Compare commits

...

38 Commits

Author SHA1 Message Date
Mostafa
fda467082c Bump rust version on CI image 2025-02-27 14:34:27 -06:00
(Alan) Jia Lin Yuan
4fdf73c599 Bump Rust Version on Dockerfile from 1.79.0 to 1.81 (#912) 2025-02-27 14:30:45 -06:00
Mostafa
3349cecc18 Add checkout_failure_limit config/feature (#911)
In a high availability deployment of PgCat, it is possible that a client may land on a container of PgCat that is very busy with clients and as such the new client might be perpetually stuck in checkout failure loop because all connections are used by other clients. This is specially true in session mode pools with long-lived client connections (e.g. FDW connections).

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

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

I was able to reproduce this issue in a control environment and was able to show this PR is able to fix it.
2025-02-27 13:17:00 -06:00
Alex Kesling
f8e2fcd0ed s/Iniitalize/Initialize/ (#897) 2025-01-09 11:59:42 -08:00
Nadav Shatz
3202f5685b Add DB activity based routing (#864) 2024-12-22 05:23:57 -06:00
Gabriel Simmer
b37d105184 chore(deps): bump sqlparser from 0.41.0 to 0.52.0 (#870)
* chore(deps): bump sqlparser from 0.41.0 to 0.52.0

Bumps [sqlparser](https://github.com/apache/datafusion-sqlparser-rs) from 0.41.0 to 0.52.0.
- [Changelog](https://github.com/apache/datafusion-sqlparser-rs/blob/main/CHANGELOG.md)
- [Commits](https://github.com/apache/datafusion-sqlparser-rs/commits)

---
updated-dependencies:
- dependency-name: sqlparser
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>

* bump

* Update to latest sqlparser version

---------

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Mostafa <mostafa.mohmmed@gmail.com>
2024-11-23 07:25:37 -06:00
Vitalii Tverdokhlib
32f4752daf DOCS: small typo LIMT (Update README.md) (#866) 2024-11-17 06:51:05 -06:00
Mostafa
3796e26402 Fix contact info for Helm chart (#861)
* Fix contact info for Helm chart
2024-11-11 09:24:17 -06:00
Mostafa
0ee59c0c40 Another no-op helm release (#853) 2024-11-08 06:07:12 -06:00
Mostafa
b61d2cc6f0 Use main branch for helm chart releases (#852) 2024-11-08 06:04:42 -06:00
Jose Fernández
c11418c083 Revert "Do not unban replicas if a primary is available" (#850)
Revert "Do not unban replicas if a primary is available (#843)"

This reverts commit cdcfa99fb9.
2024-11-07 22:00:43 +01:00
Jose Fernández
c9544bdff2 Fix default_role being ignored when query_parser_enabled was false (#847)
Fix default_role being ignored when query_parser_enabled was false
2024-11-07 11:11:49 -06:00
Jose Fernández
cdcfa99fb9 Do not unban replicas if a primary is available (#843)
Add `unban_replicas_when_all_banned` to control unbanning replicas behavior.
2024-11-07 11:11:11 -06:00
Víťa Tauer
f27dc6b483 Fixing invalid setting name in pgcat.toml (#849) 2024-11-07 06:17:09 -06:00
Mostafa
326efc22b3 Another no-op release for helm (#845)
Another no-op release
2024-11-02 18:05:41 -05:00
Mostafa
01c6afb2e5 Attempt a helm chart release (#844)
Attempt a release

Co-authored-by: Mostafa <no_reply@github.com>
2024-11-02 11:55:18 -05:00
Nicolas Vanelslande
a68071dd28 Bump bb8 from 0.8.1 to 0.8.6 (#709)
* Update bb8 to 0.8.6

To get https://github.com/djc/bb8/pull/186 and https://github.com/djc/bb8/pull/189
which fix potential deadlocks (https://github.com/djc/bb8/issues/154).

Also, this (https://github.com/djc/bb8/pull/225) was needed to prevent a connection
leak which was conveniently spotted in our integration tests.

* Ignore ./.bundle (created by dev console)

---------

Co-authored-by: Jose Fernandez (magec) <joseferper@gmail.com>
2024-10-28 06:49:36 -05:00
Mostafa
c27d801abf Rename a couple of variables (#839) 2024-10-23 06:38:07 -05:00
Javier Goday
186e72298f #829: read/write splitting on CTE mutable statements (#835) 2024-10-23 06:20:04 -05:00
Sebastian Serth
3935366d86 End Prometheus stats with a new line separator (#826)
End prometheus stats with a new line separator

According to the [OpenMetrics specification](https://github.com/OpenObservability/OpenMetrics/blob/main/specification/OpenMetrics.md#overall-structure), each line MUST end with `\n`. Previously, the last line was not ending with `\n`, so that strict parsers had issues reading the Prometheus stats.
2024-09-22 17:14:04 -05:00
Sean McGivern
b575935b1d Improve documentation for connect_timeout and add min_pool_size (#822)
Currently, `connect_timeout` sounds like it should be for connections to
the Postgres server. It's actually used for obtaining a connection from
the pool.
2024-09-18 06:56:17 -05:00
Shijun Wang
efbab1c333 Helm chart improvements including allowing user password to be pulled from K8s secret (#753)
* Make user min_pool_size configurable

* Set user server_lifetime only if specified

* Increment chart version

* Use default instea of or

* Allow enabling server_tls

* statement_timeout default value

* Allow pulling password from existing secret

---------

Co-authored-by: Mostafa Abdelraouf <mostafa.mohmmed@gmail.com>
2024-09-14 09:57:17 -05:00
Mostafa Abdelraouf
9f12d7958e Fix Ruby tests (#819)
Build is failing with this error

Downloading activerecord-3.2.14 revealed dependencies not in the API or the
lockfile (activesupport (= 3.2.14), activemodel (= 3.2.14), arel (~> 3.0.2),
tzinfo (~> 0.3.29)).
Either installing with `--full-index` or running `bundle update activerecord`
should fix the problem.

After ActiveSupport was updated.

This PR fixes that
2024-09-13 20:02:38 -05:00
dependabot[bot]
e6634ef461 chore(deps): bump activesupport from 7.0.4.1 to 7.0.7.1 in /tests/ruby (#804)
Bumps [activesupport](https://github.com/rails/rails) from 7.0.4.1 to 7.0.7.1.
- [Release notes](https://github.com/rails/rails/releases)
- [Changelog](https://github.com/rails/rails/blob/v7.2.1/activesupport/CHANGELOG.md)
- [Commits](https://github.com/rails/rails/compare/v7.0.4.1...v7.0.7.1)

---
updated-dependencies:
- dependency-name: activesupport
  dependency-type: indirect
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2024-09-13 19:43:26 -05:00
dependabot[bot]
dab2e58647 chore(deps): bump helm/chart-releaser-action from 1.5.0 to 1.6.0 (#812)
Bumps [helm/chart-releaser-action](https://github.com/helm/chart-releaser-action) from 1.5.0 to 1.6.0.
- [Release notes](https://github.com/helm/chart-releaser-action/releases)
- [Commits](be16258da8...a917fd15b2)

---
updated-dependencies:
- dependency-name: helm/chart-releaser-action
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2024-09-13 19:41:25 -05:00
dependabot[bot]
4aaa4378cf chore(deps): bump rexml from 3.2.8 to 3.3.6 in /tests/ruby (#803)
Bumps [rexml](https://github.com/ruby/rexml) from 3.2.8 to 3.3.6.
- [Release notes](https://github.com/ruby/rexml/releases)
- [Changelog](https://github.com/ruby/rexml/blob/master/NEWS.md)
- [Commits](https://github.com/ruby/rexml/compare/v3.2.8...v3.3.6)

---
updated-dependencies:
- dependency-name: rexml
  dependency-type: indirect
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2024-09-13 19:19:30 -05:00
Andrew Jackson
670311daf9 Implement Trust Authentication (#805)
* Implement Trust Authentication

* Remove remaining LDAP stuff

* Reverted LDAP changes, Cleaned up tests

---------

Co-authored-by: Andrew Jackson <andrewjackson2988@gmail.com>
Co-authored-by: CommanderKeynes <andrewjackson947@gmail.coma>
2024-09-10 09:29:45 -05:00
dependabot[bot]
b9ec7f8036 chore(deps): bump actions/setup-python from 4.1.0 to 5.1.0 (#715)
Bumps [actions/setup-python](https://github.com/actions/setup-python) from 4.1.0 to 5.1.0.
- [Release notes](https://github.com/actions/setup-python/releases)
- [Commits](https://github.com/actions/setup-python/compare/v4.1.0...v5.1.0)

---
updated-dependencies:
- dependency-name: actions/setup-python
  dependency-type: direct:production
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2024-09-07 12:21:21 -05:00
dependabot[bot]
d91d23848b chore(deps): bump helm/kind-action from 1.7.0 to 1.10.0 (#732)
Bumps [helm/kind-action](https://github.com/helm/kind-action) from 1.7.0 to 1.10.0.
- [Release notes](https://github.com/helm/kind-action/releases)
- [Commits](https://github.com/helm/kind-action/compare/v1.7.0...v1.10.0)

---
updated-dependencies:
- dependency-name: helm/kind-action
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2024-09-07 12:20:38 -05:00
dependabot[bot]
bbbc01a467 chore(deps): bump rexml from 3.2.5 to 3.2.8 in /tests/ruby (#743)
Bumps [rexml](https://github.com/ruby/rexml) from 3.2.5 to 3.2.8.
- [Release notes](https://github.com/ruby/rexml/releases)
- [Changelog](https://github.com/ruby/rexml/blob/master/NEWS.md)
- [Commits](https://github.com/ruby/rexml/compare/v3.2.5...v3.2.8)

---
updated-dependencies:
- dependency-name: rexml
  dependency-type: indirect
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2024-09-07 12:20:01 -05:00
Sebastian Serth
9bb71ede9d Automatically build deb package on a new version tag (#801)
In #796, I noticed that the deb package was not build since an automation was missing.

With this PR, I add the missing automation.

I tested the workflow in my repo...

    when starting the workflow manually: https://github.com/MrSerth/pgcat/actions/runs/10737879151/job/29780286094
    when drafting a new release: https://github.com/MrSerth/pgcat/actions/runs/10737835796/job/29780146212

Obviously, both workflows failed since I cannot upload to the APT repo. However, the version substitution for the workflow is working correctly (as shown when collapsing the first line of the "Build and release package" step).
2024-09-06 09:11:52 -05:00
Sebastian Serth
88b2afb19b Automatically start systemd service if config file is present (#800)
Previously, upgrading the deb package stopped the service but didn't reenable it after a successful upgrade. This made upgrading the package more difficult and required a second step to restart the service. With this commit, the systemd service is automatically started when the default config file is present.
2024-09-06 09:07:01 -05:00
Mostafa Abdelraouf
f0865ca616 Improve Prometheus exporter output (#795)
* Prometheus metrics updates:

 * Add username label to deconflict metrics that would otherwise
   have duplicate labels across different pools.
 * Group metrics by name and only print HELP and TYPE once per
   metric name.
 * Sort labels for a deterministic output.

