Compare commits

...

63 Commits

Author SHA1 Message Date
Lev Kokotov
a16699063e fmt 2023-06-17 08:57:19 -07:00
Juraj Bubniak
473bb3d17d Log not implemented messages as debug in prometheus metrics. (#477) 2023-06-16 18:48:38 -07:00
Lev Kokotov
c7d6273037 Support for prepared statements (#474)
* Start prepared statements

* parse

* Ok

* optional

* dont rewrite anonymous prepared stmts

* Dont rewrite anonymous prep statements

* hm?

* prep statements

* I see!

* comment

* Print config value

* Rewrite bind and add sqlx test

* fmt

* ok

* Fix

* Fix stats

* its late

* clean up PREPARE
2023-06-16 12:57:44 -07:00
Jeff Chen
94c781881f Report min_pool_size correctly (#471) 2023-06-12 09:23:56 -07:00
dependabot[bot]
a8c81e5df6 chore(deps): bump pin-project from 1.0.12 to 1.1.0 (#440)
Bumps [pin-project](https://github.com/taiki-e/pin-project) from 1.0.12 to 1.1.0.
- [Release notes](https://github.com/taiki-e/pin-project/releases)
- [Changelog](https://github.com/taiki-e/pin-project/blob/main/CHANGELOG.md)
- [Commits](https://github.com/taiki-e/pin-project/compare/v1.0.12...v1.1.0)

---
updated-dependencies:
- dependency-name: pin-project
  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>
2023-06-12 00:51:24 -07:00
dependabot[bot]
1d3746ec9e chore(deps): bump sqlparser from 0.33.0 to 0.34.0 (#448)
Bumps [sqlparser](https://github.com/sqlparser-rs/sqlparser-rs) from 0.33.0 to 0.34.0.
- [Changelog](https://github.com/sqlparser-rs/sqlparser-rs/blob/main/CHANGELOG.md)
- [Commits](https://github.com/sqlparser-rs/sqlparser-rs/compare/v0.33.0...v0.34.0)

---
updated-dependencies:
- dependency-name: sqlparser
  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>
2023-06-12 00:51:05 -07:00
dependabot[bot]
b5489dc1e6 chore(deps): bump regex from 1.8.1 to 1.8.4 (#466)
Bumps [regex](https://github.com/rust-lang/regex) from 1.8.1 to 1.8.4.
- [Release notes](https://github.com/rust-lang/regex/releases)
- [Changelog](https://github.com/rust-lang/regex/blob/master/CHANGELOG.md)
- [Commits](https://github.com/rust-lang/regex/compare/1.8.1...1.8.4)

---
updated-dependencies:
- dependency-name: regex
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2023-06-12 00:50:46 -07:00
dependabot[bot]
557b425fb1 chore(deps): bump log from 0.4.17 to 0.4.19 (#470)
Bumps [log](https://github.com/rust-lang/log) from 0.4.17 to 0.4.19.
- [Release notes](https://github.com/rust-lang/log/releases)
- [Changelog](https://github.com/rust-lang/log/blob/master/CHANGELOG.md)
- [Commits](https://github.com/rust-lang/log/compare/0.4.17...0.4.19)

---
updated-dependencies:
- dependency-name: log
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2023-06-12 00:50:26 -07:00
Zain Kabani
aca9738821 Make queue strategy configurable and default to Fifo (#463)
* Change idle timeout default to 10 minutes

* Revert lifo for now while we investigate connection thrashing issues

* Make queue strategy configurable

* test revert idle time out

* Add pgcat start to python test
2023-06-09 11:35:20 -07:00
Zain Kabani
0bc453a771 Change default server lifetime and bump bb8 version to use LIFO correctly (#462)
Change default server lifetime and idle timeouts and bump bb8 version to use LIFO correctly
2023-05-31 08:25:42 -07:00
Zain Kabani
b67c33b6d0 Use latest bb8 and use Lifo as the queue strategy in the pool (#455)
* Use git bb8

* Use latest bb8 and change pool is use stack
2023-05-28 19:46:13 -07:00
Mostafa Abdelraouf
a8a30ad43b Refactor Pool Stats to be based off of Server/Client stats (#445)
What is wrong
Stats reported by SHOW POOLS seem to be leaking. We see lingering cl_idle , cl_waiting, and similarly for sv_idle , sv_active. We confirmed that these are reporting issues not actual lingering clients.

This behavior is readily reproducible by running

while true; do
    psql "postgres://sharding_user:sharding_user@localhost:6432/sharded_db" -c "SELECT 1" > /dev/null 2>&1  &
done

Why it happens
I wasn't able to get to figure our the reason for the bug but my best guess is that we have race conditions when updating pool-level stats. So even though individual update operations are atomic, we perform a check then update sequence which is not protected by a guard.
https://github.com/postgresml/pgcat/blob/main/src/stats/pool.rs#L174-L179

I am also suspecting that using Relaxed ordering might allow this behavior (I changed all operations to use Ordering::SeqCst but still got lingering clients)

How to fix
Since SHOW POOLS/SHOW SERVER/SHOW CLIENTS all show the current state of the proxy (as opposed to SHOW STATS which show aggregate values), this PR refactors SHOW POOLS to have it construct the results directly from SHOW SERVER and SHOW CLIENT datasets. This reduces the complexity of stat updates and eliminates the need for having locks when updating pool stats as we only care about updating individual client/server states.

This will change the semantics of maxwait, so instead of it holding the maxwait time ever encountered by a client (connected or disconnected), it will only consider connected clients which should be okay given PgCat tends to hold on to client connections more than Pgbouncer.
2023-05-23 08:44:49 -05:00
dependabot[bot]
d63be9b93a chore(deps): bump toml from 0.7.3 to 0.7.4 (#447)
Bumps [toml](https://github.com/toml-rs/toml) from 0.7.3 to 0.7.4.
- [Commits](https://github.com/toml-rs/toml/compare/toml-v0.7.3...toml-v0.7.4)

---
updated-dependencies:
- dependency-name: toml
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2023-05-19 08:58:13 -07:00
Lev Kokotov
100778670c Ensure data makes it to the client (#446)
* Ensure data makes it to the client

* flush all buffers
2023-05-18 16:41:22 -07:00
Lev Kokotov
37e3349c24 Optionally clean up server connections (#444)
* Optionally clean up server connections

* move setting to pool

* fix test

* Print setting to screen

* fmt

* Fix pool_settings override in tests
2023-05-18 10:46:55 -07:00
Zain Kabani
7f57a89d75 Fix time based average stats (#442)
* keep track of current stats and zero them after updating averages

* Try tests

* typo

* remove commented test stuff

* Avoid dividing by zero

* Fix test

* refactor, get rid of iterator. do it manually

* trigger build

* Fix
2023-05-17 21:38:10 -07:00
Lev Kokotov
0898461c01 Allow to deploy pools without checking (#438) 2023-05-12 12:48:37 -07:00
Lev Kokotov
52b1b43850 Prewarmer (#435)
* Prewarmer

* hmm

* Tests

* default

* fix test

* Correct configuration

* Added minimal config example

* remove connect_timeout
2023-05-12 09:50:52 -07:00
Zain Kabani
0907f1b77f Improve logging for connection cleanup (#428)
* initial commit

* fix

* fmt
2023-05-11 17:40:10 -07:00
Zain Kabani
73260690b0 Fixes average stats bug (#436)
* Add test

* Fix test

* Add fix
2023-05-11 17:37:58 -07:00
Mostafa Abdelraouf
5056cbe8ed Fix docker-compose dev stack for Apple silicon (#432)
The docker-compose dev setup is broken under Apple silicon, starting the stack fails with the following error. Switching to a different docker image fixes the issue.
2023-05-10 10:24:35 -05:00
Lev Kokotov
571b02e178 Calculate averages correctly and preserve totals like before (#429)
* Reset totals after avg calculation

* like it used to be
2023-05-08 10:06:16 -07:00
Andrew Tanner
159eb89bf0 First try with role reset (#427)
* First try with role rest

* update

* extra line

* Update src/server.rs

Co-authored-by: Lev Kokotov <levkk@users.noreply.github.com>

* Update tests/ruby/misc_spec.rb

Co-authored-by: Lev Kokotov <levkk@users.noreply.github.com>

---------

Co-authored-by: Lev Kokotov <levkk@users.noreply.github.com>
2023-05-05 15:31:27 -07:00
Lev Kokotov
389993bf3e Accurate log messages (#425) 2023-05-05 08:27:19 -07:00
Lev Kokotov
ba5243b6dd Optionally validate config on boot (#423) 2023-05-03 17:07:23 -07:00
Lev Kokotov
128ef72911 lowercase config query (#422)
* lowercase config query

* remove debug
2023-05-03 16:47:20 -07:00
Lev Kokotov
811885f464 Actually plugins (#421)
* more plugins

* clean up

* fix tests

* fix flakey test
2023-05-03 16:13:45 -07:00
dependabot[bot]
d5e329fec5 chore(deps): bump regex from 1.8.0 to 1.8.1 (#413)
Bumps [regex](https://github.com/rust-lang/regex) from 1.8.0 to 1.8.1.
- [Release notes](https://github.com/rust-lang/regex/releases)
- [Changelog](https://github.com/rust-lang/regex/blob/master/CHANGELOG.md)
- [Commits](https://github.com/rust-lang/regex/commits/1.8.1)

---
updated-dependencies:
- dependency-name: regex
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2023-05-03 10:00:05 -07:00
Lev Kokotov
09e54e1175 Plugins! (#420)
* Some queries

* Plugins!!

* cleanup

* actual names

* the actual plugins

* comment

* fix tests

* Tests

* unused errors

* Increase reaper rate to actually enforce settings

* ok
2023-05-03 09:13:05 -07:00
dependabot[bot]
23819c8549 chore(deps): bump rustls from 0.21.0 to 0.21.1 (#419)
Bumps [rustls](https://github.com/rustls/rustls) from 0.21.0 to 0.21.1.
- [Release notes](https://github.com/rustls/rustls/releases)
- [Changelog](https://github.com/rustls/rustls/blob/main/RELEASE_NOTES.md)
- [Commits](https://github.com/rustls/rustls/compare/v/0.21.0...v/0.21.1)

---
updated-dependencies:
- dependency-name: rustls
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2023-05-02 07:32:44 -07:00
Jose Fernández
7dfbd993f2 Add dns_cache for server addresses as in pgbouncer (#249)
* Add dns_cache so server addresses are cached and invalidated when DNS changes.

Adds a module to deal with dns_cache feature. It's
main struct is CachedResolver, which is a simple thread safe
hostname <-> Ips cache with the ability to refresh resolutions
every `dns_max_ttl` seconds. This way, a client can check whether its
ip address has changed.

* Allow reloading dns cached

* Add documentation for dns_cached
2023-05-02 10:26:40 +02:00
Lev Kokotov
3601130ba1 Readme update (#418)
* Readme update

* m

* wording
2023-04-30 09:44:25 -07:00
Lev Kokotov
0d504032b2 Server TLS (#417)
* Server TLS

* Finish up TLS

* thats it

* diff

* remove dead code

* maybe?

* dirty shutdown

* skip flakey test

* remove unused error

* fetch config once
2023-04-30 09:41:46 -07:00
Lev Kokotov
4a87b4807d Add more pool settings (#416)
* Add some pool settings

* fmt
2023-04-26 16:33:26 -07:00
Shawn
cb5ff40a59 fix typo (#415)
chore: typo
2023-04-26 08:28:54 -07:00
dependabot[bot]
62b2d994c1 chore(deps): bump regex from 1.7.3 to 1.8.0 (#411)
Bumps [regex](https://github.com/rust-lang/regex) from 1.7.3 to 1.8.0.
- [Release notes](https://github.com/rust-lang/regex/releases)
- [Changelog](https://github.com/rust-lang/regex/blob/master/CHANGELOG.md)
- [Commits](https://github.com/rust-lang/regex/commits)

---
updated-dependencies:
- dependency-name: regex
  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>
2023-04-21 06:33:52 -07:00
Lev Kokotov
66805d7e77 README updates (#409)
* Better table

* add image

* promote auth passthrough to stable

* fmt
2023-04-20 07:53:55 -07:00
Lev Kokotov
4ccc1e7fa3 Fix CONFIG (#408)
Fix readme
2023-04-19 07:45:26 -07:00
Lev Kokotov
3dae3d0777 Separate server and client passwords optionally (#407)
* Separate server and user passwords

* config
2023-04-18 09:57:17 -07:00
dependabot[bot]
a18eb42df5 chore(deps): bump serde from 1.0.159 to 1.0.160 (#404)
Bumps [serde](https://github.com/serde-rs/serde) from 1.0.159 to 1.0.160.
- [Release notes](https://github.com/serde-rs/serde/releases)
- [Commits](https://github.com/serde-rs/serde/compare/v1.0.159...v1.0.160)

---
updated-dependencies:
- dependency-name: serde
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2023-04-14 10:25:00 -07:00
dependabot[bot]
6aacf1fa19 chore(deps): bump serde_derive from 1.0.159 to 1.0.160 (#403)
Bumps [serde_derive](https://github.com/serde-rs/serde) from 1.0.159 to 1.0.160.
- [Release notes](https://github.com/serde-rs/serde/releases)
- [Commits](https://github.com/serde-rs/serde/compare/v1.0.159...v1.0.160)

---
updated-dependencies:
- dependency-name: serde_derive
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2023-04-14 10:24:52 -07:00
dependabot[bot]
8e99e65215 chore(deps): bump sqlparser from 0.32.0 to 0.33.0 (#399)
Bumps [sqlparser](https://github.com/sqlparser-rs/sqlparser-rs) from 0.32.0 to 0.33.0.
- [Release notes](https://github.com/sqlparser-rs/sqlparser-rs/releases)
- [Changelog](https://github.com/sqlparser-rs/sqlparser-rs/blob/main/CHANGELOG.md)
- [Commits](https://github.com/sqlparser-rs/sqlparser-rs/compare/v0.32.0...v0.33.0)

---
updated-dependencies:
- dependency-name: sqlparser
  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>
2023-04-14 10:24:44 -07:00
dependabot[bot]
5dfbc102a9 chore(deps): bump hyper from 0.14.25 to 0.14.26 (#406)
Bumps [hyper](https://github.com/hyperium/hyper) from 0.14.25 to 0.14.26.
- [Release notes](https://github.com/hyperium/hyper/releases)
- [Changelog](https://github.com/hyperium/hyper/blob/v0.14.26/CHANGELOG.md)
- [Commits](https://github.com/hyperium/hyper/compare/v0.14.25...v0.14.26)

---
updated-dependencies:
- dependency-name: hyper
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2023-04-14 10:23:52 -07:00
Cluas
bae12fca99 feat: set keepalive for pgcat server itself (#402)
* feat: set keepalive for pgcat server self

* docs: note also set for client
2023-04-12 09:29:43 -07:00
Lev Kokotov
421c5d4b64 Load config on client connect (#401) 2023-04-11 10:32:48 -07:00
Kian-Meng Ang
d568739db9 Fix typos (#398)
Found via `typos --format brief`
2023-04-10 18:37:16 -07:00
Lev Kokotov
692353c839 A couple things (#397)
* Format cleanup

* fmt

* finally
2023-04-10 14:51:01 -07:00
Lev Kokotov
a62f6b0eea Fix port; add user pool mode (#395)
* Fix port; add user pool mode

* will probably break our session/transaction mode tests
2023-04-05 15:06:19 -07:00
dependabot[bot]
89e15f09b5 chore(deps): bump tokio-rustls from 0.23.4 to 0.24.0 (#394)
Bumps [tokio-rustls](https://github.com/tokio-rs/tls) from 0.23.4 to 0.24.0.
- [Release notes](https://github.com/tokio-rs/tls/releases)
- [Commits](https://github.com/tokio-rs/tls/commits)

---
updated-dependencies:
- dependency-name: tokio-rustls
  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>
2023-04-02 23:00:09 -07:00
Mostafa Abdelraouf
7ddd23b514 Protocol-level test helpers (#393)
I needed to have granular control over protocol message testing. For example, being able to send protocol messages one-by-one and then be able to inspect the results.

In order to do that, I created this low-level ruby client that can be used to send protocol messages in any order without blocking and also allows inspection of response messages.
2023-04-01 15:27:57 -05:00
dependabot[bot]
faa9c1f64a chore(deps): bump futures from 0.3.27 to 0.3.28 (#392)
Bumps [futures](https://github.com/rust-lang/futures-rs) from 0.3.27 to 0.3.28.
- [Release notes](https://github.com/rust-lang/futures-rs/releases)
- [Changelog](https://github.com/rust-lang/futures-rs/blob/master/CHANGELOG.md)
- [Commits](https://github.com/rust-lang/futures-rs/compare/0.3.27...0.3.28)

---
updated-dependencies:
- dependency-name: futures
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2023-03-31 09:35:01 -07:00
dependabot[bot]
9094988491 chore(deps): bump postgres-protocol from 0.6.4 to 0.6.5 (#391)
Bumps [postgres-protocol](https://github.com/sfackler/rust-postgres) from 0.6.4 to 0.6.5.
- [Release notes](https://github.com/sfackler/rust-postgres/releases)
- [Commits](https://github.com/sfackler/rust-postgres/compare/postgres-protocol-v0.6.4...postgres-protocol-v0.6.5)

---
updated-dependencies:
- dependency-name: postgres-protocol
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2023-03-31 09:34:51 -07:00
Jose Fernández
6f768a84ce Auth passthrough (auth_query) (#266)
* Add a new exec_simple_query method

This adds a new `exec_simple_query` method so we can make 'out of band'
queries to servers that don't interfere with pools at all.
In order to reuse startup code for making these simple queries,
we need to set the stats (`Reporter`) optional, so using these
simple queries wont interfere with stats.

* Add auth passthough (auth_query)

Adds a feature that allows setting auth passthrough for md5 auth.

It adds 3 new (general and pool) config parameters:

- `auth_query`: An string containing a query that will be executed on boot
to obtain the hash of a given user. This query have to use a placeholder `$1`,
so pgcat can replace it with the user its trying to fetch the hash from.
- `auth_query_user`: The user to use for connecting to the server and executing the
auth_query.
- `auth_query_password`: The password to use for connecting to the server and executing the
auth_query.

The configuration can be done either on the general config (so pools share them) or in a per-pool basis.

The behavior is, at boot time, when validating server connections, a hash is fetched per server
and stored in the pool. When new server connections are created, and no cleartext password is specified,
the obtained hash is used for creating them, if the hash could not be obtained for whatever reason, it retries
it.

When client authentication is tried, it uses cleartext passwords if specified, it not, it checks whether
we have query_auth set up, if so, it tries to use the obtained hash for making client auth. If there is no
hash (we could not obtain one when validating the connection), a new fetch is tried.

Once we have a hash, we authenticate using it against whathever the client has sent us, if there is a failure
we refetch the hash and retry auth (so password changes can be done).

The idea with this 'retrial' mechanism is to make it fault tolerant, so if for whatever reason hash could not be
obtained during connection validation, or the password has change, we can still connect later.

* Add documentation for Auth passthrough
2023-03-30 13:29:23 -07:00
dependabot[bot]
0757d7f3a0 chore(deps): bump serde from 1.0.158 to 1.0.159 (#386)
Bumps [serde](https://github.com/serde-rs/serde) from 1.0.158 to 1.0.159.
- [Release notes](https://github.com/serde-rs/serde/releases)
- [Commits](https://github.com/serde-rs/serde/compare/v1.0.158...v1.0.159)

---
updated-dependencies:
- dependency-name: serde
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2023-03-28 09:54:39 -07:00
dependabot[bot]
568f04feee chore(deps): bump serde_derive from 1.0.154 to 1.0.159 (#387)
Bumps [serde_derive](https://github.com/serde-rs/serde) from 1.0.154 to 1.0.159.
- [Release notes](https://github.com/serde-rs/serde/releases)
- [Commits](https://github.com/serde-rs/serde/compare/v1.0.154...v1.0.159)

---
updated-dependencies:
- dependency-name: serde_derive
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2023-03-28 09:54:31 -07:00
Jose Fernández
58ce76d9b9 Refactor stats to use atomics (#375)
* Refactor stats to use atomics

When we are dealing with a high number of connections, generated
stats cannot be consumed fast enough by the stats collector loop.
This makes the stats subsystem inconsistent and a log of
warning messages are thrown due to unregistered server/clients.

This change refactors the stats subsystem so it uses atomics:

- Now counters are handled using U64 atomics
- Event system is dropped and averages are calculated using a loop
  every 15 seconds.
- Now, instead of snapshots being generated ever second we keep track of servers/clients
  that have registered. Each pool/server/client has its own instance of the counter and
  makes changes directly, instead of adding an event that gets processed later.

