed).to be_within(expected_share * MARGIN_OF_ERROR).of(expected_share)
end
end
end
end
end

View File

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

View File

@@ -1,312 +0,0 @@
# frozen_string_literal: true
require_relative 'spec_helper'
describe "Miscellaneous" do
let(:processes) { Helpers::Pgcat.single_shard_setup("sharded_db", 5) }
after do
processes.all_databases.map(&:reset)
processes.pgcat.shutdown
end
context "when adding then removing instance using RELOAD" do
it "works correctly" do
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
current_configs = processes.pgcat.current_config
correct_count = current_configs["pools"]["sharded_db"]["shards"]["0"]["servers"].count
expect(admin_conn.async_exec("SHOW DATABASES").count).to eq(correct_count)
extra_replica = current_configs["pools"]["sharded_db"]["shards"]["0"]["servers"].last.clone
extra_replica[0] = "127.0.0.1"
current_configs["pools"]["sharded_db"]["shards"]["0"]["servers"] << extra_replica
processes.pgcat.update_config(current_configs) # with replica added
processes.pgcat.reload_config
correct_count = current_configs["pools"]["sharded_db"]["shards"]["0"]["servers"].count
expect(admin_conn.async_exec("SHOW DATABASES").count).to eq(correct_count)
current_configs["pools"]["sharded_db"]["shards"]["0"]["servers"].pop
processes.pgcat.update_config(current_configs) # with replica removed again
processes.pgcat.reload_config
correct_count = current_configs["pools"]["sharded_db"]["shards"]["0"]["servers"].count
expect(admin_conn.async_exec("SHOW DATABASES").count).to eq(correct_count)
end
end
context "when removing then adding instance back using RELOAD" do
it "works correctly" do
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
current_configs = processes.pgcat.current_config
correct_count = current_configs["pools"]["sharded_db"]["shards"]["0"]["servers"].count
expect(admin_conn.async_exec("SHOW DATABASES").count).to eq(correct_count)
removed_replica = current_configs["pools"]["sharded_db"]["shards"]["0"]["servers"].pop
processes.pgcat.update_config(current_configs) # with replica removed
processes.pgcat.reload_config
correct_count = current_configs["pools"]["sharded_db"]["shards"]["0"]["servers"].count
expect(admin_conn.async_exec("SHOW DATABASES").count).to eq(correct_count)
current_configs["pools"]["sharded_db"]["shards"]["0"]["servers"] << removed_replica
processes.pgcat.update_config(current_configs) # with replica added again
processes.pgcat.reload_config
correct_count = current_configs["pools"]["sharded_db"]["shards"]["0"]["servers"].count
expect(admin_conn.async_exec("SHOW DATABASES").count).to eq(correct_count)
end
end
describe "TCP Keepalives" do
# Ideally, we should block TCP traffic to the database using
# iptables to mimic passive (connection is dropped without a RST packet)
# but we cannot do this in CircleCI because iptables requires NET_ADMIN
# capability that we cannot enable in CircleCI
# Toxiproxy won't work either because it does not block keepalives
# so our best bet is to query the OS keepalive params set on the socket
context "default settings" do
it "applies default keepalive settings" do
# We query ss command to verify that we have correct keepalive values set
# we can only verify the keepalives_idle parameter but that's good enough
# example output
#Recv-Q Send-Q Local Address:Port Peer Address:Port Process
#0 0 127.0.0.1:60526 127.0.0.1:18432 timer:(keepalive,1min59sec,0)
#0 0 127.0.0.1:60664 127.0.0.1:19432 timer:(keepalive,4.123ms,0)
port_search_criteria = processes.all_databases.map { |d| "dport = :#{d.port}"}.join(" or ")
results = `ss -t4 state established -o -at '( #{port_search_criteria} )'`.lines
results.shift
results.each { |line| expect(line).to match(/timer:\(keepalive,.*ms,0\)/) }
end
end
context "changed settings" do
it "applies keepalive settings from config" do
new_configs = processes.pgcat.current_config
new_configs["general"]["tcp_keepalives_idle"] = 120
new_configs["general"]["tcp_keepalives_count"] = 1
new_configs["general"]["tcp_keepalives_interval"] = 1
processes.pgcat.update_config(new_configs)
# We need to kill the old process that was using the default configs
processes.pgcat.stop
processes.pgcat.start
processes.pgcat.wait_until_ready
port_search_criteria = processes.all_databases.map { |d| "dport = :#{d.port}"}.join(" or ")
results = `ss -t4 state established -o -at '( #{port_search_criteria} )'`.lines
results.shift
results.each { |line| expect(line).to include("timer:(keepalive,1min") }
end
end
end
describe "Extended Protocol handling" do
it "does not send packets that client does not expect during extended protocol sequence" do
new_configs = processes.pgcat.current_config
new_configs["general"]["connect_timeout"] = 500
new_configs["general"]["ban_time"] = 1
new_configs["general"]["shutdown_timeout"] = 1
new_configs["pools"]["sharded_db"]["users"]["0"]["pool_size"] = 1
processes.pgcat.update_config(new_configs)
processes.pgcat.reload_config
25.times do
Thread.new do
conn = PG::connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
conn.async_exec("SELECT pg_sleep(5)") rescue PG::SystemError
ensure
conn&.close
end
end
sleep(0.5)
conn_under_test = PG::connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
stdout, stderr = with_captured_stdout_stderr do
15.times do |i|
conn_under_test.async_exec("SELECT 1") rescue PG::SystemError
conn_under_test.exec_params("SELECT #{i} + $1", [i]) rescue PG::SystemError
sleep 1
end
end
raise StandardError, "Libpq got unexpected messages while idle" if stderr.include?("arrived from server while idle")
end
end
describe "Pool recycling after config reload" do