---------

Co-authored-by: Curtis Myzie <curtis.myzie@gmail.com>
Co-authored-by: Towhid Khan
2024-09-05 08:58:18 -05:00
Andrew Jackson
7d047c6c19 Implemented python tests with pytest (#790)
Currently the python tests act as scripts. A lot of output is generated to stdout which makes it very hard to figure out where problems were. Also if you want to run only a single test you basically need to comment out code in order to accomplish this.

This PR modifies the python tests to us the pytest python testing framework. This framework allows individual tests to be targeted via the command line, without touching the source code. It also suppressed stdout by default making the test output much easier to read. Also after the tests run it will provide a summary of what failed, what succeded, etc.


Co-authored-by: CommanderKeynes <andrewjackson947@gmail.coma>
Co-authored-by: Andrew Jackson <andrewjackson2988@gmail.com>
2024-09-05 08:16:45 -05:00
Andrew Jackson
f73d15f82c Fix CI script to allow consecutive runs locally (#793)
Co-authored-by: CommanderKeynes <andrewjackson947@gmail.coma>
2024-09-05 08:01:33 -05:00
Mostafa Abdelraouf
69af6cc5e5 Make iterating on integration tests easier (#789)
Writing and iterating on integration tests are cumbersome, having to wait 10 minutes for the test-suite to run just to see if your test works or not is unacceptable.

In this PR, I added a detailed workflow for writing tests that should shorten the feedback cycle of modifying tests to be as low as a few seconds.

It will involve opening a shell into a long-lived container that has all the setup and dependencies necessary and then running your desired tests directly there. I added a convenience script that bootstraps the environment and then opens an interactive shell into the container and you can then run tests immediately in an environment that is more or less identical to what we have running in CircleCI
2024-09-03 11:15:53 -05:00
Mostafa Abdelraouf
ca34597002 Fix broken integration test #740 (#787) 2024-08-31 17:15:13 -05:00
Mostafa Abdelraouf
2def40ea6a Add test case for issue 776 (#786)
I am adding a tiny test that uses the SQL statement that was reported to break an older version of SQL parser library

#776
2024-08-31 10:52:33 -05:00
39 changed files with 2184 additions and 659 deletions

View File

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

View File

@@ -26,6 +26,7 @@ PGPASSWORD=sharding_user pgbench -h 127.0.0.1 -U sharding_user shard1 -i
PGPASSWORD=sharding_user pgbench -h 127.0.0.1 -U sharding_user shard2 -i
# Start Toxiproxy
kill -9 $(pgrep toxiproxy) || true
LOG_LEVEL=error toxiproxy-server &
sleep 1
@@ -106,7 +107,7 @@ 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
pytest || exit 1
#
@@ -177,3 +178,6 @@ killall pgcat -s SIGINT
# Allow for graceful shutdown
sleep 1
kill -9 $(pgrep toxiproxy)
sleep 1

View File

@@ -22,7 +22,7 @@ jobs:
# Python is required because `ct lint` runs Yamale (https://github.com/23andMe/Yamale) and
# yamllint (https://github.com/adrienverge/yamllint) which require Python
- name: Set up Python
uses: actions/setup-python@v4.1.0
uses: actions/setup-python@v5.1.0
with:
python-version: 3.7
@@ -43,7 +43,7 @@ jobs:
run: ct lint --config ct.yaml
- name: Create kind cluster
uses: helm/kind-action@v1.7.0
uses: helm/kind-action@v1.10.0
if: steps.list-changed.outputs.changed == 'true'
- name: Run chart-testing (install)

View File

@@ -32,7 +32,7 @@ jobs:
version: v3.13.0
- name: Run chart-releaser
uses: helm/chart-releaser-action@be16258da8010256c6e82849661221415f031968 # v1.5.0
uses: helm/chart-releaser-action@a917fd15b20e8b64b94d9158ad54cd6345335584 # v1.6.0
with:
charts_dir: charts
config: cr.yaml