* Manually mplement Hash/Eq in `config::Address` ignoring stats

* Add tests for client connection counters

* Allow connecting to dockerized dev pgcat from the host

* stats: Decrease cl_idle when idle socket disconnects
2023-03-28 17:19:37 +02:00
dependabot[bot]
9a2076a9eb chore(deps): bump futures from 0.3.26 to 0.3.27 (#356)
Bumps [futures](https://github.com/rust-lang/futures-rs) from 0.3.26 to 0.3.27.
- [Release notes](https://github.com/rust-lang/futures-rs/releases)
- [Changelog](https://github.com/rust-lang/futures-rs/blob/master/CHANGELOG.md)
- [Commits](https://github.com/rust-lang/futures-rs/compare/0.3.26...0.3.27)

---
updated-dependencies:
- dependency-name: futures
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2023-03-27 09:13:45 -07:00
dependabot[bot]
e7e7118725 chore(deps): bump hyper from 0.14.24 to 0.14.25 (#358)
Bumps [hyper](https://github.com/hyperium/hyper) from 0.14.24 to 0.14.25.
- [Release notes](https://github.com/hyperium/hyper/releases)
- [Changelog](https://github.com/hyperium/hyper/blob/v0.14.25/CHANGELOG.md)
- [Commits](https://github.com/hyperium/hyper/compare/v0.14.24...v0.14.25)

---
updated-dependencies:
- dependency-name: hyper
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2023-03-27 09:13:36 -07:00
dependabot[bot]
99f790cacf chore(deps): bump toml from 0.7.2 to 0.7.3 (#360)
Bumps [toml](https://github.com/toml-rs/toml) from 0.7.2 to 0.7.3.
- [Release notes](https://github.com/toml-rs/toml/releases)
- [Commits](https://github.com/toml-rs/toml/compare/toml-v0.7.2...toml-v0.7.3)

---
updated-dependencies:
- dependency-name: toml
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2023-03-27 09:13:25 -07:00
dependabot[bot]
434b0bb69e chore(deps): bump serde from 1.0.154 to 1.0.158 (#376)
Bumps [serde](https://github.com/serde-rs/serde) from 1.0.154 to 1.0.158.
- [Release notes](https://github.com/serde-rs/serde/releases)
- [Commits](https://github.com/serde-rs/serde/compare/v1.0.154...v1.0.158)

---
updated-dependencies:
- dependency-name: serde
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2023-03-27 09:13:13 -07:00
dependabot[bot]
714e043ef0 chore(deps): bump async-trait from 0.1.66 to 0.1.68 (#382)
Bumps [async-trait](https://github.com/dtolnay/async-trait) from 0.1.66 to 0.1.68.
- [Release notes](https://github.com/dtolnay/async-trait/releases)
- [Commits](https://github.com/dtolnay/async-trait/compare/0.1.66...0.1.68)

---
updated-dependencies:
- dependency-name: async-trait
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2023-03-27 09:12:54 -07:00
dependabot[bot]
863104aadd chore(deps): bump regex from 1.7.1 to 1.7.3 (#385)
Bumps [regex](https://github.com/rust-lang/regex) from 1.7.1 to 1.7.3.
- [Release notes](https://github.com/rust-lang/regex/releases)
- [Changelog](https://github.com/rust-lang/regex/blob/master/CHANGELOG.md)
- [Commits](https://github.com/rust-lang/regex/compare/1.7.1...1.7.3)

---
updated-dependencies:
- dependency-name: regex
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2023-03-27 09:12:43 -07:00
Lev Kokotov
7dd96141e3 Update README.md 2023-03-26 00:33:05 -07:00
63 changed files with 7972 additions and 1901 deletions

View File

@@ -46,6 +46,14 @@ jobs:
POSTGRES_PASSWORD: postgres
POSTGRES_INITDB_ARGS: --auth-local=scram-sha-256 --auth-host=scram-sha-256 --auth=scram-sha-256
- image: postgres:14
command: ["postgres", "-p", "10432", "-c", "shared_preload_libraries=pg_stat_statements"]
environment:
POSTGRES_USER: postgres
POSTGRES_DB: postgres
POSTGRES_PASSWORD: postgres
POSTGRES_INITDB_ARGS: --auth-local=md5 --auth-host=md5 --auth=md5
# Add steps to the job
# See: https://circleci.com/docs/2.0/configuration-reference/#steps
steps:

View File

@@ -39,7 +39,7 @@ log_client_connections = false
log_client_disconnections = false
# Reload config automatically if it changes.
autoreload = true
autoreload = 15000
# TLS
tls_certificate = ".circleci/server.cert"

View File

@@ -19,6 +19,7 @@ PGPASSWORD=postgres psql -e -h 127.0.0.1 -p 5432 -U postgres -f tests/sharding/q
PGPASSWORD=postgres psql -e -h 127.0.0.1 -p 7432 -U postgres -f tests/sharding/query_routing_setup.sql
PGPASSWORD=postgres psql -e -h 127.0.0.1 -p 8432 -U postgres -f tests/sharding/query_routing_setup.sql
PGPASSWORD=postgres psql -e -h 127.0.0.1 -p 9432 -U postgres -f tests/sharding/query_routing_setup.sql
PGPASSWORD=postgres psql -e -h 127.0.0.1 -p 10432 -U postgres -f tests/sharding/query_routing_setup.sql
PGPASSWORD=sharding_user pgbench -h 127.0.0.1 -U sharding_user shard0 -i
PGPASSWORD=sharding_user pgbench -h 127.0.0.1 -U sharding_user shard1 -i

14
.editorconfig Normal file
View File

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

2
.rustfmt.toml Normal file
View File

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

View File

@@ -49,6 +49,14 @@ default: 30000 # milliseconds
How long an idle connection with a server is left open (ms).
### server_lifetime
```
path: general.server_lifetime
default: 86400000 # 24 hours
```
Max connection lifetime before it's closed, even if actively used.
### idle_client_in_transaction_timeout
```
path: general.idle_client_in_transaction_timeout
@@ -108,7 +116,7 @@ If we should log client disconnections
### autoreload
```
path: general.autoreload
default: false
default: 15000
```
When set to true, PgCat reloads configs if it detects a change in the config file.
@@ -152,7 +160,7 @@ default: <UNSET>
example: "server.cert"
```
Path to TLS Certficate file to use for TLS connections
Path to TLS Certificate file to use for TLS connections
### tls_private_key
```
@@ -180,6 +188,22 @@ default: "admin_pass"
Password to access the virtual administrative database
### dns_cache_enabled
```
path: general.dns_cache_enabled
default: false
```
When enabled, ip resolutions for server connections specified using hostnames will be cached
and checked for changes every `dns_max_ttl` seconds. If a change in the host resolution is found
old ip connections are closed (gracefully) and new connections will start using new ip.
### dns_max_ttl
```
path: general.dns_max_ttl
default: 30
```
Specifies how often (in seconds) cached ip addresses for servers are rechecked (see `dns_cache_enabled`).
## `pools.<pool_name>` Section
### pool_mode
@@ -213,7 +237,7 @@ 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.
### query_parser_enabled (experimental)
### query_parser_enabled
```
path: pools.<pool_name>.query_parser_enabled
default: true
@@ -234,7 +258,7 @@ If the query parser is enabled and this setting is enabled, the primary will be
load balancing of read queries. Otherwise, the primary will only be used for write
queries. The primary can always be explicitly selected with our custom protocol.
### sharding_key_regex (experimental)
### sharding_key_regex
```
path: pools.<pool_name>.sharding_key_regex
default: <UNSET>
@@ -256,7 +280,40 @@ Current options:
`pg_bigint_hash`: PARTITION BY HASH (Postgres hashing function)
`sha1`: A hashing function based on SHA1
### automatic_sharding_key (experimental)
### auth_query
```
path: pools.<pool_name>.auth_query
default: <UNSET>
example: "SELECT $1"
```
Query to be sent to servers to obtain the hash used for md5 authentication. The connection will be
established using the database configured in the pool. This parameter is inherited by every pool
and can be redefined in pool configuration.
### auth_query_user
```
path: pools.<pool_name>.auth_query_user
default: <UNSET>
example: "sharding_user"
```
User to be used for connecting to servers to obtain the hash used for md5 authentication by sending the query
specified in `auth_query_user`. The connection will be established using the database configured in the pool.
This parameter is inherited by every pool and can be redefined in pool configuration.
### auth_query_password
```
path: pools.<pool_name>.auth_query_password
default: <UNSET>
example: "sharding_user"
```
Password to be used for connecting to servers to obtain the hash used for md5 authentication by sending the query
specified in `auth_query_user`. The connection will be established using the database configured in the pool.
This parameter is inherited by every pool and can be redefined in pool configuration.
### automatic_sharding_key
```
path: pools.<pool_name>.automatic_sharding_key
default: <UNSET>
@@ -289,7 +346,8 @@ path: pools.<pool_name>.users.<user_index>.username
default: "sharding_user"
```
Postgresql username
PostgreSQL username used to authenticate the user and connect to the server
if `server_username` is not set.
### password
```
@@ -297,7 +355,26 @@ path: pools.<pool_name>.users.<user_index>.password
default: "sharding_user"
```
Postgresql password
PostgreSQL password used to authenticate the user and connect to the server
if `server_password` is not set.
### server_username
```
path: pools.<pool_name>.users.<user_index>.server_username
default: <UNSET>
example: "another_user"
```
PostgreSQL username used to connect to the server.
### server_password
```
path: pools.<pool_name>.users.<user_index>.server_password
default: <UNSET>
example: "another_password"
```
PostgreSQL password used to connect to the server.
### pool_size
```
@@ -328,7 +405,7 @@ default: [["127.0.0.1", 5432, "primary"], ["localhost", 5432, "replica"]]
Array of servers in the shard, each server entry is an array of `[host, port, role]`
### mirrors (experimental)
### mirrors
```
path: pools.<pool_name>.shards.<shard_index>.mirrors
default: <UNSET>

578
Cargo.lock generated
View File

@@ -4,9 +4,9 @@ version = 3
[[package]]
name = "aho-corasick"
version = "0.7.20"
version = "1.0.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "cc936419f96fa211c1b9166887b38e5e40b19958e5b895be7c1f93adec7071ac"
checksum = "67fc08ce920c31afb70f013dcce1bfc3a3195de6a228474e45e1f145b36f8d04"
dependencies = [
"memchr",
]
@@ -27,14 +27,46 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "bddcadddf5e9015d310179a59bb28c4d4b9920ad0f11e8e14dbadf654890c9a6"
[[package]]
name = "async-trait"
version = "0.1.66"
name = "async-stream"
version = "0.3.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "b84f9ebcc6c1f5b8cb160f6990096a5c127f423fcb6e1ccc46c370cbdfb75dfc"
checksum = "dad5c83079eae9969be7fadefe640a1c566901f05ff91ab221de4b6f68d9507e"
dependencies = [
"async-stream-impl",
"futures-core",
]
[[package]]
name = "async-stream-impl"
version = "0.3.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "10f203db73a71dfa2fb6dd22763990fa26f3d2625a6da2da900d23b87d26be27"
dependencies = [
"proc-macro2",
"quote",
"syn",
"syn 1.0.109",
]
[[package]]
name = "async-trait"
version = "0.1.68"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "b9ccdd8f2a161be9bd5c023df56f1b2a0bd1d83872ae53b71a84a12c9bf6e842"
dependencies = [
"proc-macro2",
"quote",
"syn 2.0.9",
]
[[package]]
name = "atomic_enum"
version = "0.2.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "6227a8d6fdb862bcb100c4314d0d9579e5cd73fa6df31a2e6f6e1acd3c5f1207"
dependencies = [
"proc-macro2",
"quote",
"syn 1.0.109",
]
[[package]]
@@ -51,9 +83,9 @@ checksum = "a4a4ddaa51a5bc52a6948f74c06d20aaaddb71924eab79b8c97a8c556e942d6a"
[[package]]
name = "bb8"
version = "0.8.0"
version = "0.8.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "1627eccf3aa91405435ba240be23513eeca466b5dc33866422672264de061582"
checksum = "98b4b0f25f18bcdc3ac72bdb486ed0acf7e185221fd4dc985bc15db5800b0ba2"
dependencies = [
"async-trait",
"futures-channel",
@@ -83,6 +115,12 @@ version = "3.12.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "0d261e256854913907f67ed06efbc3338dfe6179796deefc1ff763fc1aee5535"
[[package]]
name = "byteorder"
version = "1.4.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "14c189c53d098945499cdfa7ecc63567cf3886b3332b312a5b4585d8d3a6a610"
[[package]]
name = "bytes"
version = "1.4.0"
@@ -175,7 +213,7 @@ dependencies = [
"proc-macro2",
"quote",
"scratch",
"syn",
"syn 1.0.109",
]
[[package]]
@@ -192,9 +230,15 @@ checksum = "086c685979a698443656e5cf7856c95c642295a38599f12fb1ff76fb28d19892"
dependencies = [
"proc-macro2",
"quote",
"syn",
"syn 1.0.109",
]
[[package]]
name = "data-encoding"
version = "2.3.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "3ee2393c4a91429dffb4bedf19f4d6abf27d8a732c8ce4980305d782e5426d57"
[[package]]
name = "digest"
version = "0.10.6"
@@ -206,6 +250,24 @@ dependencies = [
"subtle",
]
[[package]]
name = "either"
version = "1.8.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "7fcaabb2fef8c910e7f4c7ce9f67a1283a1715879a7c230ca9d6d1ae31f16d91"
[[package]]
name = "enum-as-inner"
version = "0.5.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "c9720bba047d567ffc8a3cba48bf19126600e249ab7f128e9233e6376976a116"
dependencies = [
"heck",
"proc-macro2",
"quote",
"syn 1.0.109",
]
[[package]]
name = "env_logger"
version = "0.10.0"
@@ -246,6 +308,12 @@ version = "1.1.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "de853764b47027c2e862a995c34978ffa63c1501f2e15f987ba11bd4f9bba193"
[[package]]
name = "fallible-iterator"
version = "0.2.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4443176a9f2c162692bd3d352d745ef9413eec5782a80d8fd6f8a1ac692a07f7"
[[package]]
name = "fnv"
version = "1.0.7"
@@ -253,10 +321,19 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "3f9eec918d3f24069decb9af1554cad7c880e2da24a9afd88aca000531ab82c1"
[[package]]
name = "futures"
version = "0.3.26"
name = "form_urlencoded"
version = "1.1.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "13e2792b0ff0340399d58445b88fd9770e3489eff258a4cbc1523418f12abf84"
checksum = "a9c384f161156f5260c24a097c56119f9be8c798586aecc13afbcbe7b7e26bf8"
dependencies = [
"percent-encoding",
]
[[package]]
name = "futures"
version = "0.3.28"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "23342abe12aba583913b2e62f22225ff9c950774065e4bfb61a19cd9770fec40"
dependencies = [
"futures-channel",
"futures-core",
@@ -269,9 +346,9 @@ dependencies = [
[[package]]
name = "futures-channel"
version = "0.3.26"
version = "0.3.28"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "2e5317663a9089767a1ec00a487df42e0ca174b61b4483213ac24448e4664df5"
checksum = "955518d47e09b25bbebc7a18df10b81f0c766eaf4c4f1cccef2fca5f2a4fb5f2"
dependencies = [
"futures-core",
"futures-sink",
@@ -279,15 +356,15 @@ dependencies = [
[[package]]
name = "futures-core"
version = "0.3.26"
version = "0.3.28"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ec90ff4d0fe1f57d600049061dc6bb68ed03c7d2fbd697274c41805dcb3f8608"
checksum = "4bca583b7e26f571124fe5b7561d49cb2868d79116cfa0eefce955557c6fee8c"
[[package]]
name = "futures-executor"
version = "0.3.26"
version = "0.3.28"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e8de0a35a6ab97ec8869e32a2473f4b1324459e14c29275d14b10cb1fd19b50e"
checksum = "ccecee823288125bd88b4d7f565c9e58e41858e47ab72e8ea2d64e93624386e0"
dependencies = [
"futures-core",
"futures-task",
@@ -296,38 +373,38 @@ dependencies = [
[[package]]
name = "futures-io"
version = "0.3.26"
version = "0.3.28"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "bfb8371b6fb2aeb2d280374607aeabfc99d95c72edfe51692e42d3d7f0d08531"
checksum = "4fff74096e71ed47f8e023204cfd0aa1289cd54ae5430a9523be060cdb849964"
[[package]]
name = "futures-macro"
version = "0.3.26"
version = "0.3.28"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "95a73af87da33b5acf53acfebdc339fe592ecf5357ac7c0a7734ab9d8c876a70"
checksum = "89ca545a94061b6365f2c7355b4b32bd20df3ff95f02da9329b34ccc3bd6ee72"
dependencies = [
"proc-macro2",
"quote",
"syn",
"syn 2.0.9",
]
[[package]]
name = "futures-sink"
version = "0.3.26"
version = "0.3.28"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f310820bb3e8cfd46c80db4d7fb8353e15dfff853a127158425f31e0be6c8364"
checksum = "f43be4fe21a13b9781a69afa4985b0f6ee0e1afab2c6f454a8cf30e2b2237b6e"
[[package]]
name = "futures-task"
version = "0.3.26"
version = "0.3.28"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "dcf79a1bf610b10f42aea489289c5a2c478a786509693b80cd39c44ccd936366"
checksum = "76d3d132be6c0e6aa1534069c705a74a5997a356c0dc2f86a47765e5617c5b65"
[[package]]
name = "futures-util"
version = "0.3.26"
version = "0.3.28"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "9c1d6de3acfef38d2be4b1f543f553131788603495be83da675e180c8d6b7bd1"
checksum = "26b01e40b772d54cf6c6d721c1d1abd0647a0106a12ecaa1c186273392a69533"
dependencies = [
"futures-channel",
"futures-core",
@@ -364,9 +441,9 @@ dependencies = [
[[package]]
name = "h2"
version = "0.3.15"
version = "0.3.17"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "5f9f29bc9dda355256b2916cf526ab02ce0aeaaaf2bad60d65ef3f12f11dd0f4"
checksum = "66b91535aa35fea1523ad1b86cb6b53c28e0ae566ba4a460f4457e936cad7c6f"
dependencies = [
"bytes",
"fnv",
@@ -387,6 +464,12 @@ version = "0.12.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8a9ee70c43aaf417c914396645a0fa852624801b24ebb7ae78fe8272889ac888"
[[package]]
name = "heck"
version = "0.4.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "2540771e65fc8cb83cd6e8a237f70c319bd5c29f78ed1084ba5d50eeac86f7f9"
[[package]]
name = "hermit-abi"
version = "0.2.6"
@@ -411,6 +494,17 @@ dependencies = [
"digest",
]
[[package]]
name = "hostname"
version = "0.3.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "3c731c3e10504cc8ed35cfe2f1db4c9274c3d35fa486e3b31df46f068ef3e867"
dependencies = [
"libc",
"match_cfg",
"winapi",
]
[[package]]
name = "http"
version = "0.2.9"
@@ -453,9 +547,9 @@ checksum = "9a3a5bfb195931eeb336b2a7b4d761daec841b97f947d34394601737a7bba5e4"
[[package]]
name = "hyper"
version = "0.14.24"
version = "0.14.26"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "5e011372fa0b68db8350aa7a248930ecc7839bf46d8485577d69f117a75f164c"
checksum = "ab302d72a6f11a3b910431ff93aae7e773078c769f0a3ef15fb9ec692ed147d4"
dependencies = [
"bytes",
"futures-channel",
@@ -499,6 +593,27 @@ dependencies = [
"cxx-build",
]
[[package]]
name = "idna"
version = "0.2.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "418a0a6fab821475f634efe3ccc45c013f742efe03d853e8d3355d5cb850ecf8"
dependencies = [
"matches",
"unicode-bidi",
"unicode-normalization",
]
[[package]]
name = "idna"
version = "0.3.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e14ddfc70884202db2244c223200c204c2bda1bc6e0998d11b5e024d657209e6"
dependencies = [
"unicode-bidi",
"unicode-normalization",
]
[[package]]
name = "indexmap"
version = "1.9.2"
@@ -519,6 +634,24 @@ dependencies = [
"windows-sys",
]
[[package]]
name = "ipconfig"
version = "0.3.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "bd302af1b90f2463a98fa5ad469fc212c8e3175a41c3068601bfa2727591c5be"
dependencies = [
"socket2",
"widestring",
"winapi",
"winreg",
]
[[package]]
name = "ipnet"
version = "2.5.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f88c5561171189e69df9d98bcf18fd5f9558300f7ea7b801eb8a0fd748bd8745"
[[package]]
name = "is-terminal"
version = "0.4.4"
@@ -531,6 +664,15 @@ dependencies = [
"windows-sys",
]
[[package]]
name = "itertools"
version = "0.10.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "b0fd2260e829bddf4cb6ea802289de2f86d6a7a690192fbe91b3f46e0f2c8473"
dependencies = [
"either",
]
[[package]]
name = "itoa"
version = "1.0.5"
@@ -566,6 +708,12 @@ dependencies = [
"wasm-bindgen",
]
[[package]]
name = "lazy_static"
version = "1.4.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e2abad23fbc42b3700f2f279844dc832adb2b2eb069b2df918f455c4e18cc646"
[[package]]
name = "libc"
version = "0.2.139"
@@ -581,6 +729,12 @@ dependencies = [
"cc",
]
[[package]]
name = "linked-hash-map"
version = "0.5.6"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "0717cef1bc8b636c6e1c1bbdefc09e6322da8a9321966e8928ef80d20f7f770f"
[[package]]
name = "linux-raw-sys"
version = "0.1.4"
@@ -599,13 +753,31 @@ dependencies = [
[[package]]
name = "log"
version = "0.4.17"
version = "0.4.19"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "abb12e687cfb44aa40f41fc3978ef76448f9b6038cad6aef4259d3c095a2382e"
checksum = "b06a4cde4c0f271a446782e3eff8de789548ce57dbc8eca9292c27f4a42004b4"
[[package]]
name = "lru-cache"
version = "0.1.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "31e24f1ad8321ca0e8a1e0ac13f23cb668e6f5466c2c57319f6a5cf1cc8e3b1c"
dependencies = [
"cfg-if",
"linked-hash-map",
]
[[package]]
name = "match_cfg"
version = "0.1.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ffbee8634e0d45d258acb448e7eaab3fce7a0a467395d4d9f228e3c1f01fb2e4"
[[package]]
name = "matches"
version = "0.1.9"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a3e378b66a060d48947b590737b30a1be76706c8dd7b8ba0f2fe3989c68a853f"
[[package]]
name = "md-5"
version = "0.10.5"
@@ -714,21 +886,30 @@ dependencies = [
"windows-sys",
]
[[package]]
name = "percent-encoding"
version = "2.2.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "478c572c3d73181ff3c2539045f6eb99e5491218eae919370993b890cdbdd98e"
[[package]]
name = "pgcat"
version = "1.0.0"
version = "1.0.2-alpha3"
dependencies = [
"arc-swap",
"async-trait",
"atomic_enum",
"base64",
"bb8",
"bytes",
"chrono",
"env_logger",
"exitcode",
"fallible-iterator",
"futures",
"hmac",
"hyper",
"itertools",
"jemallocator",
"log",
"md-5",
@@ -737,11 +918,15 @@ dependencies = [
"once_cell",
"parking_lot",
"phf",
"pin-project",
"postgres-protocol",
"rand",
"regex",
"rustls",
"rustls-pemfile",
"serde",
"serde_derive",
"serde_json",
"sha-1",
"sha2",
"socket2",
@@ -749,7 +934,10 @@ dependencies = [
"stringprep",
"tokio",
"tokio-rustls",
"tokio-test",
"toml",
"trust-dns-resolver",
"webpki-roots",
]
[[package]]
@@ -782,7 +970,7 @@ dependencies = [
"phf_shared",
"proc-macro2",
"quote",
"syn",
"syn 1.0.109",
]
[[package]]
@@ -794,6 +982,26 @@ dependencies = [
"siphasher",
]
[[package]]
name = "pin-project"
version = "1.1.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "c95a7476719eab1e366eaf73d0260af3021184f18177925b07f54b30089ceead"
dependencies = [
"pin-project-internal",
]
[[package]]
name = "pin-project-internal"
version = "1.1.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "39407670928234ebc5e6e580247dd567ad73a3578460c5990f9503df207e8f07"
dependencies = [
"proc-macro2",
"quote",
"syn 2.0.9",
]
[[package]]
name = "pin-project-lite"
version = "0.2.9"
@@ -806,6 +1014,24 @@ version = "0.1.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8b870d8c151b6f2fb93e84a13146138f05d02ed11c7e7c54f8826aaaf7c9f184"
[[package]]
name = "postgres-protocol"
version = "0.6.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "78b7fa9f396f51dffd61546fd8573ee20592287996568e6175ceb0f8699ad75d"
dependencies = [
"base64",
"byteorder",
"bytes",
"fallible-iterator",
"hmac",
"md-5",
"memchr",
"rand",
"sha2",
"stringprep",
]
[[package]]
name = "ppv-lite86"
version = "0.2.17"
@@ -814,18 +1040,24 @@ checksum = "5b40af805b3121feab8a3c29f04d8ad262fa8e0561883e7653e024ae4479e6de"
[[package]]
name = "proc-macro2"
version = "1.0.51"
version = "1.0.53"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "5d727cae5b39d21da60fa540906919ad737832fe0b1c165da3a34d6548c849d6"
checksum = "ba466839c78239c09faf015484e5cc04860f88242cff4d03eb038f04b4699b73"
dependencies = [
"unicode-ident",
]
[[package]]
name = "quote"
version = "1.0.23"
name = "quick-error"
version = "1.2.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8856d8364d252a14d474036ea1358d63c9e6965c8e5c1885c18f73d70bff9c7b"
checksum = "a1d01941d82fa2ab50be1e79e6714289dd7cde78eba4c074bc5a4374f650dfe0"
[[package]]
name = "quote"
version = "1.0.26"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4424af4bf778aae2051a77b60283332f386554255d722233d09fbfc7e30da2fc"
dependencies = [
"proc-macro2",
]
@@ -871,9 +1103,9 @@ dependencies = [
[[package]]
name = "regex"
version = "1.7.1"
version = "1.8.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "48aaa5748ba571fb95cd2c85c09f629215d3a6ece942baa100950af03a34f733"
checksum = "d0ab3ca65655bb1e41f2a8c8cd662eb4fb035e67c3f78da1d61dffe89d07300f"
dependencies = [
"aho-corasick",
"memchr",
@@ -882,9 +1114,19 @@ dependencies = [
[[package]]
name = "regex-syntax"
version = "0.6.28"
version = "0.7.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "456c603be3e8d448b072f410900c09faf164fbce2d480456f50eea6e25f9c848"
checksum = "436b050e76ed2903236f032a59761c1eb99e1b0aead2c257922771dab1fc8c78"
[[package]]
name = "resolv-conf"
version = "0.7.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "52e44394d2086d010551b14b53b1f24e31647570cd1deb0379e2c21b329aba00"
dependencies = [
"hostname",
"quick-error",
]
[[package]]
name = "ring"
@@ -917,14 +1159,14 @@ dependencies = [
[[package]]
name = "rustls"
version = "0.20.8"
version = "0.21.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "fff78fc74d175294f4e83b28343315ffcfb114b156f0185e9741cb5570f50e2f"
checksum = "c911ba11bc8433e811ce56fde130ccf32f5127cab0e0194e9c68c5a5b671791e"
dependencies = [
"log",
"ring",
"rustls-webpki",
"sct",
"webpki",
]
[[package]]
@@ -936,6 +1178,22 @@ dependencies = [
"base64",
]
[[package]]
name = "rustls-webpki"
version = "0.100.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d6207cd5ed3d8dca7816f8f3725513a34609c0c765bf652b8c3cb4cfd87db46b"
dependencies = [
"ring",
"untrusted",
]
[[package]]
name = "ryu"
version = "1.0.13"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f91339c0467de62360649f8d3e185ca8de4224ff281f66000de5eb2a77a79041"
[[package]]
name = "scopeguard"
version = "1.1.0"
@@ -960,26 +1218,40 @@ dependencies = [
[[package]]
name = "serde"
version = "1.0.154"
version = "1.0.160"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8cdd151213925e7f1ab45a9bbfb129316bd00799784b174b7cc7bcd16961c49e"
checksum = "bb2f3770c8bce3bcda7e149193a069a0f4365bda1fa5cd88e03bca26afc1216c"
dependencies = [
"serde_derive",
]
[[package]]
name = "serde_derive"
version = "1.0.154"
version = "1.0.160"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4fc80d722935453bcafdc2c9a73cd6fac4dc1938f0346035d84bf99fa9e33217"
checksum = "291a097c63d8497e00160b166a967a4a79c64f3facdd01cbd7502231688d77df"
dependencies = [
"proc-macro2",
"quote",
"syn",
"syn 2.0.9",
]
[[package]]
name = "serde_json"
version = "1.0.96"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "057d394a50403bcac12672b2b18fb387ab6d289d957dab67dd201875391e52f1"
dependencies = [
"itoa",
"ryu",
"serde",
]
[[package]]
name = "serde_spanned"
version = "0.6.1"
version = "0.6.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "0efd8caf556a6cebd3b285caf480045fcc1ac04f6bd786b09a6f11af30c4fcf4"
checksum = "93107647184f6027e3b7dcb2e11034cf95ffa1e3a682c67951963ac69c1c007d"
dependencies = [
"serde",
]
@@ -1054,11 +1326,23 @@ checksum = "6e63cff320ae2c57904679ba7cb63280a3dc4613885beafb148ee7bf9aa9042d"
[[package]]
name = "sqlparser"
version = "0.32.0"
version = "0.34.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "0366f270dbabb5cc2e4c88427dc4c08bba144f81e32fbd459a013f26a4d16aa0"
checksum = "37d3706eefb17039056234df6b566b0014f303f867f2656108334a55b8096f59"
dependencies = [
"log",
"sqlparser_derive",
]
[[package]]
name = "sqlparser_derive"
version = "0.1.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "55fe75cb4a364c7f7ae06c7dbbc8d84bddd85d6cdf9975963c3935bc1991761e"
dependencies = [
"proc-macro2",
"quote",
"syn 1.0.109",
]
[[package]]
@@ -1094,6 +1378,17 @@ dependencies = [
"unicode-ident",
]
[[package]]
name = "syn"
version = "2.0.9"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "0da4a3c17e109f700685ec577c0f85efd9b19bcf15c913985f14dc1ac01775aa"
dependencies = [
"proc-macro2",
"quote",
"unicode-ident",
]
[[package]]
name = "termcolor"
version = "1.2.0"
@@ -1103,6 +1398,26 @@ dependencies = [
"winapi-util",
]
[[package]]
name = "thiserror"
version = "1.0.37"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "10deb33631e3c9018b9baf9dcbbc4f737320d2b576bac10f6aefa048fa407e3e"
dependencies = [
"thiserror-impl",
]
[[package]]
name = "thiserror-impl"
version = "1.0.37"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "982d17546b47146b28f7c22e3d08465f6b8903d0ea13c1660d9d84a6e7adcdbb"
dependencies = [
"proc-macro2",
"quote",
"syn 1.0.109",
]
[[package]]
name = "time"
version = "0.1.45"
@@ -1157,18 +1472,41 @@ checksum = "d266c00fde287f55d3f1c3e96c500c362a2b8c695076ec180f27918820bc6df8"
dependencies = [
"proc-macro2",
"quote",
"syn",
"syn 1.0.109",
]
[[package]]
name = "tokio-rustls"
version = "0.23.4"
version = "0.24.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "c43ee83903113e03984cb9e5cebe6c04a5116269e900e3ddba8f068a62adda59"
checksum = "e0d409377ff5b1e3ca6437aa86c1eb7d40c134bfec254e44c830defa92669db5"
dependencies = [
"rustls",
"tokio",
"webpki",
]
[[package]]
name = "tokio-stream"
version = "0.1.11"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d660770404473ccd7bc9f8b28494a811bc18542b915c0855c51e8f419d5223ce"
dependencies = [
"futures-core",
"pin-project-lite",
"tokio",
]
[[package]]
name = "tokio-test"
version = "0.4.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "53474327ae5e166530d17f2d956afcb4f8a004de581b3cae10f12006bc8163e3"
dependencies = [
"async-stream",
"bytes",
"futures-core",
"tokio",
"tokio-stream",
]
[[package]]
@@ -1187,9 +1525,9 @@ dependencies = [
[[package]]
name = "toml"
version = "0.7.2"
version = "0.7.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f7afcae9e3f0fe2c370fd4657108972cbb2fa9db1b9f84849cefd80741b01cb6"
checksum = "d6135d499e69981f9ff0ef2167955a5333c35e36f6937d382974566b3d5b94ec"
dependencies = [
"serde",
"serde_spanned",
@@ -1199,18 +1537,18 @@ dependencies = [
[[package]]
name = "toml_datetime"
version = "0.6.1"
version = "0.6.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "3ab8ed2edee10b50132aed5f331333428b011c99402b5a534154ed15746f9622"
checksum = "5a76a9312f5ba4c2dec6b9161fdf25d87ad8a09256ccea5a556fef03c706a10f"
dependencies = [
"serde",
]
[[package]]
name = "toml_edit"
version = "0.19.4"
version = "0.19.9"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "9a1eb0622d28f4b9c90adc4ea4b2b46b47663fde9ac5fafcb14a1369d5508825"
checksum = "92d964908cec0d030b812013af25a0e57fddfadb1e066ecc6681d86253129d4f"
dependencies = [
"indexmap",
"serde",
@@ -1233,9 +1571,21 @@ checksum = "8ce8c33a8d48bd45d624a6e523445fd21ec13d3653cd51f681abf67418f54eb8"
dependencies = [
"cfg-if",
"pin-project-lite",
"tracing-attributes",
"tracing-core",
]
[[package]]
name = "tracing-attributes"
version = "0.1.23"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4017f8f45139870ca7e672686113917c71c7a6e02d4924eda67186083c03081a"
dependencies = [
"proc-macro2",
"quote",
"syn 1.0.109",
]
[[package]]
name = "tracing-core"
version = "0.1.30"
@@ -1245,6 +1595,51 @@ dependencies = [
"once_cell",
]
[[package]]
name = "trust-dns-proto"
version = "0.22.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4f7f83d1e4a0e4358ac54c5c3681e5d7da5efc5a7a632c90bb6d6669ddd9bc26"
dependencies = [
"async-trait",
"cfg-if",
"data-encoding",
"enum-as-inner",
"futures-channel",
"futures-io",
"futures-util",
"idna 0.2.3",
"ipnet",
"lazy_static",
"rand",
"smallvec",
"thiserror",
"tinyvec",
"tokio",
"tracing",
"url",
]
[[package]]
name = "trust-dns-resolver"
version = "0.22.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "aff21aa4dcefb0a1afbfac26deb0adc93888c7d295fb63ab273ef276ba2b7cfe"
dependencies = [
"cfg-if",
"futures-util",
"ipconfig",
"lazy_static",
"lru-cache",
"parking_lot",
"resolv-conf",
"smallvec",
"thiserror",
"tokio",
"tracing",
"trust-dns-proto",
]
[[package]]
name = "try-lock"
version = "0.2.4"
@@ -1290,6 +1685,17 @@ version = "0.7.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a156c684c91ea7d62626509bce3cb4e1d9ed5c4d978f7b4352658f96a4c26b4a"
[[package]]
name = "url"
version = "2.3.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "0d68c799ae75762b8c3fe375feb6600ef5602c883c5d21eb51c09f22b83c4643"
dependencies = [
"form_urlencoded",
"idna 0.3.0",
"percent-encoding",
]
[[package]]
name = "version_check"
version = "0.9.4"
@@ -1339,7 +1745,7 @@ dependencies = [
"once_cell",
"proc-macro2",
"quote",
"syn",
"syn 1.0.109",
"wasm-bindgen-shared",
]
@@ -1361,7 +1767,7 @@ checksum = "2aff81306fcac3c7515ad4e177f521b5c9a15f2b08f4e32d823066102f35a5f6"
dependencies = [
"proc-macro2",
"quote",
"syn",
"syn 1.0.109",
"wasm-bindgen-backend",
"wasm-bindgen-shared",
]
@@ -1383,15 +1789,20 @@ dependencies = [
]
[[package]]
name = "webpki"
version = "0.22.0"
name = "webpki-roots"
version = "0.23.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f095d78192e208183081cc07bc5515ef55216397af48b873e5edcd72637fa1bd"
checksum = "aa54963694b65584e170cf5dc46aeb4dcaa5584e652ff5f3952e56d66aff0125"
dependencies = [
"ring",
"untrusted",
"rustls-webpki",
]
[[package]]
name = "widestring"
version = "0.5.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "17882f045410753661207383517a6f62ec3dbeb6a4ed2acce01f0728238d1983"
[[package]]
name = "winapi"
version = "0.3.9"
@@ -1491,9 +1902,18 @@ checksum = "447660ad36a13288b1db4d4248e857b510e8c3a225c822ba4fb748c0aafecffd"
[[package]]
name = "winnow"
version = "0.3.3"
version = "0.4.6"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "faf09497b8f8b5ac5d3bb4d05c0a99be20f26fd3d5f2db7b0716e946d5103658"
checksum = "61de7bac303dc551fe038e2b3cef0f571087a47571ea6e79a87692ac99b99699"
dependencies = [
"memchr",
]
[[package]]
name = "winreg"
version = "0.10.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "80d0f4e272c85def139476380b12f9ac60926689dd2e01d4923222f40580869d"
dependencies = [
"winapi",
]

View File

@@ -1,26 +1,25 @@
[package]
name = "pgcat"
version = "1.0.0"
version = "1.0.2-alpha3"
edition = "2021"
# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
[dependencies]
tokio = { version = "1", features = ["full"] }
bytes = "1"
md-5 = "0.10"
bb8 = "0.8.0"
bb8 = "0.8.1"
async-trait = "0.1"
rand = "0.8"
chrono = "0.4"
sha-1 = "0.10"
toml = "0.7"
serde = "1"
serde = { version = "1", features = ["derive"] }
serde_derive = "1"
regex = "1"
num_cpus = "1"
once_cell = "1"
sqlparser = "0.32.0"
sqlparser = {version = "0.34", features = ["visitor"] }
log = "0.4"
arc-swap = "1"
env_logger = "0.10"
@@ -29,7 +28,7 @@ hmac = "0.12"
sha2 = "0.10"
base64 = "0.21"
stringprep = "0.1"
tokio-rustls = "0.23"
tokio-rustls = "0.24"
rustls-pemfile = "1"
hyper = { version = "0.14", features = ["full"] }
phf = { version = "0.11.1", features = ["macros"] }
@@ -37,6 +36,17 @@ exitcode = "1.1.2"
futures = "0.3"
socket2 = { version = "0.4.7", features = ["all"] }
nix = "0.26.2"
atomic_enum = "0.2.0"
postgres-protocol = "0.6.5"
fallible-iterator = "0.2"
pin-project = "1"
webpki-roots = "0.23"
rustls = { version = "0.21", features = ["dangerous_configuration"] }
trust-dns-resolver = "0.22.0"
tokio-test = "0.4.2"
serde_json = "1"
itertools = "0.10"
[target.'cfg(not(target_env = "msvc"))'.dependencies]
jemallocator = "0.5.0"

View File

@@ -18,23 +18,56 @@ PostgreSQL pooler and proxy (like PgBouncer) with support for sharding, load bal
| Failover | **Stable** | Queries are automatically rerouted around broken replicas, validated by regular health checks. |
| Admin database statistics | **Stable** | Pooler statistics and administration via the `pgbouncer` and `pgcat` databases. |
| Prometheus statistics | **Stable** | Statistics are reported via a HTTP endpoint for Prometheus. |
| Client TLS | **Stable** | Clients can connect to the pooler using TLS/SSL. |
| SSL/TLS | **Stable** | Clients can connect to the pooler using TLS. Pooler can connect to Postgres servers using TLS. |
| Client/Server authentication | **Stable** | Clients can connect using MD5 authentication, supported by `libpq` and all Postgres client drivers. PgCat can connect to Postgres using MD5 and SCRAM-SHA-256. |
| Live configuration reloading | **Stable** | Identical to PgBouncer; all settings can be reloaded dynamically (except `host` and `port`). |
| Auth passthrough | **Stable** | MD5 password authentication can be configured to use an `auth_query` so no cleartext passwords are needed in the config file.|
| Sharding using extended SQL syntax | **Experimental** | Clients can dynamically configure the pooler to route queries to specific shards. |
| Sharding using comments parsing/Regex | **Experimental** | Clients can include shard information (sharding key, shard ID) in the query comments. |
| Automatic sharding | **Experimental** | PgCat can parse queries detect sharding keys automatically, and route queries to the route shard. |
| Automatic sharding | **Experimental** | PgCat can parse queries, detect sharding keys automatically, and route queries to the correct shard. |
| Mirroring | **Experimental** | Mirror queries between multiple databases in order to test servers with realistic production traffic. |
## Status
PgCat is stable and used in production to serve hundreds of thousands of queries per second. Some features remain experimental and are being actively developed. They are optional and can be enabled through configuration.
PgCat is stable and used in production to serve hundreds of thousands of queries per second.
| | |
|-|-|
|<a href="https://tech.instacart.com/adopting-pgcat-a-nextgen-postgres-proxy-3cf284e68c2f"><img src="./images/instacart.webp" height="70" width="auto"></a>|<a href="https://postgresml.org/blog/scaling-postgresml-to-one-million-requests-per-second"><img src="./images/postgresml.webp" height="70" width="auto"></a>|
| [Instacart](https://tech.instacart.com/adopting-pgcat-a-nextgen-postgres-proxy-3cf284e68c2f) | [PostgresML](https://postgresml.org/blog/scaling-postgresml-to-one-million-requests-per-second) |
<table>
<tr>
<td>
<a href="https://tech.instacart.com/adopting-pgcat-a-nextgen-postgres-proxy-3cf284e68c2f">
<img src="./images/instacart.webp" height="70" width="auto">
</a>
</td>
<td>
<a href="https://postgresml.org/blog/scaling-postgresml-to-one-million-requests-per-second">
<img src="./images/postgresml.webp" height="70" width="auto">
</a>
</td>
<td>
<a href="https://onesignal.com">
<img src="./images/one_signal.webp" height="70" width="auto">
</a>
</td>
</tr>
<tr>
<td>
<a href="https://tech.instacart.com/adopting-pgcat-a-nextgen-postgres-proxy-3cf284e68c2f">
Instacart
</a>
</td>
<td>
<a href="https://postgresml.org/blog/scaling-postgresml-to-one-million-requests-per-second">
PostgresML
</a>
</td>
<td>
OneSignal
</td>
</tr>
</table>
Some features remain experimental and are being actively developed. They are optional and can be enabled through configuration.
## Deployment
@@ -79,7 +112,7 @@ pgbench -t 1000 -p 6432 -h 127.0.0.1 --protocol extended
See [sharding README](./tests/sharding/README.md) for sharding logic testing.
Additionally, all features are tested with Ruby, Python, and Rust tests unit and integration tests.
Additionally, all features are tested with Ruby, Python, and Rust unit and integration tests.
Run `cargo test` to run Rust unit tests.
@@ -98,7 +131,7 @@ You can open a Docker development environment where you can debug tests easier.
./dev/script/console
```
This will open a terminal in an environment similar to that used in tests. In there, you can compile the pooler, run tests, do some debugging with the test environment, etc. Objects compiled inside the contaner (and bundled gems) will be placed in `dev/cache` so they don't interfere with what you have on your machine.
This will open a terminal in an environment similar to that used in tests. In there, you can compile the pooler, run tests, do some debugging with the test environment, etc. Objects compiled inside the container (and bundled gems) will be placed in `dev/cache` so they don't interfere with what you have on your machine.
## Usage