let(:processes) { Helpers::Pgcat.three_shard_setup("sharded_db", 5) }
it "should update pools for new clients and clients that are no longer in transaction" do
server_conn = PG::connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
server_conn.async_exec("BEGIN")
# No config change yet, client should set old configs
current_datebase_from_pg = server_conn.async_exec("SELECT current_database();")[0]["current_database"]
expect(current_datebase_from_pg).to eq('shard0')
# Swap shards
new_config = processes.pgcat.current_config
shard0 = new_config["pools"]["sharded_db"]["shards"]["0"]
shard1 = new_config["pools"]["sharded_db"]["shards"]["1"]
new_config["pools"]["sharded_db"]["shards"]["0"] = shard1
new_config["pools"]["sharded_db"]["shards"]["1"] = shard0
# Reload config
processes.pgcat.update_config(new_config)
processes.pgcat.reload_config
sleep 0.5
# Config changed but transaction is in progress, client should set old configs
current_datebase_from_pg = server_conn.async_exec("SELECT current_database();")[0]["current_database"]
expect(current_datebase_from_pg).to eq('shard0')
server_conn.async_exec("COMMIT")
# Transaction finished, client should get new configs
current_datebase_from_pg = server_conn.async_exec("SELECT current_database();")[0]["current_database"]
expect(current_datebase_from_pg).to eq('shard1')
# New connection should get new configs
server_conn.close()
server_conn = PG::connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
current_datebase_from_pg = server_conn.async_exec("SELECT current_database();")[0]["current_database"]
expect(current_datebase_from_pg).to eq('shard1')
end
end
describe "Clients closing connection in the middle of transaction" do
it "sends a rollback to the server" do
conn = PG::connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
conn.async_exec("SET SERVER ROLE to 'primary'")
conn.async_exec("BEGIN")
conn.close
expect(processes.primary.count_query("ROLLBACK")).to eq(1)
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"))
expect(server_conn.server_version).not_to eq(0)
server_conn.close
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
expect(admin_conn.server_version).not_to eq(0)
admin_conn.close
end
end
describe "State clearance" do
context "session mode" do
let(:processes) { Helpers::Pgcat.single_shard_setup("sharded_db", 5, "session") }
it "Clears state before connection checkin" do
# Both modes of operation should not raise
# ERROR: prepared statement "prepared_q" already exists
15.times do
conn = PG::connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
conn.async_exec("PREPARE prepared_q (int) AS SELECT $1")
conn.close
end
conn = PG::connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
initial_value = conn.async_exec("SHOW statement_timeout")[0]["statement_timeout"]
conn.async_exec("SET statement_timeout to 1000")
current_value = conn.async_exec("SHOW statement_timeout")[0]["statement_timeout"]
expect(conn.async_exec("SHOW statement_timeout")[0]["statement_timeout"]).to eq("1s")
conn.close
end
it "Does not send DISCARD ALL unless necessary" do
10.times do
conn = PG::connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
conn.async_exec("SET SERVER ROLE to 'primary'")
conn.async_exec("SELECT 1")
conn.close
end
expect(processes.primary.count_query("DISCARD ALL")).to eq(0)
10.times do
conn = PG::connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
conn.async_exec("SET SERVER ROLE to 'primary'")
conn.async_exec("SELECT 1")
conn.async_exec("SET statement_timeout to 5000")
conn.close
end
expect(processes.primary.count_query("DISCARD ALL")).to eq(10)
end
end
context "transaction mode" do
let(:processes) { Helpers::Pgcat.single_shard_setup("sharded_db", 5, "transaction") }
it "Clears state before connection checkin" do
# Both modes of operation should not raise
# ERROR: prepared statement "prepared_q" already exists
15.times do
conn = PG::connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
conn.async_exec("PREPARE prepared_q (int) AS SELECT $1")
conn.close
end
15.times do
conn = PG::connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
conn.prepare("prepared_q", "SELECT $1")
conn.close
end
end
it "Does not send DISCARD ALL unless necessary" do
10.times do
conn = PG::connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
conn.async_exec("SET SERVER ROLE to 'primary'")
conn.async_exec("SELECT 1")
conn.exec_params("SELECT $1", [1])
conn.close
end
expect(processes.primary.count_query("DISCARD ALL")).to eq(0)
10.times do
conn = PG::connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
conn.async_exec("SET SERVER ROLE to 'primary'")
conn.async_exec("SELECT 1")
conn.async_exec("SET statement_timeout to 5000")
conn.close
end
expect(processes.primary.count_query("DISCARD ALL")).to eq(10)
end
end
context "transaction mode with transactions" do
let(:processes) { Helpers::Pgcat.single_shard_setup("sharded_db", 5, "transaction") }
it "Does not clear set statement state when declared in a transaction" do
10.times do
conn = PG::connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
conn.async_exec("SET SERVER ROLE to 'primary'")
conn.async_exec("BEGIN")
conn.async_exec("SET statement_timeout to 1000")
conn.async_exec("COMMIT")
conn.close
end
expect(processes.primary.count_query("DISCARD ALL")).to eq(0)
10.times do
conn = PG::connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
conn.async_exec("SET SERVER ROLE to 'primary'")
conn.async_exec("BEGIN")
conn.async_exec("SET LOCAL statement_timeout to 1000")
conn.async_exec("COMMIT")
conn.close
end
expect(processes.primary.count_query("DISCARD ALL")).to eq(0)
end
end
end
end