View File

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

4
.gitignore vendored
View File

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

View File

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

View File

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

300
Cargo.lock generated
View File

@@ -132,7 +132,7 @@ checksum = "16e62a023e7c117e27523144c5d2459f4397fcc3cab0085af8e2224f643a0193"
dependencies = [
"proc-macro2",
"quote",
"syn 2.0.26",
"syn 2.0.87",
]
[[package]]
@@ -143,7 +143,7 @@ checksum = "a564d521dd56509c4c47480d00b80ee55f7e385ae48db5744c67ad50c92d2ebf"
dependencies = [
"proc-macro2",
"quote",
"syn 2.0.26",
"syn 2.0.87",
]
[[package]]
@@ -192,12 +192,11 @@ checksum = "604178f6c5c21f02dc555784810edfb88d34ac2c73b2eae109655649ee73ce3d"
[[package]]
name = "bb8"
version = "0.8.1"
version = "0.8.6"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "98b4b0f25f18bcdc3ac72bdb486ed0acf7e185221fd4dc985bc15db5800b0ba2"
checksum = "d89aabfae550a5c44b43ab941844ffcd2e993cb6900b342debf59e9ea74acdb8"
dependencies = [
"async-trait",
"futures-channel",
"futures-util",
"parking_lot",
"tokio",
@@ -230,6 +229,12 @@ version = "3.13.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a3e2c3daef883ecc1b5d58c15adae93470a91d425f3532ba1695849656af3fc1"
[[package]]
name = "bytecount"
version = "0.6.8"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "5ce89b21cab1437276d2650d57e971f9d548a2d9037cc231abdc0562b97498ce"
[[package]]
name = "byteorder"
version = "1.4.3"
@@ -242,6 +247,37 @@ version = "1.4.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "89b2fd2a0dcf38d7971e2194b6b6eebab45ae01067456a7fd93d5547a61b70be"
[[package]]
name = "camino"
version = "1.1.9"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8b96ec4966b5813e2c0507c1f86115c8c5abaadc3980879c3424042a02fd1ad3"
dependencies = [
"serde",
]
[[package]]
name = "cargo-platform"
version = "0.1.8"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "24b1f0365a6c6bb4020cd05806fd0d33c44d38046b8bd7f0e40814b9763cabfc"
dependencies = [
"serde",
]
[[package]]
name = "cargo_metadata"
version = "0.14.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4acbb09d9ee8e23699b9634375c72795d095bf268439da88562cf9b501f181fa"
dependencies = [
"camino",
"cargo-platform",
"semver",
"serde",
"serde_json",
]
[[package]]
name = "cc"
version = "1.0.79"
@@ -301,7 +337,7 @@ dependencies = [
"heck",
"proc-macro2",
"quote",
"syn 2.0.26",
"syn 2.0.87",
]
[[package]]
@@ -331,6 +367,21 @@ dependencies = [
"libc",
]
[[package]]
name = "crossbeam-channel"
version = "0.5.13"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "33480d6946193aa8033910124896ca395333cae7e2d1113d1fef6c3272217df2"
dependencies = [
"crossbeam-utils",
]
[[package]]
name = "crossbeam-utils"
version = "0.8.20"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "22ec99545bb0ed0ea7bb9b8e1e9122ea386ff8a48c0922e43f36d45ab09e0e80"
[[package]]
name = "crypto-common"
version = "0.1.6"
@@ -341,6 +392,19 @@ dependencies = [
"typenum",
]
[[package]]
name = "dashmap"
version = "5.5.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "978747c1d849a7d2ee5e8adc0159961c48fb7e5db2f06af6723b80123bb53856"
dependencies = [
"cfg-if",
"hashbrown",
"lock_api",
"once_cell",
"parking_lot_core",
]
[[package]]
name = "data-encoding"
version = "2.4.0"
@@ -403,6 +467,15 @@ dependencies = [
"libc",
]
[[package]]
name = "error-chain"
version = "0.12.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "2d2f06b9cac1506ece98fe3231e3cc9c4410ec3d5b1f24ae1c8946f0742cdefc"
dependencies = [
"version_check",
]
[[package]]
name = "exitcode"
version = "1.1.2"
@@ -415,6 +488,12 @@ version = "0.2.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4443176a9f2c162692bd3d352d745ef9413eec5782a80d8fd6f8a1ac692a07f7"
[[package]]
name = "fastrand"
version = "2.1.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e8c02a5121d4ea3eb16a80748c74f5549a5665e4c21333c6098f283870fbdea6"
[[package]]
name = "fnv"
version = "1.0.7"
@@ -486,7 +565,7 @@ checksum = "89ca545a94061b6365f2c7355b4b32bd20df3ff95f02da9329b34ccc3bd6ee72"
dependencies = [
"proc-macro2",
"quote",
"syn 2.0.26",
"syn 2.0.87",
]
[[package]]
@@ -546,6 +625,12 @@ version = "0.27.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "b6c80984affa11d98d1b88b66ac8853f143217b399d3c74116778ff8fdb4ed2e"
[[package]]
name = "glob"
version = "0.3.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d2fabcfbdc87f4758337ca535fb41a6d701b65693ce38287d856d1674551ec9b"
[[package]]
name = "h2"
version = "0.4.6"
@@ -919,6 +1004,21 @@ dependencies = [
"autocfg",
]
[[package]]
name = "mini-moka"
version = "0.10.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "c325dfab65f261f386debee8b0969da215b3fa0037e74c8a1234db7ba986d803"
dependencies = [
"crossbeam-channel",
"crossbeam-utils",
"dashmap",
"skeptic",
"smallvec",
"tagptr",
"triomphe",
]
[[package]]
name = "miniz_oxide"
version = "0.7.1"
@@ -993,9 +1093,9 @@ dependencies = [
[[package]]
name = "once_cell"
version = "1.18.0"
version = "1.20.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "dd8b5dd2ae5ed71462c540258bedcb51965123ad7e7ccf4b9a8cafaa4a63576d"
checksum = "1261fe7e33c73b354eab43b1273a57c8f967d0391e80353e51f764ac02cf6775"
[[package]]
name = "overload"
@@ -1034,7 +1134,7 @@ checksum = "9b2a4787296e9989611394c33f193f676704af1686e70b8f8033ab5ba9a35a94"
[[package]]
name = "pgcat"
version = "1.2.0"
version = "1.3.0"
dependencies = [
"arc-swap",
"async-trait",
@@ -1056,6 +1156,7 @@ dependencies = [
"log",
"lru",
"md-5",
"mini-moka",
"nix",
"num_cpus",
"once_cell",
@@ -1070,6 +1171,7 @@ dependencies = [
"serde",
"serde_derive",
"serde_json",
"serial_test",
"sha-1",
"sha2",
"socket2 0.4.9",
@@ -1115,7 +1217,7 @@ dependencies = [
"phf_shared",
"proc-macro2",
"quote",
"syn 2.0.26",
"syn 2.0.87",
]
[[package]]
@@ -1144,7 +1246,7 @@ checksum = "ec2e072ecce94ec471b13398d5402c188e76ac03cf74dd1a975161b23a3f6d9c"
dependencies = [
"proc-macro2",
"quote",
"syn 2.0.26",
"syn 2.0.87",
]
[[package]]
@@ -1185,13 +1287,24 @@ checksum = "5b40af805b3121feab8a3c29f04d8ad262fa8e0561883e7653e024ae4479e6de"
[[package]]
name = "proc-macro2"
version = "1.0.66"
version = "1.0.89"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "18fb31db3f9bddb2ea821cde30a9f70117e3f119938b5ee630b7403aa6e2ead9"
checksum = "f139b0662de085916d1fb67d2b4169d1addddda1919e696f3252b740b629986e"
dependencies = [
"unicode-ident",
]
[[package]]
name = "pulldown-cmark"
version = "0.9.6"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "57206b407293d2bcd3af849ce869d52068623f19e1b5ff8e8778e3309439682b"
dependencies = [
"bitflags 2.3.3",
"memchr",
"unicase",
]
[[package]]
name = "quick-error"
version = "1.2.3"
@@ -1200,9 +1313,9 @@ checksum = "a1d01941d82fa2ab50be1e79e6714289dd7cde78eba4c074bc5a4374f650dfe0"
[[package]]
name = "quote"
version = "1.0.31"
version = "1.0.37"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "5fe8a65d69dd0808184ebb5f836ab526bb259db23c657efa38711b1072ee47f0"
checksum = "b5b9d34b8991d19d98081b46eacdd8eb58c6f2b201139f7c5f643cc155a633af"
dependencies = [
"proc-macro2",
]
@@ -1381,6 +1494,24 @@ version = "1.0.15"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "1ad4cc8da4ef723ed60bced201181d83791ad433213d8c24efffda1eec85d741"
[[package]]
name = "same-file"
version = "1.0.6"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "93fc1dc3aaa9bfed95e02e6eadabb4baf7e3078b0bd1b4d7b6b0b68378900502"
dependencies = [
"winapi-util",
]
[[package]]
name = "scc"
version = "2.2.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d8d25269dd3a12467afe2e510f69fb0b46b698e5afb296b59f2145259deaf8e8"
dependencies = [
"sdd",
]
[[package]]
name = "scopeguard"
version = "1.2.0"
@@ -1398,23 +1529,38 @@ dependencies = [
]
[[package]]
name = "serde"
version = "1.0.171"
name = "sdd"
version = "3.0.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "30e27d1e4fd7659406c492fd6cfaf2066ba8773de45ca75e855590f856dc34a9"
checksum = "49c1eeaf4b6a87c7479688c6d52b9f1153cedd3c489300564f932b065c6eab95"
[[package]]
name = "semver"
version = "1.0.23"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "61697e0a1c7e512e84a621326239844a24d8207b4669b41bc18b32ea5cbf988b"
dependencies = [
"serde",
]
[[package]]
name = "serde"
version = "1.0.214"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f55c3193aca71c12ad7890f1785d2b73e1b9f63a0bbc353c08ef26fe03fc56b5"
dependencies = [
"serde_derive",
]
[[package]]
name = "serde_derive"
version = "1.0.171"
version = "1.0.214"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "389894603bd18c46fa56231694f8d827779c0951a667087194cf9de94ed24682"
checksum = "de523f781f095e28fa605cdce0f8307e451cc0fd14e2eb4cd2e98a355b147766"
dependencies = [
"proc-macro2",
"quote",
"syn 2.0.26",
"syn 2.0.87",
]
[[package]]
@@ -1437,6 +1583,31 @@ dependencies = [
"serde",
]
[[package]]
name = "serial_test"
version = "3.1.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4b4b487fe2acf240a021cf57c6b2b4903b1e78ca0ecd862a71b71d2a51fed77d"
dependencies = [
"futures",
"log",
"once_cell",
"parking_lot",
"scc",
"serial_test_derive",
]
[[package]]
name = "serial_test_derive"
version = "3.1.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "82fe9db325bcef1fbcde82e078a5cc4efdf787e96b3b9cf45b50b529f2083d67"
dependencies = [
"proc-macro2",
"quote",
"syn 2.0.87",
]
[[package]]
name = "sha-1"
version = "0.10.1"
@@ -1483,6 +1654,21 @@ version = "0.3.10"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "7bd3e3206899af3f8b12af284fafc038cc1dc2b41d1b89dd17297221c5d225de"
[[package]]
name = "skeptic"
version = "0.13.7"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "16d23b015676c90a0f01c197bfdc786c20342c73a0afdda9025adb0bc42940a8"
dependencies = [
"bytecount",
"cargo_metadata",
"error-chain",
"glob",
"pulldown-cmark",
"tempfile",
"walkdir",
]
[[package]]
name = "slab"
version = "0.4.8"
@@ -1526,9 +1712,9 @@ checksum = "6e63cff320ae2c57904679ba7cb63280a3dc4613885beafb148ee7bf9aa9042d"
[[package]]
name = "sqlparser"
version = "0.41.0"
version = "0.52.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "5cc2c25a6c66789625ef164b4c7d2e548d627902280c13710d33da8222169964"
checksum = "9a875d8cd437cc8a97e9aeaeea352ec9a19aea99c23e9effb17757291de80b08"
dependencies = [
"log",
"sqlparser_derive",
@@ -1542,7 +1728,7 @@ checksum = "01b2e185515564f15375f593fb966b5718bc624ba77fe49fa4616ad619690554"
dependencies = [
"proc-macro2",
"quote",
"syn 2.0.26",
"syn 2.0.87",
]
[[package]]
@@ -1586,15 +1772,34 @@ dependencies = [
[[package]]
name = "syn"
version = "2.0.26"
version = "2.0.87"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "45c3457aacde3c65315de5031ec191ce46604304d2446e803d71ade03308d970"
checksum = "25aa4ce346d03a6dcd68dd8b4010bcb74e54e62c90c573f394c46eae99aba32d"
dependencies = [
"proc-macro2",
"quote",
"unicode-ident",
]
[[package]]
name = "tagptr"
version = "0.2.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "7b2093cf4c8eb1e67749a6762251bc9cd836b6fc171623bd0a9d324d37af2417"
[[package]]
name = "tempfile"
version = "3.8.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "cb94d2f3cc536af71caac6b6fcebf65860b347e7ce0cc9ebe8f70d3e521054ef"
dependencies = [
"cfg-if",
"fastrand",
"redox_syscall",
"rustix",
"windows-sys",
]
[[package]]
name = "thiserror"
version = "1.0.43"
@@ -1612,7 +1817,7 @@ checksum = "463fe12d7993d3b327787537ce8dd4dfa058de32fc2b195ef3cde03dc4771e8f"
dependencies = [
"proc-macro2",
"quote",
"syn 2.0.26",
"syn 2.0.87",
]
[[package]]
@@ -1679,7 +1884,7 @@ checksum = "630bdcf245f78637c13ec01ffae6187cca34625e8c63150d424b59e55af2675e"
dependencies = [
"proc-macro2",
"quote",
"syn 2.0.26",
"syn 2.0.87",
]
[[package]]
@@ -1784,7 +1989,7 @@ checksum = "5f4f31f56159e98206da9efd823404b79b6ef3143b4a7ab76e67b1751b25a4ab"
dependencies = [
"proc-macro2",
"quote",
"syn 2.0.26",
"syn 2.0.87",
]
[[package]]
@@ -1839,6 +2044,12 @@ dependencies = [
"tracing-serde",
]
[[package]]
name = "triomphe"
version = "0.1.11"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "859eb650cfee7434994602c3a68b25d77ad9e68c8a6cd491616ef86661382eb3"
[[package]]
name = "trust-dns-proto"
version = "0.22.0"
@@ -1896,6 +2107,12 @@ version = "1.16.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "497961ef93d974e23eb6f433eb5fe1b7930b659f06d12dec6fc44a8f554c0bba"
[[package]]
name = "unicase"
version = "2.8.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "7e51b68083f157f853b6379db119d1c1be0e6e4dec98101079dec41f6f5cf6df"
[[package]]
name = "unicode-bidi"
version = "0.3.13"
@@ -1952,6 +2169,16 @@ version = "0.9.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "49874b5167b65d7193b8aba1567f5c7d93d001cafc34600cee003eda787e483f"
[[package]]
name = "walkdir"
version = "2.5.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "29790946404f91d9c5d06f9874efddea1dc06c5efe94541a7d6863108e3a5e4b"
dependencies = [
"same-file",
"winapi-util",
]
[[package]]
name = "want"
version = "0.3.1"
@@ -1994,7 +2221,7 @@ dependencies = [
"once_cell",
"proc-macro2",
"quote",
"syn 2.0.26",
"syn 2.0.87",
"wasm-bindgen-shared",
]
@@ -2016,7 +2243,7 @@ checksum = "54681b18a46765f095758388f2d0cf16eb8d4169b639ab575a8f5693af210c7b"
dependencies = [
"proc-macro2",
"quote",
"syn 2.0.26",
"syn 2.0.87",
"wasm-bindgen-backend",
"wasm-bindgen-shared",
]
@@ -2068,6 +2295,15 @@ version = "0.4.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ac3b87c63620426dd9b991e5ce0329eff545bccbbb34f3be09ff6fb6ab51b7b6"
[[package]]
name = "winapi-util"
version = "0.1.9"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "cf221c93e13a30d793f7645a0e7762c55d169dbb0a49671918a2319d289b10bb"
dependencies = [
"windows-sys",
]
[[package]]
name = "winapi-x86_64-pc-windows-gnu"
version = "0.4.0"

View File

@@ -1,6 +1,6 @@
[package]
name = "pgcat"
version = "1.2.0"
version = "1.3.0"
edition = "2021"
# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
@@ -8,7 +8,7 @@ edition = "2021"
tokio = { version = "1", features = ["full"] }
bytes = "1"
md-5 = "0.10"
bb8 = "0.8.1"
bb8 = "=0.8.6"
async-trait = "0.1"
rand = "0.8"
chrono = "0.4"
@@ -19,7 +19,7 @@ serde_derive = "1"
regex = "1"
num_cpus = "1"
once_cell = "1"
sqlparser = { version = "0.41", features = ["visitor"] }
sqlparser = { version = "0.52", features = ["visitor"] }
log = "0.4"
arc-swap = "1"
parking_lot = "0.12.1"
@@ -55,6 +55,10 @@ tracing-subscriber = { version = "0.3.17", features = [
"std",
] }
lru = "0.12.0"
mini-moka = "0.10.3"
[target.'cfg(not(target_env = "msvc"))'.dependencies]
jemallocator = "0.5.0"
[dev-dependencies]
serial_test = "*"

View File

@@ -1,4 +1,4 @@
FROM rust:1.79.0-slim-bookworm AS builder
FROM rust:1.81.0-slim-bookworm AS builder
RUN apt-get update && \
apt-get install -y build-essential

View File

@@ -1,4 +1,4 @@
FROM cimg/rust:1.79.0
FROM cimg/rust:1.81.0
COPY --from=sclevine/yj /bin/yj /bin/yj
RUN /bin/yj -h
RUN sudo apt-get update && \