View File

@@ -1,4 +1,4 @@
FROM rust:bullseye
FROM rust:1.70-bullseye
# Dependencies
RUN apt-get update -y \

View File

@@ -25,7 +25,9 @@ x-common-env-pg:
services:
main:
image: kubernetes/pause
image: gcr.io/google_containers/pause:3.2
ports:
- 6432
pg1:
<<: *common-definition-pg
@@ -56,6 +58,13 @@ services:
POSTGRES_INITDB_ARGS: --auth-local=scram-sha-256 --auth-host=scram-sha-256 --auth=scram-sha-256
PGPORT: 9432
command: ["postgres", "-p", "9432", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-c", "pg_stat_statements.max=100000"]
pg5:
<<: *common-definition-pg
environment:
<<: *common-env-pg
POSTGRES_INITDB_ARGS: --auth-local=md5 --auth-host=md5 --auth=md5
PGPORT: 10432
command: ["postgres", "-p", "10432", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-c", "pg_stat_statements.max=100000"]
toxiproxy:
build: .
@@ -69,6 +78,7 @@ services:
- pg2
- pg3
- pg4
- pg5
pgcat-shell:
stdin_open: true

View File

@@ -38,9 +38,6 @@ log_client_connections = false
# If we should log client disconnections
log_client_disconnections = false
# Reload config automatically if it changes.
autoreload = false
# TLS
# tls_certificate = "server.cert"
# tls_private_key = "server.key"
@@ -76,7 +73,7 @@ query_parser_enabled = true
# If the query parser is enabled and this setting is enabled, the primary will be part of the pool of databases used for
# load balancing of read queries. Otherwise, the primary will only be used for write
# queries. The primary can always be explicitely selected with our custom protocol.
# queries. The primary can always be explicitly selected with our custom protocol.
primary_reads_enabled = true
# So what if you wanted to implement a different hashing function,

BIN
images/one_signal.webp Normal file

Binary file not shown.