View File

@@ -1,81 +0,0 @@
# frozen_string_literal: true
require_relative 'spec_helper'
describe "Routing" do
let(:processes) { Helpers::Pgcat.single_shard_setup("sharded_db", 5) }
after do
processes.all_databases.map(&:reset)
processes.pgcat.shutdown
end
describe "SET ROLE" do
context "primary" do
it "routes queries only to primary" do
conn = PG.connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
conn.async_exec("SET SERVER ROLE to 'primary'")
query_count = 30
failed_count = 0
query_count.times do
conn.async_exec("SELECT 1 + 2")
rescue
failed_count += 1
end
expect(failed_count).to eq(0)
processes.replicas.map(&:count_select_1_plus_2).each do |instance_share|
expect(instance_share).to eq(0)
end
expect(processes.primary.count_select_1_plus_2).to eq(query_count)
end
end
context "replica" do
it "routes queries only to replicas" do
conn = PG.connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
conn.async_exec("SET SERVER ROLE to 'replica'")
expected_share = QUERY_COUNT / processes.replicas.count
failed_count = 0
QUERY_COUNT.times do
conn.async_exec("SELECT 1 + 2")
rescue
failed_count += 1
end
expect(failed_count).to eq(0)
processes.replicas.map(&:count_select_1_plus_2).each do |instance_share|
expect(instance_share).to be_within(expected_share * MARGIN_OF_ERROR).of(expected_share)
end
expect(processes.primary.count_select_1_plus_2).to eq(0)
end
end
context "any" do
it "routes queries to all instances" do
conn = PG.connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
conn.async_exec("SET SERVER ROLE to 'any'")
expected_share = QUERY_COUNT / processes.all_databases.count
failed_count = 0
QUERY_COUNT.times do
conn.async_exec("SELECT 1 + 2")
rescue
failed_count += 1
end
expect(failed_count).to eq(0)
processes.all_databases.map(&:count_select_1_plus_2).each do |instance_share|
expect(instance_share).to be_within(expected_share * MARGIN_OF_ERROR).of(expected_share)
end
end
end
end
end