View File

@@ -231,7 +231,7 @@ User.find_by_email("test@example.com")
```sql
-- Grab a bunch of users from shard 1
SET SHARD TO '1';
SELECT * FROM users LIMT 10;
SELECT * FROM users LIMIT 10;
-- Find by id
SET SHARDING KEY TO '1234';

View File

@@ -2,7 +2,7 @@ apiVersion: v2
name: pgcat
description: A Helm chart for PgCat a PostgreSQL pooler and proxy (like PgBouncer) with support for sharding, load balancing, failover and mirroring.
maintainers:
- name: Wildcard
email: support@w6d.io
appVersion: "1.2.0"
version: 0.2.0
- name: PostgresML
email: team@postgresml.org
appVersion: "1.3.0"
version: 0.2.5

View File

@@ -15,6 +15,7 @@ stringData:
connect_timeout = {{ .Values.configuration.general.connect_timeout }}
idle_timeout = {{ .Values.configuration.general.idle_timeout | int }}
server_lifetime = {{ .Values.configuration.general.server_lifetime | int }}
server_tls = {{ .Values.configuration.general.server_tls }}
idle_client_in_transaction_timeout = {{ .Values.configuration.general.idle_client_in_transaction_timeout | int }}
healthcheck_timeout = {{ .Values.configuration.general.healthcheck_timeout }}
healthcheck_delay = {{ .Values.configuration.general.healthcheck_delay }}
@@ -50,6 +51,10 @@ stringData:
query_parser_enabled = {{ default true $pool.query_parser_enabled }}
query_parser_read_write_splitting = {{ default true $pool.query_parser_read_write_splitting }}
primary_reads_enabled = {{ default true $pool.primary_reads_enabled }}
db_activity_based_routing = {{ default false $pool.db_activity_based_routing }}
db_activity_based_ms_init_delay = {{ default 100 $pool.db_activity_based_ms_init_delay }}
db_activity_ttl = {{ default 900 $pool.db_activity_ttl }}
table_mutation_cache_ttl = {{ default 50 $pool.table_mutation_cache_ttl }}
sharding_function = {{ default "pg_bigint_hash" $pool.sharding_function | quote }}
{{- range $index, $user := $pool.users }}
@@ -58,11 +63,21 @@ stringData:
##
[pools.{{ $pool.name | quote }}.users.{{ $index }}]
username = {{ $user.username | quote }}
{{- if $user.password }}
password = {{ $user.password | quote }}
{{- else if and $user.passwordSecret.name $user.passwordSecret.key }}
{{- $secret := (lookup "v1" "Secret" $.Release.Namespace $user.passwordSecret.name) }}
{{- if $secret }}
{{- $password := index $secret.data $user.passwordSecret.key | b64dec }}
password = {{ $password | quote }}
{{- end }}
{{- end }}
pool_size = {{ $user.pool_size }}
statement_timeout = {{ $user.statement_timeout }}
min_pool_size = 3
server_lifetime = 60000
statement_timeout = {{ default 0 $user.statement_timeout }}
min_pool_size = {{ default 3 $user.min_pool_size }}
{{- if $user.server_lifetime }}
server_lifetime = {{ $user.server_lifetime }}
{{- end }}
{{- if and $user.server_username $user.server_password }}
server_username = {{ $user.server_username | quote }}
server_password = {{ $user.server_password | quote }}

View File

@@ -175,6 +175,9 @@ configuration:
# Max connection lifetime before it's closed, even if actively used.
server_lifetime: 86400000 # 24 hours
# Whether to use TLS for server connections or not.
server_tls: false
# How long a client is allowed to be idle while in a transaction (ms).
idle_client_in_transaction_timeout: 0 # milliseconds
@@ -295,6 +298,22 @@ configuration:
# ## @param configuration.poolsPostgres.query_parser_read_write_splitting
# query_parser_read_write_splitting: true
# ## Db activity based routing. If enabled, we'll route queries to the primary if the table was recently mutated.
# ## @param configuration.poolsPostgres.db_activity_based_routing
# db_activity_based_routing: false
# ## DB activity based init delay. How long to wait before starting to route queries to the primary after a table mutation.
# ## @param configuration.poolsPostgres.db_activity_based_ms_init_delay
# db_activity_based_ms_init_delay: 100
# ## DB activity TTL. How long before marking the DB as inactive after no mutations or queries.
# ## @param configuration.poolsPostgres.db_activity_ttl
# db_activity_ttl: 900
# ## Table mutation cache TTL. How long to keep track of table mutations.
# ## @param configuration.poolsPostgres.table_mutation_cache_ttl
# table_mutation_cache_ttl: 50
# ## If the query parser is enabled and this setting is enabled, the primary will be part of the pool of databases used for
# ## load balancing of read queries. Otherwise, the primary will only be used for write
# ## queries. The primary can always be explicitly selected with our custom protocol.
@@ -315,7 +334,9 @@ configuration:
# ## Credentials for users that may connect to this cluster
# ## @param users [array]
# ## @param users[0].username Name of the env var (required)
# ## @param users[0].password Value for the env var (required)
# ## @param users[0].password Value for the env var (required) leave empty to use existing secret see passwordSecret.name and passwordSecret.key
# ## @param users[0].passwordSecret.name Name of the secret containing the password
# ## @param users[0].passwordSecret.key Key in the secret containing the password
# ## @param users[0].pool_size Maximum number of server connections that can be established for this user
# ## @param users[0].statement_timeout Maximum query duration. Dangerous, but protects against DBs that died in a non-obvious way.
# users: []

View File

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

View File

@@ -179,7 +179,7 @@ primary_reads_enabled = true
# `random`: picks a shard at random
# `random_healthy`: picks a shard at random favoring shards with the least number of recent errors
# `shard_<number>`: e.g. shard_0, shard_4, etc. picks a specific shard, everytime
# no_shard_specified_behavior = "shard_0"
# default_shard = "shard_0"
# So what if you wanted to implement a different hashing function,
# or you've already built one and you want this pooler to use it?

View File

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

View File

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

View File

@@ -14,7 +14,9 @@ use tokio::sync::mpsc::Sender;
use crate::admin::{generate_server_parameters_for_admin, handle_admin};
use crate::auth_passthrough::refetch_auth_hash;
use crate::config::{get_config, get_idle_client_in_transaction_timeout, Address, PoolMode};
use crate::config::{
get_config, get_idle_client_in_transaction_timeout, Address, AuthType, PoolMode,
};
use crate::constants::*;
use crate::messages::*;
use crate::plugins::PluginOutput;
@@ -463,8 +465,8 @@ where
.count()
== 1;
// Kick any client that's not admin while we're in admin-only mode.
if !admin && admin_only {
// Kick any client that's not admin while we're in admin-only mode.
debug!(
"Rejecting non-admin connection to {} when in admin only mode",
pool_name
@@ -481,72 +483,76 @@ where
let process_id: i32 = rand::random();
let secret_key: i32 = rand::random();
// Perform MD5 authentication.
// TODO: Add SASL support.
let salt = md5_challenge(&mut write).await?;
let code = match read.read_u8().await {
Ok(p) => p,
Err(_) => {
return Err(Error::ClientSocketError(
"password code".into(),
client_identifier,
))
}
};
// PasswordMessage
if code as char != 'p' {
return Err(Error::ProtocolSyncError(format!(
"Expected p, got {}",
code as char
)));
}
let len = match read.read_i32().await {
Ok(len) => len,
Err(_) => {
return Err(Error::ClientSocketError(
"password message length".into(),
client_identifier,
))
}
};
let mut password_response = vec![0u8; (len - 4) as usize];
match read.read_exact(&mut password_response).await {
Ok(_) => (),
Err(_) => {
return Err(Error::ClientSocketError(
"password message".into(),
client_identifier,
))
}
};
let mut prepared_statements_enabled = false;
// Authenticate admin user.
let (transaction_mode, mut server_parameters) = if admin {
let config = get_config();
// TODO: Add SASL support.
// Perform MD5 authentication.
match config.general.admin_auth_type {
AuthType::Trust => (),
AuthType::MD5 => {
let salt = md5_challenge(&mut write).await?;
// Compare server and client hashes.
let password_hash = md5_hash_password(
&config.general.admin_username,
&config.general.admin_password,
&salt,
);
let code = match read.read_u8().await {
Ok(p) => p,
Err(_) => {
return Err(Error::ClientSocketError(
"password code".into(),
client_identifier,
))
}
};
if password_hash != password_response {
let error = Error::ClientGeneralError("Invalid password".into(), client_identifier);
// PasswordMessage
if code as char != 'p' {
return Err(Error::ProtocolSyncError(format!(
"Expected p, got {}",
code as char
)));
}
warn!("{}", error);
wrong_password(&mut write, username).await?;
let len = match read.read_i32().await {
Ok(len) => len,
Err(_) => {
return Err(Error::ClientSocketError(
"password message length".into(),
client_identifier,
))
}
};
return Err(error);
let mut password_response = vec![0u8; (len - 4) as usize];
match read.read_exact(&mut password_response).await {
Ok(_) => (),
Err(_) => {
return Err(Error::ClientSocketError(
"password message".into(),
client_identifier,
))
}
};
// Compare server and client hashes.
let password_hash = md5_hash_password(
&config.general.admin_username,
&config.general.admin_password,
&salt,
);
if password_hash != password_response {
let error =
Error::ClientGeneralError("Invalid password".into(), client_identifier);
warn!("{}", error);
wrong_password(&mut write, username).await?;
return Err(error);
}
}
}
(false, generate_server_parameters_for_admin())
}
// Authenticate normal user.
@@ -573,92 +579,143 @@ where
// Obtain the hash to compare, we give preference to that written in cleartext in config
// if there is nothing set in cleartext and auth passthrough (auth_query) is configured, we use the hash obtained
// when the pool was created. If there is no hash there, we try to fetch it one more time.
let password_hash = if let Some(password) = &pool.settings.user.password {
Some(md5_hash_password(username, password, &salt))
} else {
if !get_config().is_auth_query_configured() {
wrong_password(&mut write, username).await?;
return Err(Error::ClientAuthImpossible(username.into()));
}
match pool.settings.user.auth_type {
AuthType::Trust => (),
AuthType::MD5 => {
// Perform MD5 authentication.
// TODO: Add SASL support.
let salt = md5_challenge(&mut write).await?;
let mut hash = (*pool.auth_hash.read()).clone();
let code = match read.read_u8().await {
Ok(p) => p,
Err(_) => {
return Err(Error::ClientSocketError(
"password code".into(),
client_identifier,
))
}
};
if hash.is_none() {
warn!(
"Query auth configured \
but no hash password found \