After

Width:  |  Height:  |  Size: 16 KiB

22
pgcat.minimal.toml Normal file
View File

@@ -0,0 +1,22 @@
# This is an example of the most basic config
# that will mimic what PgBouncer does in transaction mode with one server.
[general]
host = "0.0.0.0"
port = 6433
admin_username = "pgcat"
admin_password = "pgcat"
[pools.pgml.users.0]
username = "postgres"
password = "postgres"
pool_size = 10
min_pool_size = 1
pool_mode = "transaction"
[pools.pgml.shards.0]
servers = [
["127.0.0.1", 28815, "primary"]
]
database = "postgres"

View File

@@ -23,6 +23,9 @@ connect_timeout = 5000 # milliseconds
# How long an idle connection with a server is left open (ms).
idle_timeout = 30000 # milliseconds
# Max connection lifetime before it's closed, even if actively used.
server_lifetime = 86400000 # 24 hours
# How long a client is allowed to be idle while in a transaction (ms).
idle_client_in_transaction_timeout = 0 # milliseconds
@@ -45,7 +48,7 @@ log_client_connections = false
log_client_disconnections = false
# When set to true, PgCat reloads configs if it detects a change in the config file.
autoreload = false
autoreload = 15000
# Number of worker threads the Runtime will use (4 by default).
worker_threads = 5
@@ -57,10 +60,19 @@ tcp_keepalives_count = 5
# Number of seconds between keepalive packets.
tcp_keepalives_interval = 5
# Path to TLS Certficate file to use for TLS connections
# tls_certificate = "server.cert"
# Handle prepared statements.
prepared_statements = true
# Path to TLS Certificate file to use for TLS connections
# tls_certificate = ".circleci/server.cert"
# Path to TLS private key file to use for TLS connections
# tls_private_key = "server.key"
# tls_private_key = ".circleci/server.key"
# Enable/disable server TLS
server_tls = false
# Verify server certificate is completely authentic.
verify_server_certificate = false
# User name to access the virtual administrative database (pgbouncer or pgcat)
# Connecting to that database allows running commands like `SHOW POOLS`, `SHOW DATABASES`, etc..
@@ -68,6 +80,58 @@ admin_username = "admin_user"
# Password to access the virtual administrative database
admin_password = "admin_pass"
# Default plugins that are configured on all pools.
[plugins]
# Prewarmer plugin that runs queries on server startup, before giving the connection
# to the client.
[plugins.prewarmer]
enabled = false
queries = [
"SELECT pg_prewarm('pgbench_accounts')",
]
# Log all queries to stdout.
[plugins.query_logger]
enabled = false
# Block access to tables that Postgres does not allow us to control.
[plugins.table_access]
enabled = false
tables = [
"pg_user",
"pg_roles",
"pg_database",
]
# Intercept user queries and give a fake reply.
[plugins.intercept]
enabled = true
[plugins.intercept.queries.0]
query = "select current_database() as a, current_schemas(false) as b"
schema = [
["a", "text"],
["b", "text"],
]
result = [
["${DATABASE}", "{public}"],
]
[plugins.intercept.queries.1]
query = "select current_database(), current_schema(), current_user"
schema = [
["current_database", "text"],
["current_schema", "text"],
["current_user", "text"],
]
result = [
["${DATABASE}", "public", "${USER}"],
]
# pool configs are structured as pool.<pool_name>
# the pool_name is what clients use as database name when connecting.
# For a pool named `sharded_db`, clients access that pool using connection string like
@@ -113,6 +177,21 @@ primary_reads_enabled = true
# `sha1`: A hashing function based on SHA1
sharding_function = "pg_bigint_hash"
# Query to be sent to servers to obtain the hash used for md5 authentication. The connection will be
# established using the database configured in the pool. This parameter is inherited by every pool
# and can be redefined in pool configuration.
# auth_query = "SELECT $1"
# User to be used for connecting to servers to obtain the hash used for md5 authentication by sending the query
# specified in `auth_query_user`. The connection will be established using the database configured in the pool.
# This parameter is inherited by every pool and can be redefined in pool configuration.
# auth_query_user = "sharding_user"
# Password to be used for connecting to servers to obtain the hash used for md5 authentication by sending the query
# specified in `auth_query_user`. The connection will be established using the database configured in the pool.
# This parameter is inherited by every pool and can be redefined in pool configuration.
# auth_query_password = "sharding_user"
# Automatically parse this from queries and route queries to the right shard!
# automatic_sharding_key = "data.id"
@@ -122,18 +201,86 @@ idle_timeout = 40000
# Connect timeout can be overwritten in the pool
connect_timeout = 3000
# When enabled, ip resolutions for server connections specified using hostnames will be cached
# and checked for changes every `dns_max_ttl` seconds. If a change in the host resolution is found
# old ip connections are closed (gracefully) and new connections will start using new ip.
# dns_cache_enabled = false
# Specifies how often (in seconds) cached ip addresses for servers are rechecked (see `dns_cache_enabled`).
# dns_max_ttl = 30
# Plugins can be configured on a pool-per-pool basis. This overrides the global plugins setting,
# so all plugins have to be configured here again.
[pool.sharded_db.plugins]
[pools.sharded_db.plugins.prewarmer]
enabled = true
queries = [
"SELECT pg_prewarm('pgbench_accounts')",
]
[pools.sharded_db.plugins.query_logger]
enabled = false
[pools.sharded_db.plugins.table_access]
enabled = false
tables = [
"pg_user",
"pg_roles",
"pg_database",
]
[pools.sharded_db.plugins.intercept]
enabled = true
[pools.sharded_db.plugins.intercept.queries.0]
query = "select current_database() as a, current_schemas(false) as b"
schema = [
["a", "text"],
["b", "text"],
]
result = [
["${DATABASE}", "{public}"],
]
[pools.sharded_db.plugins.intercept.queries.1]
query = "select current_database(), current_schema(), current_user"
schema = [
["current_database", "text"],
["current_schema", "text"],
["current_user", "text"],
]
result = [
["${DATABASE}", "public", "${USER}"],
]
# User configs are structured as pool.<pool_name>.users.<user_index>
# This secion holds the credentials for users that may connect to this cluster
# This section holds the credentials for users that may connect to this cluster
[pools.sharded_db.users.0]
# Postgresql username
# PostgreSQL username used to authenticate the user and connect to the server
# if `server_username` is not set.
username = "sharding_user"
# Postgresql password
# PostgreSQL password used to authenticate the user and connect to the server
# if `server_password` is not set.
password = "sharding_user"
pool_mode = "session"
# PostgreSQL username used to connect to the server.
# server_username = "another_user"
# PostgreSQL password used to connect to the server.
# server_password = "another_password"
# Maximum number of server connections that can be established for this user
# The maximum number of connection from a single Pgcat process to any database in the cluster
# is the sum of pool_size across all users.
pool_size = 9
# Maximum query duration. Dangerous, but protects against DBs that died in a non-obvious way.
# 0 means it is disabled.
statement_timeout = 0
@@ -178,6 +325,8 @@ sharding_function = "pg_bigint_hash"
username = "simple_user"
password = "simple_user"
pool_size = 5
min_pool_size = 3
server_lifetime = 60000
statement_timeout = 0
[pools.simple_db.shards.0]

View File

@@ -1,21 +1,21 @@
use crate::pool::BanReason;
/// Admin database.
use crate::stats::pool::PoolStats;
use bytes::{Buf, BufMut, BytesMut};
use log::{error, info, trace};
use nix::sys::signal::{self, Signal};
use nix::unistd::Pid;
use std::collections::HashMap;
/// Admin database.
use std::sync::atomic::Ordering;
use std::time::{SystemTime, UNIX_EPOCH};
use tokio::time::Instant;
use crate::config::{get_config, reload_config, VERSION};
use crate::errors::Error;
use crate::messages::*;
use crate::pool::ClientServerMap;
use crate::pool::{get_all_pools, get_pool};
use crate::stats::{
get_address_stats, get_client_stats, get_pool_stats, get_server_stats, ClientState, ServerState,
};
use crate::ClientServerMap;
use crate::stats::{get_client_stats, get_server_stats, ClientState, ServerState};
pub fn generate_server_info_for_admin() -> BytesMut {
let mut server_info = BytesMut::new();
@@ -158,7 +158,14 @@ where
"free_clients".to_string(),
client_stats
.keys()
.filter(|client_id| client_stats.get(client_id).unwrap().state == ClientState::Idle)
.filter(|client_id| {
client_stats
.get(client_id)
.unwrap()
.state
.load(Ordering::Relaxed)
== ClientState::Idle
})
.count()
.to_string(),
]));
@@ -166,7 +173,14 @@ where
"used_clients".to_string(),
client_stats
.keys()
.filter(|client_id| client_stats.get(client_id).unwrap().state == ClientState::Active)
.filter(|client_id| {
client_stats
.get(client_id)
.unwrap()
.state
.load(Ordering::Relaxed)
== ClientState::Active
})
.count()
.to_string(),
]));
@@ -178,7 +192,14 @@ where
"free_servers".to_string(),
server_stats
.keys()
.filter(|server_id| server_stats.get(server_id).unwrap().state == ServerState::Idle)
.filter(|server_id| {
server_stats
.get(server_id)
.unwrap()
.state
.load(Ordering::Relaxed)
== ServerState::Idle
})
.count()
.to_string(),
]));
@@ -186,7 +207,14 @@ where
"used_servers".to_string(),
server_stats
.keys()
.filter(|server_id| server_stats.get(server_id).unwrap().state == ServerState::Active)
.filter(|server_id| {
server_stats
.get(server_id)
.unwrap()
.state
.load(Ordering::Relaxed)
== ServerState::Active
})
.count()
.to_string(),
]));
@@ -227,52 +255,12 @@ async fn show_pools<T>(stream: &mut T) -> Result<(), Error>
where
T: tokio::io::AsyncWrite + std::marker::Unpin,
{
let all_pool_stats = get_pool_stats();
let columns = vec![
("database", DataType::Text),
("user", DataType::Text),
("pool_mode", DataType::Text),
("cl_idle", DataType::Numeric),
("cl_active", DataType::Numeric),
("cl_waiting", DataType::Numeric),
("cl_cancel_req", DataType::Numeric),
("sv_active", DataType::Numeric),
("sv_idle", DataType::Numeric),
("sv_used", DataType::Numeric),
("sv_tested", DataType::Numeric),
("sv_login", DataType::Numeric),
("maxwait", DataType::Numeric),
("maxwait_us", DataType::Numeric),
];
let pool_lookup = PoolStats::construct_pool_lookup();
let mut res = BytesMut::new();
res.put(row_description(&columns));
for (user_pool, pool) in get_all_pools() {
let def = HashMap::default();
let pool_stats = all_pool_stats
.get(&(user_pool.db.clone(), user_pool.user.clone()))
.unwrap_or(&def);
let pool_config = &pool.settings;
let mut row = vec![
user_pool.db.clone(),
user_pool.user.clone(),
pool_config.pool_mode.to_string(),
];
for column in &columns[3..columns.len()] {
let value = match column.0 {
"maxwait" => (pool_stats.get("maxwait_us").unwrap_or(&0) / 1_000_000).to_string(),
"maxwait_us" => {
(pool_stats.get("maxwait_us").unwrap_or(&0) % 1_000_000).to_string()
}
_other_values => pool_stats.get(column.0).unwrap_or(&0).to_string(),
};
row.push(value);
}
res.put(data_row(&row));
}
res.put(row_description(&PoolStats::generate_header()));
pool_lookup.iter().for_each(|(_identifier, pool_stats)| {
res.put(data_row(&pool_stats.generate_row()));
});
res.put(command_complete("SHOW"));
// ReadyForQuery
@@ -320,17 +308,17 @@ where
let paused = pool.paused();
res.put(data_row(&vec![
address.name(), // name
address.host.to_string(), // host
address.port.to_string(), // port
database_name.to_string(), // database
pool_config.user.username.to_string(), // force_user
pool_config.user.pool_size.to_string(), // pool_size
"0".to_string(), // min_pool_size
"0".to_string(), // reserve_pool
pool_config.pool_mode.to_string(), // pool_mode
pool_config.user.pool_size.to_string(), // max_connections
pool_state.connections.to_string(), // current_connections
address.name(), // name
address.host.to_string(), // host
address.port.to_string(), // port
database_name.to_string(), // database
pool_config.user.username.to_string(), // force_user
pool_config.user.pool_size.to_string(), // pool_size
pool_config.user.min_pool_size.unwrap_or(0).to_string(), // min_pool_size
"0".to_string(), // reserve_pool
pool_config.pool_mode.to_string(), // pool_mode
pool_config.user.pool_size.to_string(), // max_connections
pool_state.connections.to_string(), // current_connections
match paused {
// paused
true => "1".to_string(),
@@ -400,7 +388,7 @@ where
for (id, pool) in get_all_pools().iter() {
for address in pool.get_addresses_from_host(host) {
if !pool.is_banned(&address) {
pool.ban(&address, BanReason::AdminBan(duration_seconds), -1);
pool.ban(&address, BanReason::AdminBan(duration_seconds), None);
res.put(data_row(&vec![
id.db.clone(),
id.user.clone(),
@@ -617,7 +605,6 @@ where
("avg_wait_time", DataType::Numeric),
];
let all_stats = get_address_stats();
let mut res = BytesMut::new();
res.put(row_description(&columns));
@@ -625,15 +612,10 @@ where
for shard in 0..pool.shards() {
for server in 0..pool.servers(shard) {
let address = pool.address(shard, server);
let stats = match all_stats.get(&address.id) {
Some(stats) => stats.clone(),
None => HashMap::new(),
};
let mut row = vec![address.name(), user_pool.db.clone(), user_pool.user.clone()];
for column in &columns[3..] {
row.push(stats.get(column.0).unwrap_or(&0).to_string());
}
let stats = address.stats.clone();
stats.populate_row(&mut row);
res.put(data_row(&row));
}
@@ -673,16 +655,16 @@ where
for (_, client) in new_map {
let row = vec![
format!("{:#010X}", client.client_id),
client.pool_name,
client.username,
client.application_name.clone(),
client.state.to_string(),
client.transaction_count.to_string(),
client.query_count.to_string(),
client.error_count.to_string(),
format!("{:#010X}", client.client_id()),
client.pool_name(),
client.username(),
client.application_name(),
client.state.load(Ordering::Relaxed).to_string(),
client.transaction_count.load(Ordering::Relaxed).to_string(),
client.query_count.load(Ordering::Relaxed).to_string(),
client.error_count.load(Ordering::Relaxed).to_string(),
Instant::now()
.duration_since(client.connect_time)
.duration_since(client.connect_time())
.as_secs()
.to_string(),
];
@@ -717,6 +699,8 @@ where
("bytes_sent", DataType::Numeric),
("bytes_received", DataType::Numeric),
("age_seconds", DataType::Numeric),
("prepare_cache_hit", DataType::Numeric),
("prepare_cache_miss", DataType::Numeric),
];
let new_map = get_server_stats();
@@ -724,21 +708,30 @@ where
res.put(row_description(&columns));
for (_, server) in new_map {
let application_name = server.application_name.read();
let row = vec![
format!("{:#010X}", server.server_id),
server.pool_name,
server.username,
server.address_name,
server.application_name,
server.state.to_string(),
server.transaction_count.to_string(),
server.query_count.to_string(),
server.bytes_sent.to_string(),
server.bytes_received.to_string(),
format!("{:#010X}", server.server_id()),
server.pool_name(),
server.username(),
server.address_name(),
application_name.clone(),
server.state.load(Ordering::Relaxed).to_string(),
server.transaction_count.load(Ordering::Relaxed).to_string(),
server.query_count.load(Ordering::Relaxed).to_string(),
server.bytes_sent.load(Ordering::Relaxed).to_string(),
server.bytes_received.load(Ordering::Relaxed).to_string(),
Instant::now()
.duration_since(server.connect_time)
.duration_since(server.connect_time())
.as_secs()
.to_string(),
server
.prepared_hit_count
.load(Ordering::Relaxed)
.to_string(),
server
.prepared_miss_count
.load(Ordering::Relaxed)
.to_string(),
];
res.put(data_row(&row));