View File

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

View File

@@ -1,21 +0,0 @@
# frozen_string_literal: true
require 'pg'
require_relative 'helpers/pgcat_helper'
QUERY_COUNT = 300
MARGIN_OF_ERROR = 0.35
def with_captured_stdout_stderr
sout = STDOUT.clone
serr = STDERR.clone
STDOUT.reopen("/tmp/out.txt", "w+")
STDERR.reopen("/tmp/err.txt", "w+")
STDOUT.sync = true
STDERR.sync = true
yield
return File.read('/tmp/out.txt'), File.read('/tmp/err.txt')
ensure
STDOUT.reopen(sout)
STDERR.reopen(serr)
end

View File

@@ -1,6 +1,93 @@
# frozen_string_literal: true
require 'pg'
require 'active_record'
require 'pg'
require 'toml'
$stdout.sync = true
$stderr.sync = true
class ConfigEditor
def initialize
@original_config_text = File.read('../../.circleci/pgcat.toml')
text_to_load = @original_config_text.gsub("5432", "\"5432\"")
@original_configs = TOML.load(text_to_load)
end
def original_configs
TOML.load(TOML::Generator.new(@original_configs).body)
end
def with_modified_configs(new_configs)
text_to_write = TOML::Generator.new(new_configs).body
text_to_write = text_to_write.gsub("\"5432\"", "5432")
File.write('../../.circleci/pgcat.toml', text_to_write)
yield
ensure
File.write('../../.circleci/pgcat.toml', @original_config_text)
end
end
def with_captured_stdout_stderr
sout = STDOUT.clone
serr = STDERR.clone
STDOUT.reopen("/tmp/out.txt", "w+")
STDERR.reopen("/tmp/err.txt", "w+")
STDOUT.sync = true
STDERR.sync = true
yield
return File.read('/tmp/out.txt'), File.read('/tmp/err.txt')
ensure
STDOUT.reopen(sout)
STDERR.reopen(serr)
end
def test_extended_protocol_pooler_errors
admin_conn = PG::connect("postgres://admin_user:admin_pass@127.0.0.1:6432/pgcat")
conf_editor = ConfigEditor.new
new_configs = conf_editor.original_configs
# shorter timeouts
new_configs["general"]["connect_timeout"] = 500
new_configs["general"]["ban_time"] = 1
new_configs["general"]["shutdown_timeout"] = 1
new_configs["pools"]["sharded_db"]["users"]["0"]["pool_size"] = 1
new_configs["pools"]["sharded_db"]["users"]["1"]["pool_size"] = 1
conf_editor.with_modified_configs(new_configs) { admin_conn.async_exec("RELOAD") }
conn_str = "postgres://sharding_user:sharding_user@127.0.0.1:6432/sharded_db"
10.times do
Thread.new do
conn = PG::connect(conn_str)
conn.async_exec("SELECT pg_sleep(5)") rescue PG::SystemError
ensure
conn&.close
end
end
sleep(0.5)
conn_under_test = PG::connect(conn_str)
stdout, stderr = with_captured_stdout_stderr do
5.times do |i|
conn_under_test.async_exec("SELECT 1") rescue PG::SystemError
conn_under_test.exec_params("SELECT #{i} + $1", [i]) rescue PG::SystemError
sleep 1
end
end
raise StandardError, "Libpq got unexpected messages while idle" if stderr.include?("arrived from server while idle")
puts "Pool checkout errors not breaking clients passed"
ensure
sleep 1
admin_conn.async_exec("RELOAD") # Reset state
conn_under_test&.close
end
test_extended_protocol_pooler_errors
# Uncomment these two to see all queries.
# ActiveRecord.verbose_query_logs = true
@@ -111,3 +198,68 @@ begin
rescue ActiveRecord::StatementInvalid
puts 'OK'
end
# Test evil clients
def poorly_behaved_client
conn = PG::connect("postgres://sharding_user:sharding_user@127.0.0.1:6432/sharded_db?application_name=testing_pgcat")
conn.async_exec 'BEGIN'
conn.async_exec 'SELECT 1'
conn.close
puts 'Bad client ok'
end
25.times do
poorly_behaved_client
end
def test_server_parameters
server_conn = PG::connect("postgres://sharding_user:sharding_user@127.0.0.1:6432/sharded_db?application_name=testing_pgcat")
raise StandardError, "Bad server version" if server_conn.server_version == 0
server_conn.close
admin_conn = PG::connect("postgres://admin_user:admin_pass@127.0.0.1:6432/pgcat")
raise StandardError, "Bad server version" if admin_conn.server_version == 0
admin_conn.close
puts 'Server parameters ok'
end
def test_reload_pool_recycling
admin_conn = PG::connect("postgres://admin_user:admin_pass@127.0.0.1:6432/pgcat")
server_conn = PG::connect("postgres://sharding_user:sharding_user@127.0.0.1:6432/sharded_db?application_name=testing_pgcat")
server_conn.async_exec("BEGIN")
conf_editor = ConfigEditor.new
new_configs = conf_editor.original_configs
# swap shards
new_configs["pools"]["sharded_db"]["shards"]["0"]["database"] = "shard1"
new_configs["pools"]["sharded_db"]["shards"]["1"]["database"] = "shard0"
raise StandardError if server_conn.async_exec("SELECT current_database();")[0]["current_database"] != 'shard0'
conf_editor.with_modified_configs(new_configs) { admin_conn.async_exec("RELOAD") }
raise StandardError if server_conn.async_exec("SELECT current_database();")[0]["current_database"] != 'shard0'
server_conn.async_exec("COMMIT;")
# Transaction finished, client should get new configs
raise StandardError if server_conn.async_exec("SELECT current_database();")[0]["current_database"] != 'shard1'
server_conn.close()
# New connection should get new configs
server_conn = PG::connect("postgres://sharding_user:sharding_user@127.0.0.1:6432/sharded_db?application_name=testing_pgcat")
raise StandardError if server_conn.async_exec("SELECT current_database();")[0]["current_database"] != 'shard1'
ensure
admin_conn.async_exec("RELOAD") # Go back to old state
admin_conn.close
server_conn.close
puts "Pool Recycling okay!"
end
test_reload_pool_recycling