for pool {}. Will try to refetch it.",
pool_name
);
// PasswordMessage
if code as char != 'p' {
return Err(Error::ProtocolSyncError(format!(
"Expected p, got {}",
code as char
)));
}
match refetch_auth_hash(&pool).await {
Ok(fetched_hash) => {
warn!("Password for {}, obtained. Updating.", client_identifier);
let len = match read.read_i32().await {
Ok(len) => len,
Err(_) => {
return Err(Error::ClientSocketError(
"password message length".into(),
client_identifier,
))
}
};
let mut password_response = vec![0u8; (len - 4) as usize];
match read.read_exact(&mut password_response).await {
Ok(_) => (),
Err(_) => {
return Err(Error::ClientSocketError(
"password message".into(),
client_identifier,
))
}
};
let password_hash = if let Some(password) = &pool.settings.user.password {
Some(md5_hash_password(username, password, &salt))
} else {
if !get_config().is_auth_query_configured() {
wrong_password(&mut write, username).await?;
return Err(Error::ClientAuthImpossible(username.into()));
}
let mut hash = (*pool.auth_hash.read()).clone();
if hash.is_none() {
warn!(
"Query auth configured \
but no hash password found \
for pool {}. Will try to refetch it.",
pool_name
);
match refetch_auth_hash(&pool).await {
Ok(fetched_hash) => {
warn!(
"Password for {}, obtained. Updating.",
client_identifier
);
{
let mut pool_auth_hash = pool.auth_hash.write();
*pool_auth_hash = Some(fetched_hash.clone());
}
hash = Some(fetched_hash);
}
Err(err) => {
wrong_password(&mut write, username).await?;
return Err(Error::ClientAuthPassthroughError(
err.to_string(),
client_identifier,
));
}
}
};
Some(md5_hash_second_pass(&hash.unwrap(), &salt))
};
// Once we have the resulting hash, we compare with what the client gave us.
// If they do not match and auth query is set up, we try to refetch the hash one more time
// to see if the password has changed since the pool was created.
//
// @TODO: we could end up fetching again the same password twice (see above).
if password_hash.unwrap() != password_response {
warn!(
"Invalid password {}, will try to refetch it.",
client_identifier
);
let fetched_hash = match refetch_auth_hash(&pool).await {
Ok(fetched_hash) => fetched_hash,
Err(err) => {
wrong_password(&mut write, username).await?;
return Err(err);
}
};
let new_password_hash = md5_hash_second_pass(&fetched_hash, &salt);
// Ok password changed in server an auth is possible.
if new_password_hash == password_response {
warn!(
"Password for {}, changed in server. Updating.",
client_identifier
);
{
let mut pool_auth_hash = pool.auth_hash.write();
*pool_auth_hash = Some(fetched_hash.clone());
*pool_auth_hash = Some(fetched_hash);
}
hash = Some(fetched_hash);
}
Err(err) => {
} else {
wrong_password(&mut write, username).await?;
return Err(Error::ClientAuthPassthroughError(
err.to_string(),
return Err(Error::ClientGeneralError(
"Invalid password".into(),
client_identifier,
));
}
}
};
Some(md5_hash_second_pass(&hash.unwrap(), &salt))
};
// Once we have the resulting hash, we compare with what the client gave us.
// If they do not match and auth query is set up, we try to refetch the hash one more time
// to see if the password has changed since the pool was created.
//
// @TODO: we could end up fetching again the same password twice (see above).
if password_hash.unwrap() != password_response {
warn!(
"Invalid password {}, will try to refetch it.",
client_identifier
);
let fetched_hash = match refetch_auth_hash(&pool).await {
Ok(fetched_hash) => fetched_hash,
Err(err) => {
wrong_password(&mut write, username).await?;
return Err(err);
}
};
let new_password_hash = md5_hash_second_pass(&fetched_hash, &salt);
// Ok password changed in server an auth is possible.
if new_password_hash == password_response {
warn!(
"Password for {}, changed in server. Updating.",
client_identifier
);
{
let mut pool_auth_hash = pool.auth_hash.write();
*pool_auth_hash = Some(fetched_hash);
}
} else {
wrong_password(&mut write, username).await?;
return Err(Error::ClientGeneralError(
"Invalid password".into(),
client_identifier,
));
}
}
let transaction_mode = pool.settings.pool_mode == PoolMode::Transaction;
prepared_statements_enabled =
transaction_mode && pool.prepared_statement_cache.is_some();
@@ -802,6 +859,8 @@ where
// e.g. primary, replica, which shard.
let mut query_router = QueryRouter::new();
let mut checkout_failure_count: u64 = 0;
self.stats.register(self.stats.clone());
// Result returned by one of the plugins.
@@ -824,6 +883,7 @@ where
};
query_router.update_pool_settings(&pool.settings);
query_router.set_default_role();
// Our custom protocol loop.
// We expect the client to either start a transaction with regular queries
@@ -1050,7 +1110,25 @@ where
query_router.role(),
err
);
checkout_failure_count += 1;
if let Some(limit) = pool.settings.checkout_failure_limit {
if checkout_failure_count >= limit {
error!(
"Checkout failure limit reached ({} / {}) - disconnecting client",
checkout_failure_count, limit
);
error_response_terminal(
&mut self.write,
&format!(
"checkout failure limit reached ({} / {})",
checkout_failure_count, limit
),
)
.await?;
self.stats.disconnect();
return Ok(());
}
}
continue;
}
};

View File

@@ -208,6 +208,9 @@ impl Address {
pub struct User {
pub username: String,
pub password: Option<String>,
#[serde(default = "User::default_auth_type")]
pub auth_type: AuthType,
pub server_username: Option<String>,
pub server_password: Option<String>,
pub pool_size: u32,
@@ -225,6 +228,7 @@ impl Default for User {
User {
username: String::from("postgres"),
password: None,
auth_type: AuthType::MD5,
server_username: None,
server_password: None,
pool_size: 15,
@@ -239,6 +243,10 @@ impl Default for User {
}
impl User {
pub fn default_auth_type() -> AuthType {
AuthType::MD5
}
fn validate(&self) -> Result<(), Error> {
if let Some(min_pool_size) = self.min_pool_size {
if min_pool_size > self.pool_size {
@@ -334,6 +342,9 @@ pub struct General {
pub admin_username: String,
pub admin_password: String,
#[serde(default = "General::default_admin_auth_type")]
pub admin_auth_type: AuthType,
#[serde(default = "General::default_validate_config")]
pub validate_config: bool,
@@ -348,6 +359,10 @@ impl General {
"0.0.0.0".into()
}
pub fn default_admin_auth_type() -> AuthType {
AuthType::MD5
}
pub fn default_port() -> u16 {
5432
}
@@ -456,6 +471,7 @@ impl Default for General {
verify_server_certificate: false,
admin_username: String::from("admin"),
admin_password: String::from("admin"),
admin_auth_type: AuthType::MD5,
validate_config: true,
auth_query: None,
auth_query_user: None,
@@ -476,6 +492,15 @@ pub enum PoolMode {
Session,
}
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Eq, Copy, Hash)]
pub enum AuthType {
#[serde(alias = "trust", alias = "Trust")]
Trust,
#[serde(alias = "md5", alias = "MD5")]
MD5,
}
impl std::fmt::Display for PoolMode {
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
match self {
@@ -533,6 +558,14 @@ pub struct Pool {
/// Close idle connections that have been opened for longer than this.
pub idle_timeout: Option<u64>,
/// Maximum number of checkout failures a client is allowed before it
/// gets disconnected. This is needed to prevent persistent client/server
/// imbalance in high availability setups where multiple PgCat instances are placed
/// behind a single load balancer. If for any reason a client lands on a PgCat instance that has
/// a large number of connected clients, it might get stuck in perpetual checkout failure loop especially
/// in session mode
pub checkout_failure_limit: Option<u64>,
/// Close server connections that have been opened for longer than this.
/// Only applied to idle connections. If the connection is actively used for
/// longer than this period, the pool will not interrupt it.
@@ -564,6 +597,19 @@ pub struct Pool {
#[serde(default = "Pool::default_prepared_statements_cache_size")]
pub prepared_statements_cache_size: usize,
// Support for query routing based on database activity
#[serde(default = "Pool::default_db_activity_based_routing")]
pub db_activity_based_routing: bool,
#[serde(default = "Pool::default_db_activity_init_delay")]
pub db_activity_init_delay: u64,
#[serde(default = "Pool::default_db_activity_ttl")]
pub db_activity_ttl: u64,
#[serde(default = "Pool::default_table_mutation_cache_ms_ttl")]
pub table_mutation_cache_ms_ttl: u64,
pub plugins: Option<Plugins>,
pub shards: BTreeMap<String, Shard>,
pub users: BTreeMap<String, User>,
@@ -617,6 +663,25 @@ impl Pool {
0
}
pub fn default_db_activity_based_routing() -> bool {
false
}
pub fn default_db_activity_init_delay() -> u64 {
// 100 milliseconds
100
}
pub fn default_db_activity_ttl() -> u64 {
// 15 minutes
15 * 60
}
pub fn default_table_mutation_cache_ms_ttl() -> u64 {
// 50 milliseconds
50
}
pub fn validate(&mut self) -> Result<(), Error> {
match self.default_role.as_ref() {
"any" => (),
@@ -699,6 +764,23 @@ impl Pool {
user.validate()?;
}
if self.db_activity_based_routing {
if self.db_activity_init_delay == 0 {
error!("db_activity_init_delay must be greater than 0");
return Err(Error::BadConfig);
}
if self.table_mutation_cache_ms_ttl == 0 {
error!("table_mutation_cache_ms_ttl must be greater than 0");
return Err(Error::BadConfig);
}
if self.db_activity_ttl == 0 {
error!("db_activity_ttl must be greater than 0");
return Err(Error::BadConfig);
}
}
Ok(())
}
}
@@ -708,6 +790,7 @@ impl Default for Pool {
Pool {
pool_mode: Self::default_pool_mode(),
load_balancing_mode: Self::default_load_balancing_mode(),
checkout_failure_limit: None,
default_role: String::from("any"),
query_parser_enabled: false,
query_parser_max_length: None,
@@ -728,6 +811,10 @@ impl Default for Pool {
cleanup_server_connections: true,
log_client_parameter_status_changes: false,
prepared_statements_cache_size: Self::default_prepared_statements_cache_size(),
db_activity_based_routing: Self::default_db_activity_based_routing(),
db_activity_init_delay: Self::default_db_activity_init_delay(),
db_activity_ttl: Self::default_db_activity_ttl(),
table_mutation_cache_ms_ttl: Self::default_table_mutation_cache_ms_ttl(),
plugins: None,
shards: BTreeMap::from([(String::from("1"), Shard::default())]),
users: BTreeMap::default(),
@@ -1220,6 +1307,17 @@ impl Config {
None => self.general.idle_timeout,
};
info!("[pool: {}] Idle timeout: {}ms", pool_name, idle_timeout);
match pool_config.checkout_failure_limit {
Some(checkout_failure_limit) => {
info!(
"[pool: {}] Checkout failure limit: {}",
pool_name, checkout_failure_limit
);
}
None => {
info!("[pool: {}] Checkout failure limit: not set", pool_name);
}
};
info!(