134
src/auth_passthrough.rs Normal file
View File

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

File diff suppressed because it is too large Load Diff

View File

@@ -12,9 +12,11 @@ use std::sync::Arc;
use tokio::fs::File;
use tokio::io::AsyncReadExt;
use crate::dns_cache::CachedResolver;
use crate::errors::Error;
use crate::pool::{ClientServerMap, ConnectionPool};
use crate::sharding::ShardingFunction;
use crate::stats::AddressStats;
use crate::tls::{load_certs, load_keys};
pub const VERSION: &str = env!("CARGO_PKG_VERSION");
@@ -62,7 +64,7 @@ impl PartialEq<Role> for Option<Role> {
}
/// Address identifying a PostgreSQL server uniquely.
#[derive(Clone, PartialEq, Hash, std::cmp::Eq, Debug)]
#[derive(Clone, Debug)]
pub struct Address {
/// Unique ID per addressable Postgres server.
pub id: usize,
@@ -96,6 +98,9 @@ pub struct Address {
/// List of addresses to receive mirrored traffic.
pub mirrors: Vec<Address>,
/// Address stats
pub stats: Arc<AddressStats>,
}
impl Default for Address {
@@ -112,10 +117,56 @@ impl Default for Address {
username: String::from("username"),
pool_name: String::from("pool_name"),
mirrors: Vec::new(),
stats: Arc::new(AddressStats::default()),
}
}
}
impl std::fmt::Display for Address {
fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
write!(
f,
"[address: {}:{}][database: {}][user: {}]",
self.host, self.port, self.database, self.username
)
}
}
// We need to implement PartialEq by ourselves so we skip stats in the comparison
impl PartialEq for Address {
fn eq(&self, other: &Self) -> bool {
self.id == other.id
&& self.host == other.host
&& self.port == other.port
&& self.shard == other.shard
&& self.address_index == other.address_index
&& self.replica_number == other.replica_number
&& self.database == other.database
&& self.role == other.role
&& self.username == other.username
&& self.pool_name == other.pool_name
&& self.mirrors == other.mirrors
}
}
impl Eq for Address {}
// We need to implement Hash by ourselves so we skip stats in the comparison
impl Hash for Address {
fn hash<H: Hasher>(&self, state: &mut H) {
self.id.hash(state);
self.host.hash(state);
self.port.hash(state);
self.shard.hash(state);
self.address_index.hash(state);
self.replica_number.hash(state);
self.database.hash(state);
self.role.hash(state);
self.username.hash(state);
self.pool_name.hash(state);
self.mirrors.hash(state);
}
}
impl Address {
/// Address name (aka database) used in `SHOW STATS`, `SHOW DATABASES`, and `SHOW POOLS`.
pub fn name(&self) -> String {
@@ -137,8 +188,13 @@ impl Address {
#[derive(Clone, PartialEq, Hash, Eq, Serialize, Deserialize, Debug)]
pub struct User {
pub username: String,
pub password: String,
pub password: Option<String>,
pub server_username: Option<String>,
pub server_password: Option<String>,
pub pool_size: u32,
pub min_pool_size: Option<u32>,
pub pool_mode: Option<PoolMode>,
pub server_lifetime: Option<u64>,
#[serde(default)] // 0
pub statement_timeout: u64,
}
@@ -147,13 +203,38 @@ impl Default for User {
fn default() -> User {
User {
username: String::from("postgres"),
password: String::new(),
password: None,
server_username: None,
server_password: None,
pool_size: 15,
min_pool_size: None,
statement_timeout: 0,
pool_mode: None,
server_lifetime: None,
}
}
}
impl User {
fn validate(&self) -> Result<(), Error> {
match self.min_pool_size {
Some(min_pool_size) => {
if min_pool_size > self.pool_size {
error!(
"min_pool_size of {} cannot be larger than pool_size of {}",
min_pool_size, self.pool_size
);
return Err(Error::BadConfig);
}
}
None => (),
};
Ok(())
}
}
/// General configuration.
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq)]
pub struct General {
@@ -161,9 +242,11 @@ pub struct General {
pub host: String,
#[serde(default = "General::default_port")]
pub port: i16,
pub port: u16,
pub enable_prometheus_exporter: Option<bool>,
#[serde(default = "General::default_prometheus_exporter_port")]
pub prometheus_exporter_port: i16,
#[serde(default = "General::default_connect_timeout")]
@@ -185,6 +268,12 @@ pub struct General {
#[serde(default)] // False
pub log_client_disconnections: bool,
#[serde(default)] // False
pub dns_cache_enabled: bool,
#[serde(default = "General::default_dns_max_ttl")]
pub dns_max_ttl: u64,
#[serde(default = "General::default_shutdown_timeout")]
pub shutdown_timeout: u64,
@@ -200,16 +289,40 @@ pub struct General {
#[serde(default = "General::default_idle_client_in_transaction_timeout")]
pub idle_client_in_transaction_timeout: u64,
#[serde(default = "General::default_server_lifetime")]
pub server_lifetime: u64,
#[serde(default = "General::default_server_round_robin")] // False
pub server_round_robin: bool,
#[serde(default = "General::default_worker_threads")]
pub worker_threads: usize,
#[serde(default)] // False
pub autoreload: bool,
#[serde(default)] // None
pub autoreload: Option<u64>,
pub tls_certificate: Option<String>,
pub tls_private_key: Option<String>,
#[serde(default)] // false
pub server_tls: bool,
#[serde(default)] // false
pub verify_server_certificate: bool,
pub admin_username: String,
pub admin_password: String,
#[serde(default = "General::default_validate_config")]
pub validate_config: bool,
// Support for auth query
pub auth_query: Option<String>,
pub auth_query_user: Option<String>,
pub auth_query_password: Option<String>,
#[serde(default)]
pub prepared_statements: bool,
}
impl General {
@@ -217,17 +330,21 @@ impl General {
"0.0.0.0".into()
}
pub fn default_port() -> i16 {
pub fn default_port() -> u16 {
5432
}
pub fn default_server_lifetime() -> u64 {
1000 * 60 * 60 // 1 hour
}
pub fn default_connect_timeout() -> u64 {
1000
}
// These keepalive defaults should detect a dead connection within 30 seconds.
// Tokio defaults to disabling keepalives which keeps dead connections around indefinitely.
// This can lead to permenant server pool exhaustion
// This can lead to permanent server pool exhaustion
pub fn default_tcp_keepalives_idle() -> u64 {
5 // 5 seconds
}
@@ -241,13 +358,17 @@ impl General {
}
pub fn default_idle_timeout() -> u64 {
60000 // 10 minutes
600000 // 10 minutes
}
pub fn default_shutdown_timeout() -> u64 {
60000
}
pub fn default_dns_max_ttl() -> u64 {
30
}
pub fn default_healthcheck_timeout() -> u64 {
1000
}
@@ -267,6 +388,18 @@ impl General {
pub fn default_idle_client_in_transaction_timeout() -> u64 {
0
}
pub fn default_validate_config() -> bool {
true
}
pub fn default_prometheus_exporter_port() -> i16 {
9930
}
pub fn default_server_round_robin() -> bool {
true
}
}
impl Default for General {
@@ -289,11 +422,22 @@ impl Default for General {
tcp_keepalives_interval: Self::default_tcp_keepalives_interval(),
log_client_connections: false,
log_client_disconnections: false,
autoreload: false,
autoreload: None,
dns_cache_enabled: false,
dns_max_ttl: Self::default_dns_max_ttl(),
tls_certificate: None,
tls_private_key: None,
server_tls: false,
verify_server_certificate: false,
admin_username: String::from("admin"),
admin_password: String::from("admin"),
auth_query: None,
auth_query_user: None,
auth_query_password: None,
server_lifetime: Self::default_server_lifetime(),
server_round_robin: false,
validate_config: true,
prepared_statements: false,
}
}
}
@@ -309,6 +453,7 @@ pub enum PoolMode {
#[serde(alias = "session", alias = "Session")]
Session,
}
impl ToString for PoolMode {
fn to_string(&self) -> String {
match *self {
@@ -345,6 +490,7 @@ pub struct Pool {
#[serde(default = "Pool::default_load_balancing_mode")]
pub load_balancing_mode: LoadBalancingMode,
#[serde(default = "Pool::default_default_role")]
pub default_role: String,
#[serde(default)] // False
@@ -353,10 +499,18 @@ pub struct Pool {
#[serde(default)] // False
pub primary_reads_enabled: bool,
/// Maximum time to allow for establishing a new server connection.
pub connect_timeout: Option<u64>,
/// Close idle connections that have been opened for longer than this.
pub idle_timeout: 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.
pub server_lifetime: Option<u64>,
#[serde(default = "Pool::default_sharding_function")]
pub sharding_function: ShardingFunction,
#[serde(default = "Pool::default_automatic_sharding_key")]
@@ -366,9 +520,17 @@ pub struct Pool {
pub shard_id_regex: Option<String>,
pub regex_search_limit: Option<usize>,
pub auth_query: Option<String>,
pub auth_query_user: Option<String>,
pub auth_query_password: Option<String>,
#[serde(default = "Pool::default_cleanup_server_connections")]
pub cleanup_server_connections: bool,
pub plugins: Option<Plugins>,
pub shards: BTreeMap<String, Shard>,
pub users: BTreeMap<String, User>,
// Note, don't put simple fields below these configs. There's a compatability issue with TOML that makes it
// Note, don't put simple fields below these configs. There's a compatibility issue with TOML that makes it
// incompatible to have simple fields in TOML after complex objects. See
// https://users.rust-lang.org/t/why-toml-to-string-get-error-valueaftertable/85903
}
@@ -380,6 +542,12 @@ impl Pool {
s.finish()
}
pub fn is_auth_query_configured(&self) -> bool {
self.auth_query_password.is_some()
&& self.auth_query_user.is_some()
&& self.auth_query_password.is_some()
}
pub fn default_pool_mode() -> PoolMode {
PoolMode::Transaction
}
@@ -392,6 +560,18 @@ impl Pool {
None
}
pub fn default_default_role() -> String {
"any".into()
}
pub fn default_sharding_function() -> ShardingFunction {
ShardingFunction::PgBigintHash
}
pub fn default_cleanup_server_connections() -> bool {
true
}
pub fn validate(&mut self) -> Result<(), Error> {
match self.default_role.as_ref() {
"any" => (),
@@ -451,6 +631,10 @@ impl Pool {
None => None,
};
for (_, user) in &self.users {
user.validate()?;
}
Ok(())
}
}
@@ -472,6 +656,12 @@ impl Default for Pool {
sharding_key_regex: None,
shard_id_regex: None,
regex_search_limit: Some(1000),
auth_query: None,
auth_query_user: None,
auth_query_password: None,
server_lifetime: None,
plugins: None,
cleanup_server_connections: true,
}
}
}
@@ -521,7 +711,7 @@ impl Shard {
if primary_count > 1 {
error!(
"Shard {} has more than on primary configured",
"Shard {} has more than one primary configured",
self.database
);
return Err(Error::BadConfig);
@@ -550,6 +740,76 @@ impl Default for Shard {
}
}
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Default, Hash, Eq)]
pub struct Plugins {
pub intercept: Option<Intercept>,
pub table_access: Option<TableAccess>,
pub query_logger: Option<QueryLogger>,
pub prewarmer: Option<Prewarmer>,
}
impl std::fmt::Display for Plugins {
fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
write!(
f,
"interceptor: {}, table_access: {}, query_logger: {}, prewarmer: {}",
self.intercept.is_some(),
self.table_access.is_some(),
self.query_logger.is_some(),
self.prewarmer.is_some(),
)
}
}
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Default, Hash, Eq)]
pub struct Intercept {
pub enabled: bool,
pub queries: BTreeMap<String, Query>,
}
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Default, Hash, Eq)]
pub struct TableAccess {
pub enabled: bool,
pub tables: Vec<String>,
}
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Default, Hash, Eq)]
pub struct QueryLogger {
pub enabled: bool,
}
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Default, Hash, Eq)]
pub struct Prewarmer {
pub enabled: bool,
pub queries: Vec<String>,
}
impl Intercept {
pub fn substitute(&mut self, db: &str, user: &str) {
for (_, query) in self.queries.iter_mut() {
query.substitute(db, user);
query.query = query.query.to_ascii_lowercase();
}
}
}
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Default, Hash, Eq)]
pub struct Query {
pub query: String,
pub schema: Vec<Vec<String>>,
pub result: Vec<Vec<String>>,
}
impl Query {
pub fn substitute(&mut self, db: &str, user: &str) {
for col in self.result.iter_mut() {
for i in 0..col.len() {
col[i] = col[i].replace("${USER}", user).replace("${DATABASE}", db);
}
}
}
}
/// Configuration wrapper.
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq)]
pub struct Config {
@@ -567,14 +827,42 @@ pub struct Config {
#[serde(default = "Config::default_path")]
pub path: String,
// General and global settings.
pub general: General,
// Plugins that should run in all pools.
pub plugins: Option<Plugins>,
// Connection pools.
pub pools: HashMap<String, Pool>,
}
impl Config {
pub fn is_auth_query_configured(&self) -> bool {
self.pools
.iter()
.any(|(_name, pool)| pool.is_auth_query_configured())
}
pub fn default_path() -> String {
String::from("pgcat.toml")
}
pub fn fill_up_auth_query_config(&mut self) {
for (_name, pool) in self.pools.iter_mut() {
if pool.auth_query.is_none() {
pool.auth_query = self.general.auth_query.clone();
}
if pool.auth_query_user.is_none() {
pool.auth_query_user = self.general.auth_query_user.clone();
}
if pool.auth_query_password.is_none() {
pool.auth_query_password = self.general.auth_query_password.clone();
}
}
}
}
impl Default for Config {
@@ -583,6 +871,7 @@ impl Default for Config {
path: Self::default_path(),
general: General::default(),
pools: HashMap::default(),
plugins: None,
}
}
}
@@ -702,6 +991,11 @@ impl Config {
);
info!("Shutdown timeout: {}ms", self.general.shutdown_timeout);
info!("Healthcheck delay: {}ms", self.general.healthcheck_delay);
info!(
"Default max server lifetime: {}ms",
self.general.server_lifetime
);
info!("Sever round robin: {}", self.general.server_round_robin);
match self.general.tls_certificate.clone() {
Some(tls_certificate) => {
info!("TLS certificate: {}", tls_certificate);
@@ -720,6 +1014,19 @@ impl Config {
info!("TLS support is disabled");
}
};
info!("Server TLS enabled: {}", self.general.server_tls);
info!(
"Server TLS certificate verification: {}",
self.general.verify_server_certificate
);
info!("Prepared statements: {}", self.general.prepared_statements);
info!(
"Plugins: {}",
match self.plugins {
Some(ref plugins) => plugins.to_string(),
None => "not configured".into(),
}
);
for (pool_name, pool_config) in &self.pools {
// TODO: Make this output prettier (maybe a table?)
@@ -734,8 +1041,9 @@ impl Config {
.to_string()
);
info!(
"[pool: {}] Pool mode: {:?}",
pool_name, pool_config.pool_mode
"[pool: {}] Default pool mode: {}",
pool_name,
pool_config.pool_mode.to_string()
);
info!(
"[pool: {}] Load Balancing mode: {:?}",
@@ -777,21 +1085,113 @@ impl Config {
pool_name,
pool_config.users.len()
);
info!(
"[pool: {}] Max server lifetime: {}",
pool_name,
match pool_config.server_lifetime {
Some(server_lifetime) => format!("{}ms", server_lifetime),
None => "default".to_string(),
}
);
info!(
"[pool: {}] Cleanup server connections: {}",
pool_name, pool_config.cleanup_server_connections
);
info!(
"[pool: {}] Plugins: {}",
pool_name,
match pool_config.plugins {
Some(ref plugins) => plugins.to_string(),
None => "not configured".into(),
}
);
for user in &pool_config.users {
info!(
"[pool: {}][user: {}] Pool size: {}",
pool_name, user.1.username, user.1.pool_size,
);
info!(
"[pool: {}][user: {}] Minimum pool size: {}",
pool_name,
user.1.username,
user.1.min_pool_size.unwrap_or(0)
);
info!(
"[pool: {}][user: {}] Statement timeout: {}",
pool_name, user.1.username, user.1.statement_timeout
)
);
info!(
"[pool: {}][user: {}] Pool mode: {}",
pool_name,
user.1.username,
match user.1.pool_mode {
Some(pool_mode) => pool_mode.to_string(),
None => pool_config.pool_mode.to_string(),
}
);
info!(
"[pool: {}][user: {}] Max server lifetime: {}",
pool_name,
user.1.username,
match user.1.server_lifetime {
Some(server_lifetime) => format!("{}ms", server_lifetime),
None => "default".to_string(),
}
);
}
}
}
pub fn validate(&mut self) -> Result<(), Error> {
// Validation for auth_query feature
if self.general.auth_query.is_some()
&& (self.general.auth_query_user.is_none()
|| self.general.auth_query_password.is_none())
{
error!(
"If auth_query is specified, \
you need to provide a value \
for `auth_query_user`, \
`auth_query_password`"
);
return Err(Error::BadConfig);
}
for (name, pool) in self.pools.iter() {
if pool.auth_query.is_some()
&& (pool.auth_query_user.is_none() || pool.auth_query_password.is_none())
{
error!(
"Error in pool {{ {} }}. \
If auth_query is specified, you need \
to provide a value for `auth_query_user`, \
`auth_query_password`",
name
);
return Err(Error::BadConfig);
}
for (_name, user_data) in pool.users.iter() {
if (pool.auth_query.is_none()
|| pool.auth_query_password.is_none()
|| pool.auth_query_user.is_none())
&& user_data.password.is_none()
{
error!(
"Error in pool {{ {} }}. \
You have to specify a user password \
for every pool if auth_query is not specified",
name
);
return Err(Error::BadConfig);
}
}
}
// Validate TLS!
match self.general.tls_certificate.clone() {
Some(tls_certificate) => {
@@ -844,6 +1244,10 @@ pub fn get_idle_client_in_transaction_timeout() -> u64 {
.idle_client_in_transaction_timeout
}
pub fn get_prepared_statements() -> bool {
(*(*CONFIG.load())).general.prepared_statements
}
/// Parse the configuration file located at the path.
pub async fn parse(path: &str) -> Result<(), Error> {
let mut contents = String::new();
@@ -871,6 +1275,7 @@ pub async fn parse(path: &str) -> Result<(), Error> {
}
};
config.fill_up_auth_query_config();
config.validate()?;
config.path = path.to_string();
@@ -883,6 +1288,7 @@ pub async fn parse(path: &str) -> Result<(), Error> {
pub async fn reload_config(client_server_map: ClientServerMap) -> Result<bool, Error> {
let old_config = get_config();
match parse(&old_config.path).await {
Ok(()) => (),
Err(err) => {
@@ -890,14 +1296,18 @@ pub async fn reload_config(client_server_map: ClientServerMap) -> Result<bool, E
return Err(Error::BadConfig);
}
};
let new_config = get_config();
if old_config.pools != new_config.pools {
info!("Pool configuration changed");
match CachedResolver::from_config().await {
Ok(_) => (),
Err(err) => error!("DNS cache reinitialization error: {:?}", err),
};
if old_config != new_config {
info!("Config changed, reloading");
ConnectionPool::from_config(client_server_map).await?;
Ok(true)
} else if old_config != new_config {
Ok(true)
} else {
Ok(false)
}
@@ -940,7 +1350,10 @@ mod test {
"sharding_user"
);
assert_eq!(
get_config().pools["sharded_db"].users["1"].password,
get_config().pools["sharded_db"].users["1"]
.password
.as_ref()
.unwrap(),
"other_user"
);
assert_eq!(get_config().pools["sharded_db"].users["1"].pool_size, 21);
@@ -965,10 +1378,16 @@ mod test {
"simple_user"
);
assert_eq!(
get_config().pools["simple_db"].users["0"].password,
get_config().pools["simple_db"].users["0"]
.password
.as_ref()
.unwrap(),
"simple_user"
);
assert_eq!(get_config().pools["simple_db"].users["0"].pool_size, 5);
assert_eq!(get_config().general.auth_query, None);
assert_eq!(get_config().general.auth_query_user, None);
assert_eq!(get_config().general.auth_query_password, None);
}
#[tokio::test]

410
src/dns_cache.rs Normal file
View File

@@ -0,0 +1,410 @@
use crate::config::get_config;
use crate::errors::Error;
use arc_swap::ArcSwap;
use log::{debug, error, info, warn};
use once_cell::sync::Lazy;
use std::collections::{HashMap, HashSet};
use std::io;
use std::net::IpAddr;
use std::sync::Arc;
use std::sync::RwLock;
use tokio::time::{sleep, Duration};
use trust_dns_resolver::error::{ResolveError, ResolveResult};
use trust_dns_resolver::lookup_ip::LookupIp;
use trust_dns_resolver::TokioAsyncResolver;
/// Cached Resolver Globally available
pub static CACHED_RESOLVER: Lazy<ArcSwap<CachedResolver>> =
Lazy::new(|| ArcSwap::from_pointee(CachedResolver::default()));
// Ip addressed are returned as a set of addresses
// so we can compare.
#[derive(Clone, PartialEq, Debug)]
pub struct AddrSet {
set: HashSet<IpAddr>,
}
impl AddrSet {
fn new() -> AddrSet {
AddrSet {
set: HashSet::new(),
}
}
}
impl From<LookupIp> for AddrSet {
fn from(lookup_ip: LookupIp) -> Self {
let mut addr_set = AddrSet::new();
for address in lookup_ip.iter() {
addr_set.set.insert(address);
}
addr_set
}
}
///
/// A CachedResolver is a DNS resolution cache mechanism with customizable expiration time.
///
/// The system works as follows:
///
/// When a host is to be resolved, if we have not resolved it before, a new resolution is
/// executed and stored in the internal cache. Concurrently, every `dns_max_ttl` time, the
/// cache is refreshed.
///
/// # Example:
///
/// ```
/// use pgcat::dns_cache::{CachedResolverConfig, CachedResolver};
///
/// # tokio_test::block_on(async {
/// let config = CachedResolverConfig::default();
/// let resolver = CachedResolver::new(config, None).await.unwrap();
/// let addrset = resolver.lookup_ip("www.example.com.").await.unwrap();
/// # })
/// ```
///
/// // Now the ip resolution is stored in local cache and subsequent
/// // calls will be returned from cache. Also, the cache is refreshed
/// // and updated every 10 seconds.
///
/// // You can now check if an 'old' lookup differs from what it's currently
/// // store in cache by using `has_changed`.
/// resolver.has_changed("www.example.com.", addrset)
#[derive(Default)]
pub struct CachedResolver {
// The configuration of the cached_resolver.
config: CachedResolverConfig,
// This is the hash that contains the hash.
data: Option<RwLock<HashMap<String, AddrSet>>>,
// The resolver to be used for DNS queries.
resolver: Option<TokioAsyncResolver>,
// The RefreshLoop
refresh_loop: RwLock<Option<tokio::task::JoinHandle<()>>>,
}
///
/// Configuration
#[derive(Clone, Debug, Default, PartialEq)]
pub struct CachedResolverConfig {
/// Amount of time in secods that a resolved dns address is considered stale.
dns_max_ttl: u64,
/// Enabled or disabled? (this is so we can reload config)
enabled: bool,
}
impl CachedResolverConfig {
fn new(dns_max_ttl: u64, enabled: bool) -> Self {
CachedResolverConfig {
dns_max_ttl,
enabled,
}
}
}
impl From<crate::config::Config> for CachedResolverConfig {
fn from(config: crate::config::Config) -> Self {
CachedResolverConfig::new(config.general.dns_max_ttl, config.general.dns_cache_enabled)
}
}
impl CachedResolver {
///
/// Returns a new Arc<CachedResolver> based on passed configuration.
/// It also starts the loop that will refresh cache entries.
///
/// # Arguments:
///
/// * `config` - The `CachedResolverConfig` to be used to create the resolver.
///
/// # Example:
///
/// ```
/// use pgcat::dns_cache::{CachedResolverConfig, CachedResolver};
///
/// # tokio_test::block_on(async {
/// let config = CachedResolverConfig::default();
/// let resolver = CachedResolver::new(config, None).await.unwrap();
/// # })
/// ```
///
pub async fn new(
config: CachedResolverConfig,
data: Option<HashMap<String, AddrSet>>,
) -> Result<Arc<Self>, io::Error> {
// Construct a new Resolver with default configuration options
let resolver = Some(TokioAsyncResolver::tokio_from_system_conf()?);
let data = if let Some(hash) = data {
Some(RwLock::new(hash))
} else {
Some(RwLock::new(HashMap::new()))
};
let instance = Arc::new(Self {
config,
resolver,
data,
refresh_loop: RwLock::new(None),
});
if instance.enabled() {
info!("Scheduling DNS refresh loop");
let refresh_loop = tokio::task::spawn({
let instance = instance.clone();
async move {
instance.refresh_dns_entries_loop().await;
}
});
*(instance.refresh_loop.write().unwrap()) = Some(refresh_loop);
}
Ok(instance)
}
pub fn enabled(&self) -> bool {
self.config.enabled
}
// Schedules the refresher
async fn refresh_dns_entries_loop(&self) {
let resolver = TokioAsyncResolver::tokio_from_system_conf().unwrap();
let interval = Duration::from_secs(self.config.dns_max_ttl);
loop {
debug!("Begin refreshing cached DNS addresses.");
// To minimize the time we hold the lock, we first create
// an array with keys.
let mut hostnames: Vec<String> = Vec::new();
{
if let Some(ref data) = self.data {
for hostname in data.read().unwrap().keys() {
hostnames.push(hostname.clone());
}
}
}
for hostname in hostnames.iter() {
let addrset = self
.fetch_from_cache(hostname.as_str())
.expect("Could not obtain expected address from cache, this should not happen");
match resolver.lookup_ip(hostname).await {
Ok(lookup_ip) => {
let new_addrset = AddrSet::from(lookup_ip);
debug!(
"Obtained address for host ({}) -> ({:?})",
hostname, new_addrset
);
if addrset != new_addrset {
debug!(
"Addr changed from {:?} to {:?} updating cache.",
addrset, new_addrset
);
self.store_in_cache(hostname, new_addrset);
}
}
Err(err) => {
error!(
"There was an error trying to resolv {}: ({}).",
hostname, err
);
}
}
}
debug!("Finished refreshing cached DNS addresses.");
sleep(interval).await;
}
}
/// Returns a `AddrSet` given the specified hostname.
///
/// This method first tries to fetch the value from the cache, if it misses
/// then it is resolved and stored in the cache. TTL from records is ignored.
///
/// # Arguments
///
/// * `host` - A string slice referencing the hostname to be resolved.
///
/// # Example:
///
/// ```
/// use pgcat::dns_cache::{CachedResolverConfig, CachedResolver};
///
/// # tokio_test::block_on(async {
/// let config = CachedResolverConfig::default();
/// let resolver = CachedResolver::new(config, None).await.unwrap();
/// let response = resolver.lookup_ip("www.google.com.");
/// # })
/// ```
///
pub async fn lookup_ip(&self, host: &str) -> ResolveResult<AddrSet> {
debug!("Lookup up {} in cache", host);
match self.fetch_from_cache(host) {
Some(addr_set) => {
debug!("Cache hit!");
Ok(addr_set)
}
None => {
debug!("Not found, executing a dns query!");
if let Some(ref resolver) = self.resolver {
let addr_set = AddrSet::from(resolver.lookup_ip(host).await?);
debug!("Obtained: {:?}", addr_set);
self.store_in_cache(host, addr_set.clone());
Ok(addr_set)
} else {
Err(ResolveError::from("No resolver available"))
}
}
}
}
//
// Returns true if the stored host resolution differs from the AddrSet passed.
pub fn has_changed(&self, host: &str, addr_set: &AddrSet) -> bool {
if let Some(fetched_addr_set) = self.fetch_from_cache(host) {
return fetched_addr_set != *addr_set;
}
false
}
// Fetches an AddrSet from the inner cache adquiring the read lock.
fn fetch_from_cache(&self, key: &str) -> Option<AddrSet> {
if let Some(ref hash) = self.data {
if let Some(addr_set) = hash.read().unwrap().get(key) {
return Some(addr_set.clone());
}
}
None
}
// Sets up the global CACHED_RESOLVER static variable so we can globally use DNS
// cache.
pub async fn from_config() -> Result<(), Error> {
let cached_resolver = CACHED_RESOLVER.load();
let desired_config = CachedResolverConfig::from(get_config());
if cached_resolver.config != desired_config {
if let Some(ref refresh_loop) = *(cached_resolver.refresh_loop.write().unwrap()) {
warn!("Killing Dnscache refresh loop as its configuration is being reloaded");
refresh_loop.abort()
}
let new_resolver = if let Some(ref data) = cached_resolver.data {
let data = Some(data.read().unwrap().clone());
CachedResolver::new(desired_config, data).await
} else {
CachedResolver::new(desired_config, None).await
};
match new_resolver {
Ok(ok) => {
CACHED_RESOLVER.store(ok);
Ok(())
}
Err(err) => {
let message = format!("Error setting up cached_resolver. Error: {:?}, will continue without this feature.", err);
Err(Error::DNSCachedError(message))
}
}
} else {
Ok(())
}
}
// Stores the AddrSet in cache adquiring the write lock.
fn store_in_cache(&self, host: &str, addr_set: AddrSet) {
if let Some(ref data) = self.data {
data.write().unwrap().insert(host.to_string(), addr_set);
} else {
error!("Could not insert, Hash not initialized");
}
}
}
#[cfg(test)]
mod tests {
use super::*;
use trust_dns_resolver::error::ResolveError;
#[tokio::test]
async fn new() {
let config = CachedResolverConfig {
dns_max_ttl: 10,
enabled: true,
};
let resolver = CachedResolver::new(config, None).await;
assert!(resolver.is_ok());
}
#[tokio::test]
async fn lookup_ip() {
let config = CachedResolverConfig {
dns_max_ttl: 10,
enabled: true,
};
let resolver = CachedResolver::new(config, None).await.unwrap();
let response = resolver.lookup_ip("www.google.com.").await;
assert!(response.is_ok());
}
#[tokio::test]
async fn has_changed() {
let config = CachedResolverConfig {
dns_max_ttl: 10,
enabled: true,
};
let resolver = CachedResolver::new(config, None).await.unwrap();
let hostname = "www.google.com.";
let response = resolver.lookup_ip(hostname).await;
let addr_set = response.unwrap();
assert!(!resolver.has_changed(hostname, &addr_set));
}
#[tokio::test]
async fn unknown_host() {
let config = CachedResolverConfig {
dns_max_ttl: 10,
enabled: true,
};
let resolver = CachedResolver::new(config, None).await.unwrap();
let hostname = "www.idontexists.";
let response = resolver.lookup_ip(hostname).await;
assert!(matches!(response, Err(ResolveError { .. })));
}
#[tokio::test]
async fn incorrect_address() {
let config = CachedResolverConfig {
dns_max_ttl: 10,
enabled: true,
};
let resolver = CachedResolver::new(config, None).await.unwrap();
let hostname = "w ww.idontexists.";
let response = resolver.lookup_ip(hostname).await;
assert!(matches!(response, Err(ResolveError { .. })));
assert!(!resolver.has_changed(hostname, &AddrSet::new()));
}
#[tokio::test]
// Ok, this test is based on the fact that google does DNS RR
// and does not responds with every available ip everytime, so
// if I cache here, it will miss after one cache iteration or two.
async fn thread() {
let config = CachedResolverConfig {
dns_max_ttl: 10,
enabled: true,
};
let resolver = CachedResolver::new(config, None).await.unwrap();
let hostname = "www.google.com.";
let response = resolver.lookup_ip(hostname).await;
let addr_set = response.unwrap();
assert!(!resolver.has_changed(hostname, &addr_set));
let resolver_for_refresher = resolver.clone();
let _thread_handle = tokio::task::spawn(async move {
resolver_for_refresher.refresh_dns_entries_loop().await;
});
assert!(!resolver.has_changed(hostname, &addr_set));
}
}