View File

@@ -70,35 +70,23 @@ GRANT CONNECT ON DATABASE shard2 TO other_user;
GRANT CONNECT ON DATABASE some_db TO simple_user;
\c shard0
CREATE EXTENSION IF NOT EXISTS pg_stat_statements;
GRANT EXECUTE ON FUNCTION pg_stat_statements_reset TO sharding_user;
GRANT ALL ON SCHEMA public TO sharding_user;
GRANT ALL ON TABLE data TO sharding_user;
GRANT ALL ON SCHEMA public TO other_user;
GRANT ALL ON TABLE data TO other_user;
GRANT EXECUTE ON FUNCTION pg_stat_statements_reset TO other_user;
\c shard1
CREATE EXTENSION IF NOT EXISTS pg_stat_statements;
GRANT EXECUTE ON FUNCTION pg_stat_statements_reset TO sharding_user;
GRANT ALL ON SCHEMA public TO sharding_user;
GRANT ALL ON TABLE data TO sharding_user;
GRANT ALL ON SCHEMA public TO other_user;
GRANT ALL ON TABLE data TO other_user;
GRANT EXECUTE ON FUNCTION pg_stat_statements_reset TO other_user;
\c shard2
CREATE EXTENSION IF NOT EXISTS pg_stat_statements;
GRANT EXECUTE ON FUNCTION pg_stat_statements_reset TO sharding_user;
GRANT ALL ON SCHEMA public TO sharding_user;
GRANT ALL ON TABLE data TO sharding_user;
GRANT ALL ON SCHEMA public TO other_user;
GRANT ALL ON TABLE data TO other_user;
GRANT EXECUTE ON FUNCTION pg_stat_statements_reset TO other_user;
\c some_db
CREATE EXTENSION IF NOT EXISTS pg_stat_statements;
GRANT EXECUTE ON FUNCTION pg_stat_statements_reset TO simple_user;
GRANT ALL ON SCHEMA public TO simple_user;
GRANT ALL ON TABLE data TO simple_user;

View File

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