"[pool: {}] Sharding function: {}",
pool_name,
@@ -1264,6 +1362,22 @@ impl Config {
"[pool: {}] Cleanup server connections: {}",
pool_name, pool_config.cleanup_server_connections
);
info!(
"[pool: {}] DB activity based routing: {}",
pool_name, pool_config.db_activity_based_routing
);
info!(
"[pool: {}] DB activity init delay: {}",
pool_name, pool_config.db_activity_init_delay
);
info!(
"[pool: {}] DB activity TTL: {}",
pool_name, pool_config.db_activity_ttl
);
info!(
"[pool: {}] Table mutation cache TTL: {}",
pool_name, pool_config.table_mutation_cache_ms_ttl
);
info!(
"[pool: {}] Log client parameter status changes: {}",
pool_name, pool_config.log_client_parameter_status_changes

View File

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

View File

@@ -152,6 +152,14 @@ pub struct PoolSettings {
/// Random or LeastOutstandingConnections.
pub load_balancing_mode: LoadBalancingMode,
/// Maximum number of checkout failures a client is allowed before it
/// gets disconnected. This is needed to prevent persistent client/server
/// imbalance in high availability setups where multiple PgCat instances are placed
/// behind a single load balancer. If for any reason a client lands on a PgCat instance that has
/// a large number of connected clients, it might get stuck in perpetual checkout failure loop especially
/// in session mode
pub checkout_failure_limit: Option<u64>,
// Number of shards.
pub shards: usize,
@@ -174,6 +182,18 @@ pub struct PoolSettings {
// Read from the primary as well or not.
pub primary_reads_enabled: bool,
// Automatic primary/replica selection based on recent activity.
pub db_activity_based_routing: bool,
// DB activity init delay
pub db_activity_init_delay: u64,
// DB activity TTL
pub db_activity_ttl: u64,
// Table mutation cache TTL
pub table_mutation_cache_ms_ttl: u64,
// Sharding function.
pub sharding_function: ShardingFunction,
@@ -215,6 +235,7 @@ impl Default for PoolSettings {
PoolSettings {
pool_mode: PoolMode::Transaction,
load_balancing_mode: LoadBalancingMode::Random,
checkout_failure_limit: None,
shards: 1,
user: User::default(),
db: String::default(),
@@ -223,6 +244,10 @@ impl Default for PoolSettings {
query_parser_max_length: None,
query_parser_read_write_splitting: false,
primary_reads_enabled: true,
db_activity_based_routing: false,
db_activity_init_delay: 100,
db_activity_ttl: 15 * 60,
table_mutation_cache_ms_ttl: 50,
sharding_function: ShardingFunction::PgBigintHash,
automatic_sharding_key: None,
healthcheck_delay: General::default_healthcheck_delay(),
@@ -521,6 +546,7 @@ impl ConnectionPool {
None => pool_config.pool_mode,
},
load_balancing_mode: pool_config.load_balancing_mode,
checkout_failure_limit: pool_config.checkout_failure_limit,
// shards: pool_config.shards.clone(),
shards: shard_ids.len(),
user: user.clone(),
@@ -537,6 +563,10 @@ impl ConnectionPool {
.query_parser_read_write_splitting,
primary_reads_enabled: pool_config.primary_reads_enabled,
sharding_function: pool_config.sharding_function,
db_activity_based_routing: pool_config.db_activity_based_routing,
db_activity_init_delay: pool_config.db_activity_init_delay,
db_activity_ttl: pool_config.db_activity_ttl,
table_mutation_cache_ms_ttl: pool_config.table_mutation_cache_ms_ttl,
automatic_sharding_key: pool_config.automatic_sharding_key.clone(),
healthcheck_delay: config.general.healthcheck_delay,
healthcheck_timeout: config.general.healthcheck_timeout,

View File

@@ -200,18 +200,17 @@ struct PrometheusMetric<Value: fmt::Display> {
impl<Value: fmt::Display> fmt::Display for PrometheusMetric<Value> {
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
let formatted_labels = self
.labels
let mut sorted_labels: Vec<_> = self.labels.iter().collect();
sorted_labels.sort_by_key(|&(key, _)| key);
let formatted_labels = sorted_labels
.iter()
.map(|(key, value)| format!("{}=\"{}\"", key, value))
.collect::<Vec<_>>()
.join(",");
write!(
f,
"# HELP {name} {help}\n# TYPE {name} {ty}\n{name}{{{formatted_labels}}} {value}\n",
"{name}{{{formatted_labels}}} {value}",
name = format_args!("pgcat_{}", self.name),
help = self.help,
ty = self.ty,
formatted_labels = formatted_labels,
value = self.value
)
@@ -247,7 +246,7 @@ impl<Value: fmt::Display> PrometheusMetric<Value> {
labels.insert("pool", address.pool_name.clone());
labels.insert("index", address.address_index.to_string());
labels.insert("database", address.database.to_string());
labels.insert("user", address.username.clone());
labels.insert("username", address.username.clone());
Self::from_name(&format!("databases_{}", name), value, labels)
}
@@ -264,7 +263,8 @@ impl<Value: fmt::Display> PrometheusMetric<Value> {
labels.insert("pool", address.pool_name.clone());
labels.insert("index", address.address_index.to_string());
labels.insert("database", address.database.to_string());
labels.insert("user", address.username.clone());
labels.insert("username", address.username.clone());
Self::from_name(&format!("servers_{}", name), value, labels)
}
@@ -276,7 +276,7 @@ impl<Value: fmt::Display> PrometheusMetric<Value> {
labels.insert("role", address.role.to_string());
labels.insert("index", address.address_index.to_string());
labels.insert("database", address.database.to_string());
labels.insert("user", address.username.clone());
labels.insert("username", address.username.clone());
Self::from_name(&format!("stats_{}", name), value, labels)
}
@@ -288,6 +288,15 @@ impl<Value: fmt::Display> PrometheusMetric<Value> {
Self::from_name(&format!("pools_{}", name), value, labels)
}
fn get_header(&self) -> String {
format!(
"\n# HELP {name} {help}\n# TYPE {name} {ty}",
name = format_args!("pgcat_{}", self.name),
help = self.help,
ty = self.ty,
)
}
}
async fn prometheus_stats(
@@ -300,6 +309,7 @@ async fn prometheus_stats(
push_pool_stats(&mut lines);
push_server_stats(&mut lines);
push_database_stats(&mut lines);
lines.push("".to_string()); // Ensure to end the stats with a line terminator as required by the specification.
Response::builder()
.header("content-type", "text/plain; version=0.0.4")
@@ -313,6 +323,7 @@ async fn prometheus_stats(
// Adds metrics shown in a SHOW STATS admin command.
fn push_address_stats(lines: &mut Vec<String>) {
let mut grouped_metrics: HashMap<String, Vec<PrometheusMetric<u64>>> = HashMap::new();
for (_, pool) in get_all_pools() {
for shard in 0..pool.shards() {
for server in 0..pool.servers(shard) {
@@ -322,7 +333,10 @@ fn push_address_stats(lines: &mut Vec<String>) {
if let Some(prometheus_metric) =
PrometheusMetric::<u64>::from_address(address, &key, value)
{
lines.push(prometheus_metric.to_string());
grouped_metrics
.entry(key)
.or_default()
.push(prometheus_metric);
} else {
debug!("Metric {} not implemented for {}", key, address.name());
}
@@ -330,33 +344,53 @@ fn push_address_stats(lines: &mut Vec<String>) {
}
}
}
for (_key, metrics) in grouped_metrics {
if !metrics.is_empty() {
lines.push(metrics[0].get_header());
for metric in metrics {
lines.push(metric.to_string());
}
}
}
}
// Adds relevant metrics shown in a SHOW POOLS admin command.
fn push_pool_stats(lines: &mut Vec<String>) {
let mut grouped_metrics: HashMap<String, Vec<PrometheusMetric<u64>>> = HashMap::new();
let pool_stats = PoolStats::construct_pool_lookup();
for (pool_id, stats) in pool_stats.iter() {
for (name, value) in stats.clone() {
if let Some(prometheus_metric) =
PrometheusMetric::<u64>::from_pool(pool_id.clone(), &name, value)
{
lines.push(prometheus_metric.to_string());
grouped_metrics
.entry(name)
.or_default()
.push(prometheus_metric);
} else {
debug!("Metric {} not implemented for ({})", name, *pool_id);
}
}
}
for (_key, metrics) in grouped_metrics {
if !metrics.is_empty() {
lines.push(metrics[0].get_header());
for metric in metrics {
lines.push(metric.to_string());
}
}
}
}
// Adds relevant metrics shown in a SHOW DATABASES admin command.
fn push_database_stats(lines: &mut Vec<String>) {
let mut grouped_metrics: HashMap<String, Vec<PrometheusMetric<u32>>> = HashMap::new();
for (_, pool) in get_all_pools() {
let pool_config = pool.settings.clone();
for shard in 0..pool.shards() {
for server in 0..pool.servers(shard) {
let address = pool.address(shard, server);
let pool_state = pool.pool_state(shard, server);
let metrics = vec![
("pool_size", pool_config.user.pool_size),
("current_connections", pool_state.connections),
@@ -365,7 +399,10 @@ fn push_database_stats(lines: &mut Vec<String>) {
if let Some(prometheus_metric) =
PrometheusMetric::<u32>::from_database_info(address, key, value)
{
lines.push(prometheus_metric.to_string());
grouped_metrics
.entry(key.to_string())
.or_default()
.push(prometheus_metric);
} else {
debug!("Metric {} not implemented for {}", key, address.name());
}
@@ -373,6 +410,14 @@ fn push_database_stats(lines: &mut Vec<String>) {
}
}
}
for (_key, metrics) in grouped_metrics {
if !metrics.is_empty() {
lines.push(metrics[0].get_header());
for metric in metrics {
lines.push(metric.to_string());
}
}
}
}
// Adds relevant metrics shown in a SHOW SERVERS admin command.
@@ -405,7 +450,7 @@ fn push_server_stats(lines: &mut Vec<String>) {
crate::stats::ServerState::Idle => entry.idle_count += 1,
}
}
let mut grouped_metrics: HashMap<String, Vec<PrometheusMetric<u64>>> = HashMap::new();
for (_, pool) in get_all_pools() {
for shard in 0..pool.shards() {
for server in 0..pool.servers(shard) {
@@ -428,7 +473,10 @@ fn push_server_stats(lines: &mut Vec<String>) {
if let Some(prometheus_metric) =
PrometheusMetric::<u64>::from_server_info(address, key, value)
{
lines.push(prometheus_metric.to_string());
grouped_metrics
.entry(key.to_string())
.or_default()
.push(prometheus_metric);
} else {
debug!("Metric {} not implemented for {}", key, address.name());
}
@@ -437,6 +485,14 @@ fn push_server_stats(lines: &mut Vec<String>) {
}
}
}
for (_key, metrics) in grouped_metrics {
if !metrics.is_empty() {
lines.push(metrics[0].get_header());
for metric in metrics {
lines.push(metric.to_string());
}
}
}
}
pub async fn start_metric_server(http_addr: SocketAddr) {

View File

@@ -2,6 +2,7 @@
/// or implied query characteristics.
use bytes::{Buf, BytesMut};
use log::{debug, error};
use mini_moka::sync::Cache;
use once_cell::sync::OnceCell;
use regex::{Regex, RegexSet};
use sqlparser::ast::Statement::{Delete, Insert, Query, StartTransaction, Update};
@@ -11,6 +12,7 @@ use sqlparser::ast::{
};
use sqlparser::dialect::PostgreSqlDialect;
use sqlparser::parser::Parser;
use std::sync::OnceLock;
use crate::config::Role;
use crate::errors::Error;
@@ -21,6 +23,7 @@ use crate::sharding::Sharder;