View File

@@ -1,18 +1,130 @@
/// Errors.
//! Errors.
/// Various errors.
#[derive(Debug, PartialEq)]
#[derive(Debug, PartialEq, Clone)]
pub enum Error {
SocketError(String),
ClientSocketError(String, ClientIdentifier),
ClientGeneralError(String, ClientIdentifier),
ClientAuthImpossible(String),
ClientAuthPassthroughError(String, ClientIdentifier),
ClientBadStartup,
ProtocolSyncError(String),
BadQuery(String),
ServerError,
ServerStartupError(String, ServerIdentifier),
ServerAuthError(String, ServerIdentifier),
BadConfig,
AllServersDown,
ClientError(String),
TlsError,
StatementTimeout,
DNSCachedError(String),
ShuttingDown,
ParseBytesError(String),
AuthError(String),
AuthPassthroughError(String),
UnsupportedStatement,
QueryRouterParserError(String),
QueryRouterError(String),
}
#[derive(Clone, PartialEq, Debug)]
pub struct ClientIdentifier {
pub application_name: String,
pub username: String,
pub pool_name: String,
}
impl ClientIdentifier {
pub fn new(application_name: &str, username: &str, pool_name: &str) -> ClientIdentifier {
ClientIdentifier {
application_name: application_name.into(),
username: username.into(),
pool_name: pool_name.into(),
}
}
}
impl std::fmt::Display for ClientIdentifier {
fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
write!(
f,
"{{ application_name: {}, username: {}, pool_name: {} }}",
self.application_name, self.username, self.pool_name
)
}
}
#[derive(Clone, PartialEq, Debug)]
pub struct ServerIdentifier {
pub username: String,
pub database: String,
}
impl ServerIdentifier {
pub fn new(username: &str, database: &str) -> ServerIdentifier {
ServerIdentifier {
username: username.into(),
database: database.into(),
}
}
}
impl std::fmt::Display for ServerIdentifier {
fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
write!(
f,
"{{ username: {}, database: {} }}",
self.username, self.database
)
}
}
impl std::fmt::Display for Error {
fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
match &self {
&Error::ClientSocketError(error, client_identifier) => write!(
f,
"Error reading {} from client {}",
error, client_identifier
),
&Error::ClientGeneralError(error, client_identifier) => {
write!(f, "{} {}", error, client_identifier)
}
&Error::ClientAuthImpossible(username) => write!(
f,
"Client auth not possible, \
no cleartext password set for username: {} \
in config and auth passthrough (query_auth) \
is not set up.",
username
),
&Error::ClientAuthPassthroughError(error, client_identifier) => write!(
f,
"No cleartext password set, \
and no auth passthrough could not \
obtain the hash from server for {}, \
the error was: {}",
client_identifier, error
),
&Error::ServerStartupError(error, server_identifier) => write!(
f,
"Error reading {} on server startup {}",
error, server_identifier,
),
&Error::ServerAuthError(error, server_identifier) => {
write!(f, "{} for {}", error, server_identifier,)
}
// The rest can use Debug.
err => write!(f, "{:?}", err),
}
}
}
impl From<std::ffi::NulError> for Error {
fn from(err: std::ffi::NulError) -> Self {
Error::QueryRouterError(err.to_string())
}
}

View File

@@ -1,10 +1,17 @@
pub mod admin;
pub mod auth_passthrough;
pub mod client;
pub mod config;
pub mod constants;
pub mod dns_cache;
pub mod errors;
pub mod messages;
pub mod mirrors;
pub mod multi_logger;
pub mod plugins;
pub mod pool;
pub mod prometheus;
pub mod query_router;
pub mod scram;
pub mod server;
pub mod sharding;

View File

@@ -36,6 +36,7 @@ extern crate sqlparser;
extern crate tokio;
extern crate tokio_rustls;
extern crate toml;
extern crate trust_dns_resolver;
#[cfg(not(target_env = "msvc"))]
use jemallocator::Jemalloc;
@@ -60,34 +61,19 @@ use std::str::FromStr;
use std::sync::Arc;
use tokio::sync::broadcast;
mod admin;
mod client;
mod config;
mod constants;
mod errors;
mod messages;
mod mirrors;
mod multi_logger;
mod pool;
mod prometheus;
mod query_router;
mod scram;
mod server;
mod sharding;
mod stats;
mod tls;
use crate::config::{get_config, reload_config, VERSION};
use crate::pool::{ClientServerMap, ConnectionPool};
use crate::prometheus::start_metric_server;
use crate::stats::{Collector, Reporter, REPORTER};
use pgcat::config::{get_config, reload_config, VERSION};
use pgcat::dns_cache;
use pgcat::messages::configure_socket;
use pgcat::pool::{ClientServerMap, ConnectionPool};
use pgcat::prometheus::start_metric_server;
use pgcat::stats::{Collector, Reporter, REPORTER};
fn main() -> Result<(), Box<dyn std::error::Error>> {
multi_logger::MultiLogger::init().unwrap();
pgcat::multi_logger::MultiLogger::init().unwrap();
info!("Welcome to PgCat! Meow. (Version {})", VERSION);
if !query_router::QueryRouter::setup() {
if !pgcat::query_router::QueryRouter::setup() {
error!("Could not setup query router");
std::process::exit(exitcode::CONFIG);
}
@@ -105,7 +91,7 @@ fn main() -> Result<(), Box<dyn std::error::Error>> {
let runtime = Builder::new_multi_thread().worker_threads(1).build()?;
runtime.block_on(async {
match config::parse(&config_file).await {
match pgcat::config::parse(&config_file).await {
Ok(_) => (),
Err(err) => {
error!("Config parse error: {:?}", err);
@@ -162,8 +148,13 @@ fn main() -> Result<(), Box<dyn std::error::Error>> {
let client_server_map: ClientServerMap = Arc::new(Mutex::new(HashMap::new()));
// Statistics reporting.
let (stats_tx, stats_rx) = mpsc::channel(500_000);
REPORTER.store(Arc::new(Reporter::new(stats_tx.clone())));
REPORTER.store(Arc::new(Reporter::default()));
// Starts (if enabled) dns cache before pools initialization
match dns_cache::CachedResolver::from_config().await {
Ok(_) => (),
Err(err) => error!("DNS cache initialization error: {:?}", err),
};
// Connection pool that allows to query all shards and replicas.
match ConnectionPool::from_config(client_server_map.clone()).await {
@@ -175,20 +166,23 @@ fn main() -> Result<(), Box<dyn std::error::Error>> {
};
tokio::task::spawn(async move {
let mut stats_collector = Collector::new(stats_rx, stats_tx.clone());
let mut stats_collector = Collector::default();
stats_collector.collect().await;
});
info!("Config autoreloader: {}", config.general.autoreload);
info!("Config autoreloader: {}", match config.general.autoreload {
Some(interval) => format!("{} ms", interval),
None => "disabled".into(),
});
let mut autoreload_interval = tokio::time::interval(tokio::time::Duration::from_millis(15_000));
let autoreload_client_server_map = client_server_map.clone();
if let Some(interval) = config.general.autoreload {
let mut autoreload_interval = tokio::time::interval(tokio::time::Duration::from_millis(interval));
let autoreload_client_server_map = client_server_map.clone();
tokio::task::spawn(async move {
loop {
autoreload_interval.tick().await;
if config.general.autoreload {
info!("Automatically reloading config");
tokio::task::spawn(async move {
loop {
autoreload_interval.tick().await;
debug!("Automatically reloading config");
if let Ok(changed) = reload_config(autoreload_client_server_map.clone()).await {
if changed {
@@ -196,8 +190,10 @@ fn main() -> Result<(), Box<dyn std::error::Error>> {
}
};
}
}
});
});
};
#[cfg(windows)]
let mut term_signal = win_signal::ctrl_close().unwrap();
@@ -282,18 +278,20 @@ fn main() -> Result<(), Box<dyn std::error::Error>> {
let drain_tx = drain_tx.clone();
let client_server_map = client_server_map.clone();
let tls_certificate = config.general.tls_certificate.clone();
let tls_certificate = get_config().general.tls_certificate.clone();
configure_socket(&socket);
tokio::task::spawn(async move {
let start = chrono::offset::Utc::now().naive_utc();
match client::client_entrypoint(
match pgcat::client::client_entrypoint(
socket,
client_server_map,
shutdown_rx,
drain_tx,
admin_only,
tls_certificate.clone(),
tls_certificate,
config.general.log_client_connections,
)
.await
@@ -301,7 +299,7 @@ fn main() -> Result<(), Box<dyn std::error::Error>> {
Ok(()) => {
let duration = chrono::offset::Utc::now().naive_utc() - start;
if config.general.log_client_disconnections {
if get_config().general.log_client_disconnections {
info!(
"Client {:?} disconnected, session duration: {}",
addr,
@@ -318,7 +316,7 @@ fn main() -> Result<(), Box<dyn std::error::Error>> {
Err(err) => {
match err {
errors::Error::ClientBadStartup => debug!("Client disconnected with error {:?}", err),
pgcat::errors::Error::ClientBadStartup => debug!("Client disconnected with error {:?}", err),
_ => warn!("Client disconnected with error {:?}", err),
}

View File

@@ -7,11 +7,15 @@ use socket2::{SockRef, TcpKeepalive};
use tokio::io::{AsyncReadExt, AsyncWriteExt};
use tokio::net::TcpStream;
use crate::client::PREPARED_STATEMENT_COUNTER;
use crate::config::get_config;
use crate::errors::Error;
use std::collections::HashMap;
use std::ffi::CString;
use std::io::{BufRead, Cursor};
use std::mem;
use std::sync::atomic::Ordering;
use std::time::Duration;
/// Postgres data type mappings
@@ -20,6 +24,10 @@ pub enum DataType {
Text,
Int4,
Numeric,
Bool,
Oid,
AnyArray,
Any,
}
impl From<&DataType> for i32 {
@@ -28,6 +36,10 @@ impl From<&DataType> for i32 {
DataType::Text => 25,
DataType::Int4 => 23,
DataType::Numeric => 1700,
DataType::Bool => 16,
DataType::Oid => 26,
DataType::AnyArray => 2277,
DataType::Any => 2276,
}
}
}
@@ -116,7 +128,10 @@ where
/// Send the startup packet the server. We're pretending we're a Pg client.
/// This tells the server which user we are and what database we want.
pub async fn startup(stream: &mut TcpStream, user: &str, database: &str) -> Result<(), Error> {
pub async fn startup<S>(stream: &mut S, user: &str, database: &str) -> Result<(), Error>
where
S: tokio::io::AsyncWrite + std::marker::Unpin,
{
let mut bytes = BytesMut::with_capacity(25);
bytes.put_i32(196608); // Protocol number
@@ -150,6 +165,21 @@ pub async fn startup(stream: &mut TcpStream, user: &str, database: &str) -> Resu
}
}
pub async fn ssl_request(stream: &mut TcpStream) -> Result<(), Error> {
let mut bytes = BytesMut::with_capacity(12);
bytes.put_i32(8);
bytes.put_i32(80877103);
match stream.write_all(&bytes).await {
Ok(_) => Ok(()),
Err(err) => Err(Error::SocketError(format!(
"Error writing SSLRequest to server socket - Error: {:?}",
err
))),
}
}
/// Parse the params the server sends as a key/value format.
pub fn parse_params(mut bytes: BytesMut) -> Result<HashMap<String, String>, Error> {
let mut result = HashMap::new();
@@ -213,7 +243,13 @@ pub fn md5_hash_password(user: &str, password: &str, salt: &[u8]) -> Vec<u8> {
let output = md5.finalize_reset();
// Second pass
md5.update(format!("{:x}", output));
md5_hash_second_pass(&(format!("{:x}", output)), salt)
}
pub fn md5_hash_second_pass(hash: &str, salt: &[u8]) -> Vec<u8> {
let mut md5 = Md5::new();
// Second pass
md5.update(hash);
md5.update(salt);
let mut password = format!("md5{:x}", md5.finalize())
@@ -247,6 +283,20 @@ where
write_all(stream, message).await
}
pub async fn md5_password_with_hash<S>(stream: &mut S, hash: &str, salt: &[u8]) -> Result<(), Error>
where
S: tokio::io::AsyncWrite + std::marker::Unpin,
{
let password = md5_hash_second_pass(hash, salt);
let mut message = BytesMut::with_capacity(password.len() as usize + 5);
message.put_u8(b'p');
message.put_i32(password.len() as i32 + 4);
message.put_slice(&password[..]);
write_all(stream, message).await
}
/// Implements a response to our custom `SET SHARDING KEY`
/// and `SET SERVER ROLE` commands.
/// This tells the client we're ready for the next query.
@@ -384,7 +434,7 @@ pub fn row_description(columns: &Vec<(&str, DataType)>) -> BytesMut {
let mut res = BytesMut::new();
let mut row_desc = BytesMut::new();
// how many colums we are storing
// how many columns we are storing
row_desc.put_i16(columns.len() as i16);
for (name, data_type) in columns {
@@ -405,6 +455,10 @@ pub fn row_description(columns: &Vec<(&str, DataType)>) -> BytesMut {
DataType::Text => -1,
DataType::Int4 => 4,
DataType::Numeric => -1,
DataType::Bool => 1,
DataType::Oid => 4,
DataType::AnyArray => -1,
DataType::Any => -1,
};
row_desc.put_i16(type_size);
@@ -443,6 +497,29 @@ pub fn data_row(row: &Vec<String>) -> BytesMut {
res
}
pub fn data_row_nullable(row: &Vec<Option<String>>) -> BytesMut {
let mut res = BytesMut::new();
let mut data_row = BytesMut::new();
data_row.put_i16(row.len() as i16);
for column in row {
if let Some(column) = column {
let column = column.as_bytes();
data_row.put_i32(column.len() as i32);
data_row.put_slice(column);
} else {
data_row.put_i32(-1 as i32);
}
}
res.put_u8(b'D');
res.put_i32(data_row.len() as i32 + 4);
res.put(data_row);
res
}
/// Create a CommandComplete message.
pub fn command_complete(command: &str) -> BytesMut {
let cmd = BytesMut::from(format!("{}\0", command).as_bytes());
@@ -453,6 +530,13 @@ pub fn command_complete(command: &str) -> BytesMut {
res
}
pub fn flush() -> BytesMut {
let mut bytes = BytesMut::new();
bytes.put_u8(b'H');
bytes.put_i32(4);
bytes
}
/// Write all data in the buffer to the TcpStream.
pub async fn write_all<S>(stream: &mut S, buf: BytesMut) -> Result<(), Error>
where
@@ -485,6 +569,29 @@ where
}
}
pub async fn write_all_flush<S>(stream: &mut S, buf: &[u8]) -> Result<(), Error>
where
S: tokio::io::AsyncWrite + std::marker::Unpin,
{
match stream.write_all(buf).await {
Ok(_) => match stream.flush().await {
Ok(_) => Ok(()),
Err(err) => {
return Err(Error::SocketError(format!(
"Error flushing socket - Error: {:?}",
err
)))
}
},
Err(err) => {
return Err(Error::SocketError(format!(
"Error writing to socket - Error: {:?}",
err
)))
}
}
}
/// Read a complete message from the socket.
pub async fn read_message<S>(stream: &mut S) -> Result<BytesMut, Error>
where
@@ -593,3 +700,285 @@ impl BytesMutReader for Cursor<&BytesMut> {
}
}
}
/// Parse (F) message.
/// See: <https://www.postgresql.org/docs/current/protocol-message-formats.html>
#[derive(Clone, Debug)]
pub struct Parse {
code: char,
#[allow(dead_code)]
len: i32,
pub name: String,
pub generated_name: String,
query: String,
num_params: i16,
param_types: Vec<i32>,
}
impl TryFrom<&BytesMut> for Parse {
type Error = Error;
fn try_from(buf: &BytesMut) -> Result<Parse, Error> {
let mut cursor = Cursor::new(buf);
let code = cursor.get_u8() as char;
let len = cursor.get_i32();
let name = cursor.read_string()?;
let query = cursor.read_string()?;
let num_params = cursor.get_i16();
let mut param_types = Vec::new();
for _ in 0..num_params {
param_types.push(cursor.get_i32());
}
Ok(Parse {
code,
len,
name,
generated_name: prepared_statement_name(),
query,
num_params,
param_types,
})
}
}
impl TryFrom<Parse> for BytesMut {
type Error = Error;
fn try_from(parse: Parse) -> Result<BytesMut, Error> {
let mut bytes = BytesMut::new();
let name_binding = CString::new(parse.name)?;
let name = name_binding.as_bytes_with_nul();
let query_binding = CString::new(parse.query)?;
let query = query_binding.as_bytes_with_nul();
// Recompute length of the message.
let len = 4 // self
+ name.len()
+ query.len()
+ 2
+ 4 * parse.num_params as usize;
bytes.put_u8(parse.code as u8);
bytes.put_i32(len as i32);
bytes.put_slice(name);
bytes.put_slice(query);
bytes.put_i16(parse.num_params);
for param in parse.param_types {
bytes.put_i32(param);
}
Ok(bytes)
}
}
impl TryFrom<&Parse> for BytesMut {
type Error = Error;
fn try_from(parse: &Parse) -> Result<BytesMut, Error> {
parse.clone().try_into()
}
}
impl Parse {
pub fn rename(mut self) -> Self {
self.name = self.generated_name.to_string();
self
}
pub fn anonymous(&self) -> bool {
self.name.is_empty()
}
}
/// Bind (B) message.
/// See: <https://www.postgresql.org/docs/current/protocol-message-formats.html>
#[derive(Clone, Debug)]
pub struct Bind {
code: char,
#[allow(dead_code)]
len: i64,
portal: String,
pub prepared_statement: String,
num_param_format_codes: i16,
param_format_codes: Vec<i16>,
num_param_values: i16,
param_values: Vec<(i32, BytesMut)>,
num_result_column_format_codes: i16,
result_columns_format_codes: Vec<i16>,
}
impl TryFrom<&BytesMut> for Bind {
type Error = Error;
fn try_from(buf: &BytesMut) -> Result<Bind, Error> {
let mut cursor = Cursor::new(buf);
let code = cursor.get_u8() as char;
let len = cursor.get_i32();
let portal = cursor.read_string()?;
let prepared_statement = cursor.read_string()?;
let num_param_format_codes = cursor.get_i16();
let mut param_format_codes = Vec::new();
for _ in 0..num_param_format_codes {
param_format_codes.push(cursor.get_i16());
}
let num_param_values = cursor.get_i16();
let mut param_values = Vec::new();
for _ in 0..num_param_values {
let param_len = cursor.get_i32();
let mut param = BytesMut::with_capacity(param_len as usize);
param.resize(param_len as usize, b'0');
cursor.copy_to_slice(&mut param);
param_values.push((param_len, param));
}
let num_result_column_format_codes = cursor.get_i16();
let mut result_columns_format_codes = Vec::new();
for _ in 0..num_result_column_format_codes {
result_columns_format_codes.push(cursor.get_i16());
}
Ok(Bind {
code,
len: len as i64,
portal,
prepared_statement,
num_param_format_codes,
param_format_codes,
num_param_values,
param_values,
num_result_column_format_codes,
result_columns_format_codes,
})
}
}
impl TryFrom<Bind> for BytesMut {
type Error = Error;
fn try_from(bind: Bind) -> Result<BytesMut, Error> {
let mut bytes = BytesMut::new();
let portal_binding = CString::new(bind.portal)?;
let portal = portal_binding.as_bytes_with_nul();
let prepared_statement_binding = CString::new(bind.prepared_statement)?;
let prepared_statement = prepared_statement_binding.as_bytes_with_nul();
let mut len = 4 // self
+ portal.len()
+ prepared_statement.len()
+ 2 // num_param_format_codes
+ 2 * bind.num_param_format_codes as usize // num_param_format_codes
+ 2; // num_param_values
for (param_len, _) in &bind.param_values {
len += 4 + *param_len as usize;
}
len += 2; // num_result_column_format_codes
len += 2 * bind.num_result_column_format_codes as usize;
bytes.put_u8(bind.code as u8);
bytes.put_i32(len as i32);
bytes.put_slice(portal);
bytes.put_slice(prepared_statement);
bytes.put_i16(bind.num_param_format_codes);
for param_format_code in bind.param_format_codes {
bytes.put_i16(param_format_code);
}
bytes.put_i16(bind.num_param_values);
for (param_len, param) in bind.param_values {
bytes.put_i32(param_len);
bytes.put_slice(&param);
}
bytes.put_i16(bind.num_result_column_format_codes);
for result_column_format_code in bind.result_columns_format_codes {
bytes.put_i16(result_column_format_code);
}
Ok(bytes)
}
}
impl Bind {
pub fn reassign(mut self, parse: &Parse) -> Self {
self.prepared_statement = parse.name.clone();
self
}
pub fn anonymous(&self) -> bool {
self.prepared_statement.is_empty()
}
}
#[derive(Debug, Clone)]
pub struct Describe {
code: char,
#[allow(dead_code)]
len: i32,
target: char,
pub statement_name: String,
}
impl TryFrom<&BytesMut> for Describe {
type Error = Error;
fn try_from(bytes: &BytesMut) -> Result<Describe, Error> {
let mut cursor = Cursor::new(bytes);
let code = cursor.get_u8() as char;
let len = cursor.get_i32();
let target = cursor.get_u8() as char;
let statement_name = cursor.read_string()?;
Ok(Describe {
code,
len,
target,
statement_name,
})
}
}
impl TryFrom<Describe> for BytesMut {
type Error = Error;
fn try_from(describe: Describe) -> Result<BytesMut, Error> {
let mut bytes = BytesMut::new();
let statement_name_binding = CString::new(describe.statement_name)?;
let statement_name = statement_name_binding.as_bytes_with_nul();
let len = 4 + 1 + statement_name.len();
bytes.put_u8(describe.code as u8);
bytes.put_i32(len as i32);
bytes.put_u8(describe.target as u8);
bytes.put_slice(statement_name);
Ok(bytes)
}
}
impl Describe {
pub fn rename(mut self, name: &str) -> Self {
self.statement_name = name.to_string();
self
}
pub fn anonymous(&self) -> bool {
self.statement_name.is_empty()
}
}
pub fn prepared_statement_name() -> String {
format!(
"P_{}",
PREPARED_STATEMENT_COUNTER.fetch_add(1, Ordering::SeqCst)
)
}

View File

@@ -1,11 +1,13 @@
use std::sync::Arc;
/// A mirrored PostgreSQL client.
/// Packets arrive to us through a channel from the main client and we send them to the server.
use bb8::Pool;
use bytes::{Bytes, BytesMut};
use parking_lot::RwLock;
use crate::config::{get_config, Address, Role, User};
use crate::pool::{ClientServerMap, ServerPool};
use crate::stats::get_reporter;
use log::{error, info, trace, warn};
use tokio::sync::mpsc::{channel, Receiver, Sender};
@@ -21,20 +23,24 @@ impl MirroredClient {
async fn create_pool(&self) -> Pool<ServerPool> {
let config = get_config();
let default = std::time::Duration::from_millis(10_000).as_millis() as u64;
let (connection_timeout, idle_timeout) = match config.pools.get(&self.address.pool_name) {
Some(cfg) => (
cfg.connect_timeout.unwrap_or(default),
cfg.idle_timeout.unwrap_or(default),
),
None => (default, default),
};
let (connection_timeout, idle_timeout, _cfg) =
match config.pools.get(&self.address.pool_name) {
Some(cfg) => (
cfg.connect_timeout.unwrap_or(default),
cfg.idle_timeout.unwrap_or(default),
cfg.clone(),
),
None => (default, default, crate::config::Pool::default()),
};
let manager = ServerPool::new(
self.address.clone(),
self.user.clone(),
self.database.as_str(),
ClientServerMap::default(),
get_reporter(),
Arc::new(RwLock::new(None)),
None,
true,
);
Pool::builder()

View File

@@ -17,7 +17,7 @@ use log::{Level, Log, Metadata, Record, SetLoggerError};
//
// So to summarize, if no `STDOUT_LOG` env var is present, the logger is the default logger. If `STDOUT_LOG` is set, everything
// but errors, that matches the log level set in the `STDOUT_LOG` env var is sent to stdout. You can have also some esoteric configuration
// where you set `RUST_LOG=debug` and `STDOUT_LOG=info`, in here, erros will go to stderr, warns and infos to stdout and debugs to stderr.
// where you set `RUST_LOG=debug` and `STDOUT_LOG=info`, in here, errors will go to stderr, warns and infos to stdout and debugs to stderr.
//
pub struct MultiLogger {
stderr_logger: env_logger::Logger,

120
src/plugins/intercept.rs Normal file
View File

@@ -0,0 +1,120 @@
//! The intercept plugin.
//!
//! It intercepts queries and returns fake results.
use async_trait::async_trait;
use bytes::{BufMut, BytesMut};
use serde::{Deserialize, Serialize};
use sqlparser::ast::Statement;
use log::debug;
use crate::{
config::Intercept as InterceptConfig,
errors::Error,
messages::{command_complete, data_row_nullable, row_description, DataType},
plugins::{Plugin, PluginOutput},
query_router::QueryRouter,
};
// TODO: use these structs for deserialization
#[derive(Serialize, Deserialize)]
pub struct Rule {
query: String,
schema: Vec<Column>,
result: Vec<Vec<String>>,
}
#[derive(Serialize, Deserialize)]
pub struct Column {
name: String,
data_type: String,
}
/// The intercept plugin.
pub struct Intercept<'a> {
pub enabled: bool,
pub config: &'a InterceptConfig,
}
#[async_trait]
impl<'a> Plugin for Intercept<'a> {
async fn run(
&mut self,
query_router: &QueryRouter,
ast: &Vec<Statement>,
) -> Result<PluginOutput, Error> {
if !self.enabled || ast.is_empty() {
return Ok(PluginOutput::Allow);
}
let mut config = self.config.clone();
config.substitute(
&query_router.pool_settings().db,
&query_router.pool_settings().user.username,
);
let mut result = BytesMut::new();
for q in ast {
// Normalization
let q = q.to_string().to_ascii_lowercase();
for (_, target) in config.queries.iter() {
if target.query.as_str() == q {
debug!("Intercepting query: {}", q);
let rd = target
.schema
.iter()
.map(|row| {
let name = &row[0];
let data_type = &row[1];
(
name.as_str(),
match data_type.as_str() {
"text" => DataType::Text,
"anyarray" => DataType::AnyArray,
"oid" => DataType::Oid,
"bool" => DataType::Bool,
"int4" => DataType::Int4,
_ => DataType::Any,
},
)
})
.collect::<Vec<(&str, DataType)>>();
result.put(row_description(&rd));
target.result.iter().for_each(|row| {
let row = row
.iter()
.map(|s| {
let s = s.as_str().to_string();
if s == "" {
None
} else {
Some(s)
}
})
.collect::<Vec<Option<String>>>();
result.put(data_row_nullable(&row));
});
result.put(command_complete("SELECT"));
}
}
}
if !result.is_empty() {
result.put_u8(b'Z');
result.put_i32(5);
result.put_u8(b'I');
return Ok(PluginOutput::Intercept(result));
} else {
Ok(PluginOutput::Allow)
}
}
}

44
src/plugins/mod.rs Normal file
View File

@@ -0,0 +1,44 @@
//! The plugin ecosystem.
//!
//! Currently plugins only grant access or deny access to the database for a particual query.
//! Example use cases:
//! - block known bad queries
//! - block access to system catalogs
//! - block dangerous modifications like `DROP TABLE`
//! - etc
//!
pub mod intercept;
pub mod prewarmer;
pub mod query_logger;
pub mod table_access;
use crate::{errors::Error, query_router::QueryRouter};
use async_trait::async_trait;
use bytes::BytesMut;
use sqlparser::ast::Statement;
pub use intercept::Intercept;
pub use query_logger::QueryLogger;
pub use table_access::TableAccess;
#[derive(Clone, Debug, PartialEq)]
pub enum PluginOutput {
Allow,
Deny(String),
Overwrite(Vec<Statement>),
Intercept(BytesMut),
}
#[async_trait]
pub trait Plugin {
// Run before the query is sent to the server.
async fn run(
&mut self,
query_router: &QueryRouter,
ast: &Vec<Statement>,
) -> Result<PluginOutput, Error>;
// TODO: run after the result is returned
// async fn callback(&mut self, query_router: &QueryRouter);
}

28
src/plugins/prewarmer.rs Normal file
View File

@@ -0,0 +1,28 @@
//! Prewarm new connections before giving them to the client.
use crate::{errors::Error, server::Server};
use log::info;
pub struct Prewarmer<'a> {
pub enabled: bool,
pub server: &'a mut Server,
pub queries: &'a Vec<String>,
}
impl<'a> Prewarmer<'a> {
pub async fn run(&mut self) -> Result<(), Error> {
if !self.enabled {
return Ok(());
}
for query in self.queries {
info!(
"{} Prewarning with query: `{}`",
self.server.address(),
query
);
self.server.query(&query).await?;
}
Ok(())
}
}

View File

@@ -0,0 +1,38 @@
//! Log all queries to stdout (or somewhere else, why not).
use crate::{
errors::Error,
plugins::{Plugin, PluginOutput},
query_router::QueryRouter,
};
use async_trait::async_trait;
use log::info;
use sqlparser::ast::Statement;
pub struct QueryLogger<'a> {
pub enabled: bool,
pub user: &'a str,
pub db: &'a str,
}
#[async_trait]
impl<'a> Plugin for QueryLogger<'a> {
async fn run(
&mut self,
_query_router: &QueryRouter,
ast: &Vec<Statement>,
) -> Result<PluginOutput, Error> {
if !self.enabled {
return Ok(PluginOutput::Allow);
}
let query = ast
.iter()
.map(|q| q.to_string())
.collect::<Vec<String>>()
.join("; ");
info!("[pool: {}][user: {}] {}", self.user, self.db, query);
Ok(PluginOutput::Allow)
}
}

View File

@@ -0,0 +1,59 @@
//! This query router plugin will check if the user can access a particular
//! table as part of their query. If they can't, the query will not be routed.
use async_trait::async_trait;
use sqlparser::ast::{visit_relations, Statement};
use crate::{
errors::Error,
plugins::{Plugin, PluginOutput},
query_router::QueryRouter,
};
use log::debug;
use core::ops::ControlFlow;
pub struct TableAccess<'a> {
pub enabled: bool,
pub tables: &'a Vec<String>,
}
#[async_trait]
impl<'a> Plugin for TableAccess<'a> {
async fn run(
&mut self,
_query_router: &QueryRouter,
ast: &Vec<Statement>,
) -> Result<PluginOutput, Error> {
if !self.enabled {
return Ok(PluginOutput::Allow);
}
let mut found = None;
visit_relations(ast, |relation| {
let relation = relation.to_string();
let parts = relation.split(".").collect::<Vec<&str>>();
let table_name = parts.last().unwrap();
if self.tables.contains(&table_name.to_string()) {
found = Some(table_name.to_string());
ControlFlow::<()>::Break(())
} else {
ControlFlow::<()>::Continue(())
}
});
if let Some(found) = found {
debug!("Blocking access to table \"{}\"", found);
Ok(PluginOutput::Deny(format!(
"permission for table \"{}\" denied",
found
)))
} else {
Ok(PluginOutput::Allow)
}
}
}

View File

@@ -1,6 +1,6 @@
use arc_swap::ArcSwap;
use async_trait::async_trait;
use bb8::{ManageConnection, Pool, PooledConnection};
use bb8::{ManageConnection, Pool, PooledConnection, QueueStrategy};
use bytes::{BufMut, BytesMut};
use chrono::naive::NaiveDateTime;
use log::{debug, error, info, warn};
@@ -10,6 +10,7 @@ use rand::seq::SliceRandom;
use rand::thread_rng;
use regex::Regex;
use std::collections::HashMap;
use std::fmt::{Display, Formatter};
use std::sync::{
atomic::{AtomicBool, Ordering},
Arc,
@@ -17,12 +18,16 @@ use std::sync::{
use std::time::Instant;
use tokio::sync::Notify;
use crate::config::{get_config, Address, General, LoadBalancingMode, PoolMode, Role, User};
use crate::config::{
get_config, Address, General, LoadBalancingMode, Plugins, PoolMode, Role, User,
};
use crate::errors::Error;
use crate::auth_passthrough::AuthPassthrough;
use crate::plugins::prewarmer;
use crate::server::Server;
use crate::sharding::ShardingFunction;
use crate::stats::{get_reporter, Reporter};
use crate::stats::{AddressStats, ClientStats, ServerStats};
pub type ProcessId = i32;
pub type SecretKey = i32;
@@ -51,7 +56,7 @@ pub enum BanReason {
/// An identifier for a PgCat pool,
/// a database visible to clients.
#[derive(Hash, Debug, Clone, PartialEq, Eq)]
#[derive(Hash, Debug, Clone, PartialEq, Eq, Default)]
pub struct PoolIdentifier {
// The name of the database clients want to connect to.
pub db: String,
@@ -60,6 +65,8 @@ pub struct PoolIdentifier {
pub user: String,
}
static POOL_REAPER_RATE: u64 = 30_000; // 30 seconds by default
impl PoolIdentifier {
/// Create a new user/pool identifier.
pub fn new(db: &str, user: &str) -> PoolIdentifier {
@@ -70,6 +77,12 @@ impl PoolIdentifier {
}
}
impl Display for PoolIdentifier {
fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
write!(f, "{}@{}", self.user, self.db)
}
}
impl From<&Address> for PoolIdentifier {
fn from(address: &Address) -> PoolIdentifier {
PoolIdentifier::new(&address.database, &address.username)
@@ -90,6 +103,7 @@ pub struct PoolSettings {
// Connecting user.
pub user: User,
pub db: String,
// Default server role to connect to.
pub default_role: Option<Role>,
@@ -123,6 +137,14 @@ pub struct PoolSettings {
// Limit how much of each query is searched for a potential shard regex match
pub regex_search_limit: usize,
// Auth query parameters
pub auth_query: Option<String>,
pub auth_query_user: Option<String>,
pub auth_query_password: Option<String>,
/// Plugins
pub plugins: Option<Plugins>,
}
impl Default for PoolSettings {
@@ -132,6 +154,7 @@ impl Default for PoolSettings {
load_balancing_mode: LoadBalancingMode::Random,
shards: 1,
user: User::default(),
db: String::default(),
default_role: None,
query_parser_enabled: false,
primary_reads_enabled: true,
@@ -143,6 +166,10 @@ impl Default for PoolSettings {
sharding_key_regex: None,
shard_id_regex: None,
regex_search_limit: 1000,
auth_query: None,
auth_query_user: None,
auth_query_password: None,
plugins: None,
}
}
}
@@ -161,10 +188,6 @@ pub struct ConnectionPool {
/// that should not be queried.
banlist: BanList,
/// The statistics aggregator runs in a separate task
/// and receives stats from clients, servers, and the pool.
stats: Reporter,
/// The server information (K messages) have to be passed to the
/// clients on startup. We pre-connect to all shards and replicas
/// on pool creation and save the K messages here.
@@ -185,6 +208,9 @@ pub struct ConnectionPool {
/// If the pool has been paused or not.
paused: Arc<AtomicBool>,
paused_waiter: Arc<Notify>,
/// AuthInfo
pub auth_hash: Arc<RwLock<Option<String>>>,
}
impl ConnectionPool {
@@ -201,6 +227,7 @@ impl ConnectionPool {
// There is one pool per database/user pair.
for user in pool_config.users.values() {
let old_pool_ref = get_pool(pool_name, &user.username);
let identifier = PoolIdentifier::new(pool_name, &user.username);
match old_pool_ref {
Some(pool) => {
@@ -211,10 +238,7 @@ impl ConnectionPool {
"[pool: {}][user: {}] has not changed",
pool_name, user.username
);
new_pools.insert(
PoolIdentifier::new(pool_name, &user.username),
pool.clone(),
);
new_pools.insert(identifier.clone(), pool.clone());
continue;
}
}
@@ -237,6 +261,7 @@ impl ConnectionPool {
// Sort by shard number to ensure consistency.
shard_ids.sort_by_key(|k| k.parse::<i64>().unwrap());
let pool_auth_hash: Arc<RwLock<Option<String>>> = Arc::new(RwLock::new(None));
for shard_idx in &shard_ids {
let shard = &pool_config.shards[shard_idx];
@@ -266,6 +291,7 @@ impl ConnectionPool {
username: user.username.clone(),
pool_name: pool_name.clone(),
mirrors: vec![],
stats: Arc::new(AddressStats::default()),
});
address_id += 1;
}
@@ -283,6 +309,7 @@ impl ConnectionPool {
username: user.username.clone(),
pool_name: pool_name.clone(),
mirrors: mirror_addresses,
stats: Arc::new(AddressStats::default()),
};
address_id += 1;
@@ -291,12 +318,52 @@ impl ConnectionPool {
replica_number += 1;
}
// We assume every server in the pool share user/passwords
let auth_passthrough = AuthPassthrough::from_pool_config(pool_config);
if let Some(apt) = &auth_passthrough {
match apt.fetch_hash(&address).await {
Ok(ok) => {
if let Some(ref pool_auth_hash_value) = *(pool_auth_hash.read())
{
if ok != *pool_auth_hash_value {
warn!(
"Hash is not the same across shards \
of the same pool, client auth will \
be done using last obtained hash. \
Server: {}:{}, Database: {}",
server.host, server.port, shard.database,
);
}
}
debug!("Hash obtained for {:?}", address);
{
let mut pool_auth_hash = pool_auth_hash.write();
*pool_auth_hash = Some(ok.clone());
}
}
Err(err) => warn!(
"Could not obtain password hashes \
using auth_query config, ignoring. \
Error: {:?}",
err,
),
}
}
let manager = ServerPool::new(
address.clone(),
user.clone(),
&shard.database,
client_server_map.clone(),
get_reporter(),
pool_auth_hash.clone(),
match pool_config.plugins {
Some(ref plugins) => Some(plugins.clone()),
None => config.plugins.clone(),
},
pool_config.cleanup_server_connections,
);
let connect_timeout = match pool_config.connect_timeout {
@@ -309,14 +376,44 @@ impl ConnectionPool {
None => config.general.idle_timeout,
};
let server_lifetime = match user.server_lifetime {
Some(server_lifetime) => server_lifetime,
None => match pool_config.server_lifetime {
Some(server_lifetime) => server_lifetime,
None => config.general.server_lifetime,
},
};
let reaper_rate = *vec![idle_timeout, server_lifetime, POOL_REAPER_RATE]
.iter()
.min()
.unwrap();
let queue_strategy = match config.general.server_round_robin {
true => QueueStrategy::Fifo,
false => QueueStrategy::Lifo,
};
debug!(
"[pool: {}][user: {}] Pool reaper rate: {}ms",
pool_name, user.username, reaper_rate
);
let pool = Pool::builder()
.max_size(user.pool_size)
.min_idle(user.min_pool_size)
.connection_timeout(std::time::Duration::from_millis(connect_timeout))
.idle_timeout(Some(std::time::Duration::from_millis(idle_timeout)))
.test_on_check_out(false)
.build(manager)
.await
.unwrap();
.max_lifetime(Some(std::time::Duration::from_millis(server_lifetime)))
.reaper_rate(std::time::Duration::from_millis(reaper_rate))
.queue_strategy(queue_strategy)
.test_on_check_out(false);
let pool = if config.general.validate_config {
pool.build(manager).await?
} else {
pool.build_unchecked(manager)
};
pools.push(pool);
servers.push(address);
@@ -328,20 +425,30 @@ impl ConnectionPool {
}
assert_eq!(shards.len(), addresses.len());
if let Some(ref _auth_hash) = *(pool_auth_hash.clone().read()) {
info!(
"Auth hash obtained from query_auth for pool {{ name: {}, user: {} }}",
pool_name, user.username
);
}
let pool = ConnectionPool {
databases: shards,
addresses,
banlist: Arc::new(RwLock::new(banlist)),
stats: get_reporter(),
config_hash: new_pool_hash_value,
server_info: Arc::new(RwLock::new(BytesMut::new())),
auth_hash: pool_auth_hash,
settings: PoolSettings {
pool_mode: pool_config.pool_mode,
pool_mode: match user.pool_mode {
Some(pool_mode) => pool_mode,
None => pool_config.pool_mode,
},
load_balancing_mode: pool_config.load_balancing_mode,
// shards: pool_config.shards.clone(),
shards: shard_ids.len(),
user: user.clone(),
db: pool_name.clone(),
default_role: match pool_config.default_role.as_str() {
"any" => None,
"replica" => Some(Role::Replica),
@@ -364,6 +471,13 @@ impl ConnectionPool {
.clone()
.map(|regex| Regex::new(regex.as_str()).unwrap()),
regex_search_limit: pool_config.regex_search_limit.unwrap_or(1000),
auth_query: pool_config.auth_query.clone(),
auth_query_user: pool_config.auth_query_user.clone(),
auth_query_password: pool_config.auth_query_password.clone(),
plugins: match pool_config.plugins {
Some(ref plugins) => Some(plugins.clone()),
None => config.plugins.clone(),
},
},
validated: Arc::new(AtomicBool::new(false)),
paused: Arc::new(AtomicBool::new(false)),
@@ -373,10 +487,12 @@ impl ConnectionPool {
// Connect to the servers to make sure pool configuration is valid
// before setting it globally.
// Do this async and somewhere else, we don't have to wait here.
let mut validate_pool = pool.clone();
tokio::task::spawn(async move {
let _ = validate_pool.validate().await;
});
if config.general.validate_config {
let mut validate_pool = pool.clone();
tokio::task::spawn(async move {
let _ = validate_pool.validate().await;
});
}
// There is one pool per database/user pair.
new_pools.insert(PoolIdentifier::new(pool_name, &user.username), pool);
@@ -387,7 +503,8 @@ impl ConnectionPool {
Ok(())
}
/// Connect to all shards and grab server information.
/// Connect to all shards, grab server information, and possibly
/// passwords to use in client auth.
/// Return server information we will pass to the clients
/// when they connect.
/// This also warms up the pool for clients that connect when
@@ -476,9 +593,9 @@ impl ConnectionPool {
/// Get a connection from the pool.
pub async fn get(
&self,
shard: usize, // shard number
role: Option<Role>, // primary or replica
client_process_id: i32, // client id
shard: usize, // shard number
role: Option<Role>, // primary or replica
client_stats: &ClientStats, // client id
) -> Result<(PooledConnection<'_, ServerPool>, Address), Error> {
let mut candidates: Vec<&Address> = self.addresses[shard]
.iter()
@@ -497,6 +614,10 @@ impl ConnectionPool {
});
}
// Indicate we're waiting on a server connection from a pool.
let now = Instant::now();
client_stats.waiting();
while !candidates.is_empty() {
// Get the next candidate
let address = match candidates.pop() {
@@ -515,10 +636,6 @@ impl ConnectionPool {
}
}
// Indicate we're waiting on a server connection from a pool.
let now = Instant::now();
self.stats.client_waiting(client_process_id);
// Check if we can connect
let mut conn = match self.databases[address.shard][address.address_index]
.get()
@@ -526,10 +643,14 @@ impl ConnectionPool {
{
Ok(conn) => conn,
Err(err) => {
error!("Banning instance {:?}, error: {:?}", address, err);
self.ban(address, BanReason::FailedCheckout, client_process_id);
self.stats
.client_checkout_error(client_process_id, address.id);
error!(
"Connection checkout error for instance {:?}, error: {:?}",
address, err
);
self.ban(address, BanReason::FailedCheckout, Some(client_stats));
address.stats.error();
client_stats.idle();
client_stats.checkout_error();
continue;
}
};
@@ -546,26 +667,33 @@ impl ConnectionPool {
// since we last checked the server is ok.
// Health checks are pretty expensive.
if !require_healthcheck {
self.stats.checkout_time(
now.elapsed().as_micros(),
client_process_id,
server.server_id(),
);
self.stats
.server_active(client_process_id, server.server_id());
let checkout_time: u64 = now.elapsed().as_micros() as u64;
client_stats.checkout_time(checkout_time);
server
.stats()
.checkout_time(checkout_time, client_stats.application_name());
server.stats().active(client_stats.application_name());
client_stats.active();
return Ok((conn, address.clone()));
}
if self
.run_health_check(address, server, now, client_process_id)
.run_health_check(address, server, now, client_stats)
.await
{
let checkout_time: u64 = now.elapsed().as_micros() as u64;
client_stats.checkout_time(checkout_time);
server
.stats()
.checkout_time(checkout_time, client_stats.application_name());
server.stats().active(client_stats.application_name());
client_stats.active();
return Ok((conn, address.clone()));
} else {
continue;
}
}
client_stats.idle();
Err(Error::AllServersDown)
}
@@ -574,11 +702,11 @@ impl ConnectionPool {
address: &Address,
server: &mut Server,
start: Instant,
client_process_id: i32,
client_info: &ClientStats,
) -> bool {
debug!("Running health check on server {:?}", address);
self.stats.server_tested(server.server_id());
server.stats().tested();
match tokio::time::timeout(
tokio::time::Duration::from_millis(self.settings.healthcheck_timeout),
@@ -589,20 +717,20 @@ impl ConnectionPool {
// Check if health check succeeded.
Ok(res) => match res {
Ok(_) => {
self.stats.checkout_time(
start.elapsed().as_micros(),
client_process_id,
server.server_id(),
);
self.stats
.server_active(client_process_id, server.server_id());
let checkout_time: u64 = start.elapsed().as_micros() as u64;
client_info.checkout_time(checkout_time);
server
.stats()
.checkout_time(checkout_time, client_info.application_name());
server.stats().active(client_info.application_name());
return true;
}
// Health check failed.
Err(err) => {
error!(
"Banning instance {:?} because of failed health check, {:?}",
"Failed health check on instance {:?}, error: {:?}",
address, err
);
}
@@ -611,7 +739,7 @@ impl ConnectionPool {
// Health check timed out.
Err(err) => {
error!(
"Banning instance {:?} because of health check timeout, {:?}",
"Health check timeout on instance {:?}, error: {:?}",
address, err
);
}
@@ -620,23 +748,29 @@ impl ConnectionPool {
// Don't leave a bad connection in the pool.
server.mark_bad();
self.ban(&address, BanReason::FailedHealthCheck, client_process_id);
self.ban(&address, BanReason::FailedHealthCheck, Some(client_info));
return false;
}
/// Ban an address (i.e. replica). It no longer will serve
/// traffic for any new transactions. Existing transactions on that replica
/// will finish successfully or error out to the clients.
pub fn ban(&self, address: &Address, reason: BanReason, client_id: i32) {
pub fn ban(&self, address: &Address, reason: BanReason, client_info: Option<&ClientStats>) {
// Primary can never be banned
if address.role == Role::Primary {
return;
}
error!("Banning instance {:?}, reason: {:?}", address, reason);
let now = chrono::offset::Utc::now().naive_utc();
let mut guard = self.banlist.write();
error!("Banning {:?}", address);
self.stats.client_ban_error(client_id, address.id);
if let Some(client_info) = client_info {
client_info.ban_error();
address.stats.error();
}
guard[address.shard].insert(address.clone(), (reason, now));
}
@@ -793,11 +927,26 @@ impl ConnectionPool {
/// Wrapper for the bb8 connection pool.
pub struct ServerPool {
/// Server address.
address: Address,
/// Server Postgres user.
user: User,
/// Server database.
database: String,
/// Client/server mapping.
client_server_map: ClientServerMap,
stats: Reporter,
/// Server auth hash (for auth passthrough).
auth_hash: Arc<RwLock<Option<String>>>,
/// Server plugins.
plugins: Option<Plugins>,
/// Should we clean up dirty connections before putting them into the pool?
cleanup_connections: bool,
}
impl ServerPool {
@@ -806,14 +955,18 @@ impl ServerPool {
user: User,
database: &str,
client_server_map: ClientServerMap,
stats: Reporter,
auth_hash: Arc<RwLock<Option<String>>>,
plugins: Option<Plugins>,
cleanup_connections: bool,
) -> ServerPool {
ServerPool {
address,
user,
user: user.clone(),
database: database.to_string(),
client_server_map,
stats,
auth_hash,
plugins,
cleanup_connections,
}
}
}
@@ -826,34 +979,44 @@ impl ManageConnection for ServerPool {
/// Attempts to create a new connection.
async fn connect(&self) -> Result<Self::Connection, Self::Error> {
info!("Creating a new server connection {:?}", self.address);
let server_id = rand::random::<i32>();
self.stats.server_register(
server_id,
self.address.id,
self.address.name(),
self.address.pool_name.clone(),
self.address.username.clone(),
);
self.stats.server_login(server_id);
let stats = Arc::new(ServerStats::new(
self.address.clone(),
tokio::time::Instant::now(),
));
stats.register(stats.clone());
// Connect to the PostgreSQL server.
match Server::startup(
server_id,
&self.address,
&self.user,
&self.database,
self.client_server_map.clone(),
self.stats.clone(),
stats.clone(),
self.auth_hash.clone(),
self.cleanup_connections,
)
.await
{
Ok(conn) => {
self.stats.server_idle(server_id);
Ok(mut conn) => {
if let Some(ref plugins) = self.plugins {
if let Some(ref prewarmer) = plugins.prewarmer {
let mut prewarmer = prewarmer::Prewarmer {
enabled: prewarmer.enabled,
server: &mut conn,
queries: &prewarmer.queries,
};
prewarmer.run().await?;
}
}
stats.idle();
Ok(conn)
}
Err(err) => {
self.stats.server_disconnecting(server_id);
stats.disconnect();
Err(err)
}
}
@@ -881,11 +1044,3 @@ pub fn get_pool(db: &str, user: &str) -> Option<ConnectionPool> {
pub fn get_all_pools() -> HashMap<PoolIdentifier, ConnectionPool> {
(*(*POOLS.load())).clone()
}
/// How many total servers we have in the config.
pub fn get_number_of_addresses() -> usize {
get_all_pools()
.iter()
.map(|(_, pool)| pool.databases())
.sum()
}