use std::collections::BTreeSet;
use std::io::Cursor;
use std::time::Duration;
use std::{cmp, mem};
/// Regexes used to parse custom commands.
@@ -66,6 +69,18 @@ static CUSTOM_SQL_REGEX_SET: OnceCell<RegexSet> = OnceCell::new();
// Get the value inside the custom command.
static CUSTOM_SQL_REGEX_LIST: OnceCell<Vec<Regex>> = OnceCell::new();
#[derive(Debug, Clone, PartialEq)]
enum DatabaseActivityState {
Active,
Initializing,
}
// A moka cache for the databases
// the key is the database name and the value is the database activity state
static DATABASE_ACTIVITY_CACHE: OnceLock<Cache<String, DatabaseActivityState>> = OnceLock::new();
// A moka cache for the tables, the key is the db_table.
static TABLE_MUTATIONS_CACHE: OnceLock<Cache<String, bool>> = OnceLock::new();
/// The query router.
pub struct QueryRouter {
/// Which shard we should be talking to right now.
@@ -87,6 +102,12 @@ pub struct QueryRouter {
placeholders: Vec<i16>,
}
struct ExtractedExprsAndTables<'a> {
exprs: Vec<Expr>,
table_names: Vec<Vec<Ident>>,
assignments_opt: Option<&'a Vec<Assignment>>,
}
impl QueryRouter {
/// One-time initialization of regexes
/// that parse our custom SQL protocol.
@@ -386,6 +407,53 @@ impl QueryRouter {
}
}
/// Determines if a query is a mutation or not.
fn is_mutation_query(q: &sqlparser::ast::Query) -> bool {
use sqlparser::ast::*;
match q.body.as_ref() {
SetExpr::Insert(_) => true,
SetExpr::Update(_) => true,
SetExpr::Query(q) => Self::is_mutation_query(q),
_ => false,
}
}
fn database_activity_cache(&self) -> Cache<String, DatabaseActivityState> {
DATABASE_ACTIVITY_CACHE
.get_or_init(|| {
Cache::builder()
.time_to_idle(Duration::from_secs(self.pool_settings.db_activity_ttl))
.build()
})
.clone()
}
/// Check database activity state and reset it if necessary
fn database_activity_state(&self, db: &String) -> DatabaseActivityState {
let cache = self.database_activity_cache();
// Exists in cache
if cache.contains_key(db) {
return cache.get(db).unwrap();
}
// Not in cache
debug!("Adding database to cache: {}", db);
cache.insert(db.to_string(), DatabaseActivityState::Initializing);
// Set a timer to update the cache
let db = db.clone();
let db_activity_init_delay = self.pool_settings.db_activity_init_delay;
tokio::spawn(async move {
tokio::time::sleep(Duration::from_millis(db_activity_init_delay)).await;
cache.insert(db, DatabaseActivityState::Active);
});
DatabaseActivityState::Initializing
}
/// Try to infer which server to connect to based on the contents of the query.
pub fn infer(&mut self, ast: &Vec<sqlparser::ast::Statement>) -> Result<(), Error> {
if !self.pool_settings.query_parser_read_write_splitting {
@@ -400,9 +468,23 @@ impl QueryRouter {
return Err(Error::QueryRouterParserError("empty query".into()));
}
let mut primary_set_based_on_activity = false;
let mut visited_write_statement = false;
let mut prev_inferred_shard = None;
if self.pool_settings.db_activity_based_routing {
let db = self.pool_settings.db.clone();
let state = self.database_activity_state(&db);
debug!("Database activity state: {:?}", state);
if let DatabaseActivityState::Initializing = state {
debug!("Database is initializing, going to primary");
self.active_role = Some(Role::Primary);
primary_set_based_on_activity = true;
}
}
for q in ast {
match q {
// All transactions go to the primary, probably a write.
@@ -413,6 +495,22 @@ impl QueryRouter {
// Likely a read-only query
Query(query) => {
if primary_set_based_on_activity {
// If we already set the role based on activity, we don't need to do it again
continue;
}
if self.pool_settings.db_activity_based_routing {
// Check if the tables in the query have been written to recently
if self.query_handles_tables_in_mutation_cache(query) {
debug!("Query handles tables in mutation cache, going to primary");
self.active_role = Some(Role::Primary);
primary_set_based_on_activity = true;
continue;
}
}
match &self.pool_settings.automatic_sharding_key {
Some(_) => {
// TODO: if we have multiple queries in the same message,
@@ -428,8 +526,9 @@ impl QueryRouter {
};
let has_locks = !query.locks.is_empty();
let has_mutation = Self::is_mutation_query(query);
if has_locks {
if has_locks || has_mutation {
self.active_role = Some(Role::Primary);
} else if !visited_write_statement {
// If we already visited a write statement, we should be going to the primary.
@@ -442,6 +541,13 @@ impl QueryRouter {
// Likely a write
_ => {
debug!("Write statement found, going to primary");
if self.pool_settings.db_activity_based_routing {
// add all of the query tables to the mutation cache
self.update_mutation_cache_on_write(q);
}
match &self.pool_settings.automatic_sharding_key {
Some(_) => {
// TODO: similar to the above, if we have multiple queries in the
@@ -484,62 +590,69 @@ impl QueryRouter {
Ok(())
}
fn infer_shard_on_write(&mut self, q: &Statement) -> Result<Option<usize>, Error> {
let mut exprs = Vec::new();
fn table_mutations_cache(&self) -> Cache<String, bool> {
TABLE_MUTATIONS_CACHE
.get_or_init(|| {
Cache::builder()
.time_to_live(Duration::from_millis(
self.pool_settings.table_mutation_cache_ms_ttl,
))
.build()
})
.clone()
}
// Collect all table names from the query.
let mut table_names = Vec::new();
fn query_handles_tables_in_mutation_cache(&self, query: &sqlparser::ast::Query) -> bool {
let table_mutations_cache = self.table_mutations_cache();
debug!("Checking if query handles tables in mutation cache");
debug!("Table mutations cache: {:?}", table_mutations_cache);
match q {
Insert {
or,
into: _,
table_name,
columns,
overwrite: _,
source,
partitioned,
after_columns,
table: _,
on: _,
returning: _,
ignore: _,
} => {
// Not supported in postgres.
assert!(or.is_none());
assert!(partitioned.is_none());
assert!(after_columns.is_empty());
Self::process_table(table_name, &mut table_names);
if let Some(source) = source {
Self::process_query(source, &mut exprs, &mut table_names, &Some(columns));
for tables in self.table_names(query) {
for table in tables {
if table_mutations_cache.contains_key(&self.table_mutation_cache_key(table)) {
return true;
}
}
Delete {
tables,
from,
using,
selection,
returning: _,
order_by: _,
limit: _,
} => {
if let Some(expr) = selection {
}
false
}
fn extract_exprs_and_table_names<'a>(
&'a self,
q: &'a Statement,
) -> Option<ExtractedExprsAndTables<'a>> {
let mut exprs = Vec::new();
let mut table_names = Vec::new();
let mut assignments_opt = None;
match q {
Insert(i) => {
// Not supported in postgres.
assert!(i.or.is_none());
assert!(i.partitioned.is_none());
assert!(i.after_columns.is_empty());
Self::process_table(&i.table_name, &mut table_names);
if let Some(source) = &i.source {
Self::process_query(source, &mut exprs, &mut table_names, &Some(&i.columns));
}
}
Delete(d) => {
if let Some(expr) = &d.selection {
exprs.push(expr.clone());
}
// Multi tables delete are not supported in postgres.
assert!(tables.is_empty());
// Multi-tables delete are not supported in postgres.
assert!(d.tables.is_empty());
Self::process_tables_with_join(from, &mut exprs, &mut table_names);
if let Some(using_tbl_with_join) = using {
if let Some(using_tbl_with_join) = &d.using {
Self::process_tables_with_join(
using_tbl_with_join,
&mut exprs,
&mut table_names,
);
}
Self::process_selection(selection, &mut exprs);
Self::process_selection(&d.selection, &mut exprs);
}
Update {
table,
@@ -553,14 +666,55 @@ impl QueryRouter {
Self::process_table_with_join(from_tbl, &mut exprs, &mut table_names);
}
Self::process_selection(selection, &mut exprs);
self.assignment_parser(assignments)?;
}
_ => {
return Ok(None);
assignments_opt = Some(assignments);
}
_ => return None,
};
Ok(self.infer_shard_from_exprs(exprs, table_names))
Some(ExtractedExprsAndTables {
exprs,
table_names,
assignments_opt,
})
}
fn infer_shard_on_write(&mut self, q: &Statement) -> Result<Option<usize>, Error> {
if let Some(extracted) = self.extract_exprs_and_table_names(q) {
let exprs = extracted.exprs;
let table_names = extracted.table_names;
let assignments_opt = extracted.assignments_opt;
if let Some(assignments) = assignments_opt {
self.assignment_parser(assignments)?;
}
Ok(self.infer_shard_from_exprs(exprs, table_names))
} else {
Ok(None)
}
}
fn update_mutation_cache_on_write(&self, q: &Statement) {
if let Some(extracted) = self.extract_exprs_and_table_names(q) {
debug!("Updating mutation cache on write");
let table_names = extracted.table_names;
debug!("Table names in mutation query: {:?}", table_names);
let table_mutations_cache = self.table_mutations_cache();
for tables in table_names {
for table in tables {
table_mutations_cache.insert(self.table_mutation_cache_key(table), true);
}
}
}
}
// combines the database name and table name into a single string
// to be used as the key in the table mutation cache
// e.g. "mydb.mytable"
fn table_mutation_cache_key(&self, table: Ident) -> String {
format!("{}.{}", self.pool_settings.db, table.value)
}
fn process_query(
@@ -809,7 +963,13 @@ impl QueryRouter {
for a in assignments {
if sharding_key[0].value == "*"
&& sharding_key[1].value == a.id.last().unwrap().value.to_lowercase()
&& sharding_key[1].value
== a.target
.to_string()
.split('.')
.last()
.unwrap()
.to_lowercase()
{
return Err(Error::QueryRouterParserError(
"Sharding key cannot be updated.".into(),
@@ -942,6 +1102,18 @@ impl QueryRouter {
self.infer_shard_from_exprs(exprs, table_names)
}
/// get table names from query
fn table_names(&self, query: &sqlparser::ast::Query) -> Vec<Vec<Ident>> {
let mut exprs = Vec::new();
let mut table_names = Vec::new();
Self::process_query(query, &mut exprs, &mut table_names, &None);
debug!("Table names in query: {:?}", table_names);
table_names
}
fn infer_shard_from_exprs(
&mut self,
exprs: Vec<Expr>,
@@ -1048,6 +1220,11 @@ impl QueryRouter {
self.active_shard
}
/// Set active_role as the default_role specified in the pool.
pub fn set_default_role(&mut self) {
self.active_role = self.pool_settings.default_role;
}
/// Get the current desired server role we should be talking to.
pub fn role(&self) -> Option<Role> {
self.active_role
@@ -1104,6 +1281,7 @@ mod test {
use crate::messages::simple_query;
use crate::sharding::ShardingFunction;
use bytes::BufMut;
use serial_test::serial;
#[test]
fn test_defaults() {
@@ -1113,6 +1291,26 @@ mod test {
assert_eq!(qr.role(), None);
}
#[test]
fn test_split_cte_queries() {
QueryRouter::setup();
let mut qr = QueryRouter::new();
qr.pool_settings.query_parser_read_write_splitting = true;
qr.pool_settings.query_parser_enabled = true;
let query = simple_query(
"WITH t AS (
SELECT id FROM users WHERE name ILIKE '%ja%'
)
UPDATE user_languages
SET settings = '{}'
FROM t WHERE t.id = user_id;",
);
let ast = qr.parse(&query).unwrap();
assert!(qr.infer(&ast).is_ok());
assert_eq!(qr.role(), Some(Role::Primary));
}
#[test]
fn test_infer_replica() {
QueryRouter::setup();
@@ -1399,6 +1597,19 @@ mod test {
assert!(!qr.query_parser_enabled());
}
#[test]
fn test_query_parser() {
QueryRouter::setup();
let mut qr = QueryRouter::new();
qr.pool_settings.query_parser_read_write_splitting = true;
let query = simple_query("SELECT req_tab_0.* FROM validation req_tab_0 WHERE array['http://www.w3.org/ns/shacl#ValidationResult'] && req_tab_0.type::text[] AND ( ( (req_tab_0.focusnode = 'DataSource_Credilogic_DataSourceAddress_144959227') ) )");
assert!(qr.infer(&qr.parse(&query).unwrap()).is_ok());
let query = simple_query("WITH EmployeeSalaries AS (SELECT Department, Salary FROM Employees) SELECT Department, AVG(Salary) AS AverageSalary FROM EmployeeSalaries GROUP BY Department;");
assert!(qr.infer(&qr.parse(&query).unwrap()).is_ok());
}
#[test]
fn test_update_from_pool_settings() {
QueryRouter::setup();
@@ -1406,6 +1617,7 @@ mod test {
let pool_settings = PoolSettings {
pool_mode: PoolMode::Transaction,
load_balancing_mode: crate::config::LoadBalancingMode::Random,
checkout_failure_limit: None,
shards: 2,
user: crate::config::User::default(),
default_role: Some(Role::Replica),
@@ -1426,6 +1638,10 @@ mod test {
auth_query_password: None,
auth_query_user: None,
db: "test".to_string(),
db_activity_based_routing: PoolSettings::default().db_activity_based_routing,
db_activity_init_delay: PoolSettings::default().db_activity_init_delay,
db_activity_ttl: PoolSettings::default().db_activity_ttl,
table_mutation_cache_ms_ttl: PoolSettings::default().table_mutation_cache_ms_ttl,
plugins: None,
};
let mut qr = QueryRouter::new();
@@ -1484,6 +1700,7 @@ mod test {
let pool_settings = PoolSettings {
pool_mode: PoolMode::Transaction,
load_balancing_mode: crate::config::LoadBalancingMode::Random,
checkout_failure_limit: Some(10),
shards: 5,
user: crate::config::User::default(),
default_role: Some(Role::Replica),
@@ -1504,6 +1721,10 @@ mod test {
auth_query_password: None,
auth_query_user: None,
db: "test".to_string(),
db_activity_based_routing: PoolSettings::default().db_activity_based_routing,
db_activity_init_delay: PoolSettings::default().db_activity_init_delay,
db_activity_ttl: PoolSettings::default().db_activity_ttl,
table_mutation_cache_ms_ttl: PoolSettings::default().table_mutation_cache_ms_ttl,
plugins: None,
};
@@ -1919,4 +2140,150 @@ mod test {
assert_eq!(res, Ok(PluginOutput::Allow));
}
#[tokio::test]
#[serial]
async fn test_db_activity_based_routing_initializing_state() {
QueryRouter::setup();
let mut qr = QueryRouter::new();
qr.pool_settings.db_activity_based_routing = true;
qr.pool_settings.query_parser_read_write_splitting = true;
qr.pool_settings.query_parser_enabled = true;
qr.pool_settings.db = "test_table_mutation_cache".to_string();
qr.database_activity_cache()
.invalidate(&qr.pool_settings.db.clone());
let query = simple_query("SELECT * FROM some_table");
let ast = qr.parse(&query).unwrap();
// Initially, the database activity should be in the "Initializing" state
let state = qr.database_activity_state(&qr.pool_settings.db.clone());
assert_eq!(state, DatabaseActivityState::Initializing);
// Check that the router chooses the primary role due to "Initializing" state
assert!(qr.infer(&ast).is_ok());
assert_eq!(qr.role(), Some(Role::Primary));
}
#[tokio::test]
#[serial]
async fn test_db_activity_based_routing_active_state() {
QueryRouter::setup();
let mut qr = QueryRouter::new();
qr.pool_settings.db_activity_based_routing = true;
qr.pool_settings.query_parser_read_write_splitting = true;
qr.pool_settings.query_parser_enabled = true;
qr.pool_settings.db = "test_table_mutation_cache".to_string();
let db_name = qr.pool_settings.db.clone();
let cache = qr.database_activity_cache();
cache.insert(db_name.clone(), DatabaseActivityState::Active);
let query = simple_query("SELECT * FROM some_table");
let ast = qr.parse(&query).unwrap();
// Check that the router can choose a replica role when in "Active" state
assert!(qr.infer(&ast).is_ok());
assert_eq!(qr.role(), None); // Default should allow replica due to active state
}
#[tokio::test]
#[serial]
async fn test_table_mutation_cache_on_write() {
QueryRouter::setup();
let mut qr = QueryRouter::new();
qr.pool_settings.db_activity_based_routing = true;
qr.pool_settings.table_mutation_cache_ms_ttl = 20_000; // 20 seconds in milliseconds
qr.pool_settings.query_parser_enabled = true;
qr.pool_settings.query_parser_read_write_splitting = true;
qr.pool_settings.db = "test_table_mutation_cache".to_string();
qr.database_activity_cache()
.invalidate(&qr.pool_settings.db.clone());
let query = simple_query("UPDATE some_table SET col1 = 'value' WHERE col2 = 1");
let ast = qr.parse(&query).unwrap();
// Simulate the mutation query which should populate the mutation cache
assert!(qr.infer(&ast).is_ok());
assert_eq!(qr.role(), Some(Role::Primary));
let table_cache_key = qr.table_mutation_cache_key(Ident::new("some_table"));
let cache = qr.table_mutations_cache();
// Ensure the table mutation cache contains the table with recent write
assert!(cache.contains_key(&table_cache_key));
}
#[tokio::test]
#[serial]
async fn test_db_activity_based_routing_multi_query() {
use super::*;
use crate::messages::simple_query;
use tokio::time::Duration;
QueryRouter::setup();
let mut qr = QueryRouter::new();
// Configure the pool settings for db_activity_based_routing
qr.pool_settings.query_parser_read_write_splitting = true;
qr.pool_settings.query_parser_enabled = true;
qr.pool_settings.db_activity_based_routing = true;
qr.pool_settings.db = "test_db_activity_routing".to_string();
qr.database_activity_cache()
.invalidate(&qr.pool_settings.db.clone());
// First query when database is initializing
let query = simple_query("SELECT * FROM test_table");
let ast = qr.parse(&query).unwrap();
assert!(qr.infer(&ast).is_ok());
// Should route to primary because database is initializing
assert_eq!(qr.role(), Some(Role::Primary));
// Wait for the initialization delay to pass
tokio::time::sleep(Duration::from_millis(
qr.pool_settings.db_activity_init_delay * 2,
))
.await;
// Next query after database is active
let query = simple_query("SELECT * FROM test_table");
let ast = qr.parse(&query).unwrap();
qr.active_role = None; // Reset the active_role
assert!(qr.infer(&ast).is_ok());
// Should route to replica because database is active and no recent mutations
assert_eq!(qr.role(), None);
// Simulate a write query to update the mutation cache
let query = simple_query("INSERT INTO test_table (id, name) VALUES (1, 'test')");
let ast = qr.parse(&query).unwrap();
qr.active_role = None; // Reset the active_role
assert!(qr.infer(&ast).is_ok());
// Should route to primary because it's a write operation
assert_eq!(qr.role(), Some(Role::Primary));
// Immediately run a read query on the same table
let query = simple_query("SELECT * FROM test_table WHERE id = 1");
let ast = qr.parse(&query).unwrap();
qr.active_role = None; // Reset the active_role
assert!(qr.infer(&ast).is_ok());
// Should route to primary because the table was recently mutated
assert_eq!(qr.role(), Some(Role::Primary));
// Wait for the mutation cache TTL to expire
tokio::time::sleep(Duration::from_millis(
qr.pool_settings.table_mutation_cache_ms_ttl * 2,
))
.await;
// Run the read query again after cache expiration
let query = simple_query("SELECT * FROM test_table WHERE id = 1");
let ast = qr.parse(&query).unwrap();
qr.active_role = None; // Reset the active_role
assert!(qr.infer(&ast).is_ok());
// Should route to replica because mutation cache has expired
assert_eq!(qr.role(), None);
}
}

34
start_test_env.sh Executable file
View File

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

View File

@@ -1,4 +1,4 @@
FROM rust:bullseye
FROM rust:1.81.0-slim-bookworm
COPY --from=sclevine/yj /bin/yj /bin/yj
RUN /bin/yj -h

View File

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

View File

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

View File

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

71
tests/python/test_auth.py Normal file
View File

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

View File

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

110
tests/python/utils.py Normal file
View File

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

View File

@@ -1,22 +1,33 @@
GEM
remote: https://rubygems.org/
specs:
activemodel (7.0.4.1)
activesupport (= 7.0.4.1)
activerecord (7.0.4.1)
activemodel (= 7.0.4.1)
activesupport (= 7.0.4.1)
activesupport (7.0.4.1)
activemodel (7.1.4)
activesupport (= 7.1.4)
activerecord (7.1.4)
activemodel (= 7.1.4)
activesupport (= 7.1.4)
timeout (>= 0.4.0)
activesupport (7.1.4)
base64
bigdecimal
concurrent-ruby (~> 1.0, >= 1.0.2)
connection_pool (>= 2.2.5)
drb
i18n (>= 1.6, < 2)
minitest (>= 5.1)
mutex_m
tzinfo (~> 2.0)
ast (2.4.2)
concurrent-ruby (1.1.10)
base64 (0.2.0)
bigdecimal (3.1.8)
concurrent-ruby (1.3.4)
connection_pool (2.4.1)
diff-lcs (1.5.0)
i18n (1.12.0)
drb (2.2.1)
i18n (1.14.5)
concurrent-ruby (~> 1.0)
minitest (5.17.0)
minitest (5.25.1)
mutex_m (0.2.0)
parallel (1.22.1)
parser (3.1.2.0)
ast (~> 2.4.1)
@@ -24,7 +35,8 @@ GEM
pg (1.3.2)
rainbow (3.1.1)
regexp_parser (2.3.1)
rexml (3.2.5)
rexml (3.3.6)
strscan
rspec (3.11.0)
rspec-core (~> 3.11.0)
rspec-expectations (~> 3.11.0)
@@ -50,10 +62,12 @@ GEM
rubocop-ast (1.17.0)
parser (>= 3.1.1.0)
ruby-progressbar (1.11.0)
strscan (3.1.0)
timeout (0.4.1)
toml (0.3.0)
parslet (>= 1.8.0, < 3.0.0)
toxiproxy (2.0.1)
tzinfo (2.0.5)
tzinfo (2.0.6)
concurrent-ruby (~> 1.0)
unicode-display_width (2.1.0)

View File

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

View File

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

682
tests/rust/Cargo.lock generated

File diff suppressed because it is too large Load Diff

View File

@@ -15,13 +15,11 @@ async fn test_prepared_statements() {
for _ in 0..5 {
let pool = pool.clone();
let handle = tokio::task::spawn(async move {
for _ in 0..1000 {
match sqlx::query("SELECT one").fetch_all(&pool).await {
for i in 0..1000 {
match sqlx::query(&format!("SELECT {:?}", i % 5)).fetch_all(&pool).await {
Ok(_) => (),
Err(err) => {
if err.to_string().contains("prepared statement") {
panic!("prepared statement error: {}", err);
}
panic!("prepared statement error: {}", err);
}
}
}