View File

@@ -1,14 +1,17 @@
use hyper::service::{make_service_fn, service_fn};
use hyper::{Body, Method, Request, Response, Server, StatusCode};
use log::{error, info, warn};
use log::{debug, error, info};
use phf::phf_map;
use std::collections::HashMap;
use std::fmt;
use std::net::SocketAddr;
use std::sync::atomic::Ordering;
use std::sync::Arc;
use crate::config::Address;
use crate::pool::get_all_pools;
use crate::stats::{get_address_stats, get_pool_stats, get_server_stats, ServerInformation};
use crate::pool::{get_all_pools, PoolIdentifier};
use crate::stats::pool::PoolStats;
use crate::stats::{get_server_stats, ServerStats};
struct MetricHelpType {
help: &'static str,
@@ -220,7 +223,7 @@ impl<Value: fmt::Display> PrometheusMetric<Value> {
Self::from_name(&format!("servers_{}", name), value, labels)
}
fn from_address(address: &Address, name: &str, value: i64) -> Option<PrometheusMetric<i64>> {
fn from_address(address: &Address, name: &str, value: u64) -> Option<PrometheusMetric<u64>> {
let mut labels = HashMap::new();
labels.insert("host", address.host.clone());
labels.insert("shard", address.shard.to_string());
@@ -231,10 +234,10 @@ impl<Value: fmt::Display> PrometheusMetric<Value> {
Self::from_name(&format!("stats_{}", name), value, labels)
}
fn from_pool(pool: &(String, String), name: &str, value: i64) -> Option<PrometheusMetric<i64>> {
fn from_pool(pool_id: PoolIdentifier, name: &str, value: u64) -> Option<PrometheusMetric<u64>> {
let mut labels = HashMap::new();
labels.insert("pool", pool.0.clone());
labels.insert("user", pool.1.clone());
labels.insert("pool", pool_id.db);
labels.insert("user", pool_id.user);
Self::from_name(&format!("pools_{}", name), value, labels)
}
@@ -261,20 +264,18 @@ async fn prometheus_stats(request: Request<Body>) -> Result<Response<Body>, hype
// Adds metrics shown in a SHOW STATS admin command.
fn push_address_stats(lines: &mut Vec<String>) {
let address_stats: HashMap<usize, HashMap<String, i64>> = get_address_stats();
for (_, pool) in get_all_pools() {
for shard in 0..pool.shards() {
for server in 0..pool.servers(shard) {
let address = pool.address(shard, server);
if let Some(address_stats) = address_stats.get(&address.id) {
for (key, value) in address_stats.iter() {
if let Some(prometheus_metric) =
PrometheusMetric::<i64>::from_address(address, key, *value)
{
lines.push(prometheus_metric.to_string());
} else {
warn!("Metric {} not implemented for {}", key, address.name());
}
let stats = &*address.stats;
for (key, value) in stats.clone() {
if let Some(prometheus_metric) =
PrometheusMetric::<u64>::from_address(address, &key, value)
{
lines.push(prometheus_metric.to_string());
} else {
debug!("Metric {} not implemented for {}", key, address.name());
}
}
}
@@ -284,17 +285,15 @@ fn push_address_stats(lines: &mut Vec<String>) {
// Adds relevant metrics shown in a SHOW POOLS admin command.
fn push_pool_stats(lines: &mut Vec<String>) {
let pool_stats = get_pool_stats();
for (pool, stats) in pool_stats.iter() {
for (name, value) in stats.iter() {
if let Some(prometheus_metric) = PrometheusMetric::<i64>::from_pool(pool, name, *value)
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());
} else {
warn!(
"Metric {} not implemented for ({},{})",
name, pool.0, pool.1
);
debug!("Metric {} not implemented for ({})", name, *pool_id);
}
}
}
@@ -319,7 +318,7 @@ fn push_database_stats(lines: &mut Vec<String>) {
{
lines.push(prometheus_metric.to_string());
} else {
warn!("Metric {} not implemented for {}", key, address.name());
debug!("Metric {} not implemented for {}", key, address.name());
}
}
}
@@ -330,9 +329,9 @@ fn push_database_stats(lines: &mut Vec<String>) {
// Adds relevant metrics shown in a SHOW SERVERS admin command.
fn push_server_stats(lines: &mut Vec<String>) {
let server_stats = get_server_stats();
let mut server_stats_by_addresses = HashMap::<String, ServerInformation>::new();
for (_, info) in server_stats {
server_stats_by_addresses.insert(info.address_name.clone(), info);
let mut server_stats_by_addresses = HashMap::<String, Arc<ServerStats>>::new();
for (_, stats) in server_stats {
server_stats_by_addresses.insert(stats.address_name(), stats);
}
for (_, pool) in get_all_pools() {
@@ -341,11 +340,23 @@ fn push_server_stats(lines: &mut Vec<String>) {
let address = pool.address(shard, server);
if let Some(server_info) = server_stats_by_addresses.get(&address.name()) {
let metrics = [
("bytes_received", server_info.bytes_received),
("bytes_sent", server_info.bytes_sent),
("transaction_count", server_info.transaction_count),
("query_count", server_info.query_count),
("error_count", server_info.error_count),
(
"bytes_received",
server_info.bytes_received.load(Ordering::Relaxed),
),
("bytes_sent", server_info.bytes_sent.load(Ordering::Relaxed)),
(
"transaction_count",
server_info.transaction_count.load(Ordering::Relaxed),
),
(
"query_count",
server_info.query_count.load(Ordering::Relaxed),
),
(
"error_count",
server_info.error_count.load(Ordering::Relaxed),
),
];
for (key, value) in metrics {
if let Some(prometheus_metric) =

View File

@@ -1,4 +1,4 @@
/// Route queries automatically based on explicitely requested
/// Route queries automatically based on explicitly requested
/// or implied query characteristics.
use bytes::{Buf, BytesMut};
use log::{debug, error};
@@ -6,13 +6,16 @@ use once_cell::sync::OnceCell;
use regex::{Regex, RegexSet};
use sqlparser::ast::Statement::{Query, StartTransaction};
use sqlparser::ast::{
BinaryOperator, Expr, Ident, JoinConstraint, JoinOperator, SetExpr, TableFactor, Value,
BinaryOperator, Expr, Ident, JoinConstraint, JoinOperator, SetExpr, Statement, TableFactor,
Value,
};
use sqlparser::dialect::PostgreSqlDialect;
use sqlparser::parser::Parser;
use crate::config::Role;
use crate::errors::Error;
use crate::messages::BytesMutReader;
use crate::plugins::{Intercept, Plugin, PluginOutput, QueryLogger, TableAccess};
use crate::pool::PoolSettings;
use crate::sharding::Sharder;
@@ -129,6 +132,10 @@ impl QueryRouter {
self.pool_settings = pool_settings;
}
pub fn pool_settings<'a>(&'a self) -> &'a PoolSettings {
&self.pool_settings
}
/// Try to parse a command and execute it.
pub fn try_execute_command(&mut self, message_buffer: &BytesMut) -> Option<(Command, String)> {
let mut message_cursor = Cursor::new(message_buffer);
@@ -324,10 +331,7 @@ impl QueryRouter {
Some((command, value))
}
/// Try to infer which server to connect to based on the contents of the query.
pub fn infer(&mut self, message: &BytesMut) -> bool {
debug!("Inferring role");
pub fn parse(message: &BytesMut) -> Result<Vec<Statement>, Error> {
let mut message_cursor = Cursor::new(message);
let code = message_cursor.get_u8() as char;
@@ -344,37 +348,39 @@ impl QueryRouter {
// Parse (prepared statement)
'P' => {
// Reads statement name
message_cursor.read_string().unwrap();
let _name = message_cursor.read_string().unwrap();
// Reads query string
let query = message_cursor.read_string().unwrap();
debug!("Prepared statement: '{}'", query);
query
}
_ => return false,
_ => return Err(Error::UnsupportedStatement),
};
let ast = match Parser::parse_sql(&PostgreSqlDialect {}, &query) {
Ok(ast) => ast,
match Parser::parse_sql(&PostgreSqlDialect {}, &query) {
Ok(ast) => Ok(ast),
Err(err) => {
// SELECT ... FOR UPDATE won't get parsed correctly.
debug!("{}: {}", err, query);
self.active_role = Some(Role::Primary);
return false;
Err(Error::QueryRouterParserError(err.to_string()))
}
};
}
}
debug!("AST: {:?}", ast);
/// 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> {
debug!("Inferring role");
if ast.is_empty() {
// That's weird, no idea, let's go to primary
self.active_role = Some(Role::Primary);
return false;
return Err(Error::QueryRouterParserError("empty query".into()));
}
for q in &ast {
for q in ast {
match q {
// All transactions go to the primary, probably a write.
StartTransaction { .. } => {
@@ -418,7 +424,7 @@ impl QueryRouter {
};
}
true
Ok(())
}
/// Parse the shard number from the Bind message
@@ -783,6 +789,52 @@ impl QueryRouter {
}
}
/// Add your plugins here and execute them.
pub async fn execute_plugins(&self, ast: &Vec<Statement>) -> Result<PluginOutput, Error> {
let plugins = match self.pool_settings.plugins {
Some(ref plugins) => plugins,
None => return Ok(PluginOutput::Allow),
};
if let Some(ref query_logger) = plugins.query_logger {
let mut query_logger = QueryLogger {
enabled: query_logger.enabled,
user: &self.pool_settings.user.username,
db: &self.pool_settings.db,
};
let _ = query_logger.run(&self, ast).await;
}
if let Some(ref intercept) = plugins.intercept {
let mut intercept = Intercept {
enabled: intercept.enabled,
config: &intercept,
};
let result = intercept.run(&self, ast).await;
if let Ok(PluginOutput::Intercept(output)) = result {
return Ok(PluginOutput::Intercept(output));
}
}
if let Some(ref table_access) = plugins.table_access {
let mut table_access = TableAccess {
enabled: table_access.enabled,
tables: &table_access.tables,
};
let result = table_access.run(&self, ast).await;
if let Ok(PluginOutput::Deny(error)) = result {
return Ok(PluginOutput::Deny(error));
}
}
Ok(PluginOutput::Allow)
}
fn set_sharding_key(&mut self, sharding_key: i64) -> Option<usize> {
let sharder = Sharder::new(
self.pool_settings.shards,
@@ -810,11 +862,22 @@ impl QueryRouter {
/// Should we attempt to parse queries?
pub fn query_parser_enabled(&self) -> bool {
let enabled = match self.query_parser_enabled {
None => self.pool_settings.query_parser_enabled,
Some(value) => value,
};
None => {
debug!(
"Using pool settings, query_parser_enabled: {}",
self.pool_settings.query_parser_enabled
);
self.pool_settings.query_parser_enabled
}
debug!("Query parser enabled: {}", enabled);
Some(value) => {
debug!(
"Using query parser override, query_parser_enabled: {}",
value
);
value
}
};
enabled
}
@@ -862,7 +925,7 @@ mod test {
for query in queries {
// It's a recognized query
assert!(qr.infer(&query));
assert!(qr.infer(&QueryRouter::parse(&query).unwrap()).is_ok());
assert_eq!(qr.role(), Some(Role::Replica));
}
}
@@ -881,7 +944,7 @@ mod test {
for query in queries {
// It's a recognized query
assert!(qr.infer(&query));
assert!(qr.infer(&QueryRouter::parse(&query).unwrap()).is_ok());
assert_eq!(qr.role(), Some(Role::Primary));
}
}
@@ -893,7 +956,7 @@ mod test {
let query = simple_query("SELECT * FROM items WHERE id = 5");
assert!(qr.try_execute_command(&simple_query("SET PRIMARY READS TO on")) != None);
assert!(qr.infer(&query));
assert!(qr.infer(&QueryRouter::parse(&query).unwrap()).is_ok());
assert_eq!(qr.role(), None);
}
@@ -913,7 +976,7 @@ mod test {
res.put(prepared_stmt);
res.put_i16(0);
assert!(qr.infer(&res));
assert!(qr.infer(&QueryRouter::parse(&res).unwrap()).is_ok());
assert_eq!(qr.role(), Some(Role::Replica));
}
@@ -1077,11 +1140,11 @@ mod test {
assert_eq!(qr.role(), None);
let query = simple_query("INSERT INTO test_table VALUES (1)");
assert!(qr.infer(&query));
assert!(qr.infer(&QueryRouter::parse(&query).unwrap()).is_ok());
assert_eq!(qr.role(), Some(Role::Primary));
let query = simple_query("SELECT * FROM test_table");
assert!(qr.infer(&query));
assert!(qr.infer(&QueryRouter::parse(&query).unwrap()).is_ok());
assert_eq!(qr.role(), Some(Role::Replica));
assert!(qr.query_parser_enabled());
@@ -1110,6 +1173,11 @@ mod test {
sharding_key_regex: None,
shard_id_regex: None,
regex_search_limit: 1000,
auth_query: None,
auth_query_password: None,
auth_query_user: None,
db: "test".to_string(),
plugins: None,
};
let mut qr = QueryRouter::new();
assert_eq!(qr.active_role, None);
@@ -1139,15 +1207,24 @@ mod test {
QueryRouter::setup();
let mut qr = QueryRouter::new();
assert!(qr.infer(&simple_query("BEGIN; SELECT 1; COMMIT;")));
assert!(qr
.infer(&QueryRouter::parse(&simple_query("BEGIN; SELECT 1; COMMIT;")).unwrap())
.is_ok());
assert_eq!(qr.role(), Role::Primary);
assert!(qr.infer(&simple_query("SELECT 1; SELECT 2;")));
assert!(qr
.infer(&QueryRouter::parse(&simple_query("SELECT 1; SELECT 2;")).unwrap())
.is_ok());
assert_eq!(qr.role(), Role::Replica);
assert!(qr.infer(&simple_query(
"SELECT 123; INSERT INTO t VALUES (5); SELECT 1;"
)));
assert!(qr
.infer(
&QueryRouter::parse(&simple_query(
"SELECT 123; INSERT INTO t VALUES (5); SELECT 1;"
))
.unwrap()
)
.is_ok());
assert_eq!(qr.role(), Role::Primary);
}
@@ -1171,7 +1248,13 @@ mod test {
sharding_key_regex: Some(Regex::new(r"/\* sharding_key: (\d+) \*/").unwrap()),
shard_id_regex: Some(Regex::new(r"/\* shard_id: (\d+) \*/").unwrap()),
regex_search_limit: 1000,
auth_query: None,
auth_query_password: None,
auth_query_user: None,
db: "test".to_string(),
plugins: None,
};
let mut qr = QueryRouter::new();
qr.update_pool_settings(pool_settings.clone());
@@ -1202,47 +1285,84 @@ mod test {
qr.pool_settings.automatic_sharding_key = Some("data.id".to_string());
qr.pool_settings.shards = 3;
assert!(qr.infer(&simple_query("SELECT * FROM data WHERE id = 5")));
assert!(qr
.infer(&QueryRouter::parse(&simple_query("SELECT * FROM data WHERE id = 5")).unwrap())
.is_ok());
assert_eq!(qr.shard(), 2);
assert!(qr.infer(&simple_query(
"SELECT one, two, three FROM public.data WHERE id = 6"
)));
assert!(qr
.infer(
&QueryRouter::parse(&simple_query(
"SELECT one, two, three FROM public.data WHERE id = 6"
))
.unwrap()
)
.is_ok());
assert_eq!(qr.shard(), 0);
assert!(qr.infer(&simple_query(
"SELECT * FROM data
assert!(qr
.infer(
&QueryRouter::parse(&simple_query(
"SELECT * FROM data
INNER JOIN t2 ON data.id = 5
AND t2.data_id = data.id
WHERE data.id = 5"
)));
))
.unwrap()
)
.is_ok());
assert_eq!(qr.shard(), 2);
// Shard did not move because we couldn't determine the sharding key since it could be ambiguous
// in the query.
assert!(qr.infer(&simple_query(
"SELECT * FROM t2 INNER JOIN data ON id = 6 AND data.id = t2.data_id"
)));
assert!(qr
.infer(
&QueryRouter::parse(&simple_query(
"SELECT * FROM t2 INNER JOIN data ON id = 6 AND data.id = t2.data_id"
))
.unwrap()
)
.is_ok());
assert_eq!(qr.shard(), 2);
assert!(qr.infer(&simple_query(
r#"SELECT * FROM "public"."data" WHERE "id" = 6"#
)));
assert!(qr
.infer(
&QueryRouter::parse(&simple_query(
r#"SELECT * FROM "public"."data" WHERE "id" = 6"#
))
.unwrap()
)
.is_ok());
assert_eq!(qr.shard(), 0);
assert!(qr.infer(&simple_query(
r#"SELECT * FROM "public"."data" WHERE "data"."id" = 5"#
)));
assert!(qr
.infer(
&QueryRouter::parse(&simple_query(
r#"SELECT * FROM "public"."data" WHERE "data"."id" = 5"#
))
.unwrap()
)
.is_ok());
assert_eq!(qr.shard(), 2);
// Super unique sharding key
qr.pool_settings.automatic_sharding_key = Some("*.unique_enough_column_name".to_string());
assert!(qr.infer(&simple_query(
"SELECT * FROM table_x WHERE unique_enough_column_name = 6"
)));
assert!(qr
.infer(
&QueryRouter::parse(&simple_query(
"SELECT * FROM table_x WHERE unique_enough_column_name = 6"
))
.unwrap()
)
.is_ok());
assert_eq!(qr.shard(), 0);
assert!(qr.infer(&simple_query("SELECT * FROM table_y WHERE another_key = 5")));
assert!(qr
.infer(
&QueryRouter::parse(&simple_query("SELECT * FROM table_y WHERE another_key = 5"))
.unwrap()
)
.is_ok());
assert_eq!(qr.shard(), 0);
}
@@ -1266,11 +1386,61 @@ mod test {
qr.pool_settings.automatic_sharding_key = Some("data.id".to_string());
qr.pool_settings.shards = 3;
assert!(qr.infer(&simple_query(stmt)));
assert!(qr
.infer(&QueryRouter::parse(&simple_query(stmt)).unwrap())
.is_ok());
assert_eq!(qr.placeholders.len(), 1);
assert!(qr.infer_shard_from_bind(&bind));
assert_eq!(qr.shard(), 2);
assert!(qr.placeholders.is_empty());
}
#[tokio::test]
async fn test_table_access_plugin() {
use crate::config::{Plugins, TableAccess};
let table_access = TableAccess {
enabled: true,
tables: vec![String::from("pg_database")],
};
let plugins = Plugins {
table_access: Some(table_access),
intercept: None,
query_logger: None,
prewarmer: None,
};
QueryRouter::setup();
let mut pool_settings = PoolSettings::default();
pool_settings.query_parser_enabled = true;
pool_settings.plugins = Some(plugins);
let mut qr = QueryRouter::new();
qr.update_pool_settings(pool_settings);
let query = simple_query("SELECT * FROM pg_database");
let ast = QueryRouter::parse(&query).unwrap();
let res = qr.execute_plugins(&ast).await;
assert_eq!(
res,
Ok(PluginOutput::Deny(
"permission for table \"pg_database\" denied".to_string()
))
);
}
#[tokio::test]
async fn test_plugins_disabled_by_defaault() {
QueryRouter::setup();
let qr = QueryRouter::new();
let query = simple_query("SELECT * FROM pg_database");
let ast = QueryRouter::parse(&query).unwrap();
let res = qr.execute_plugins(&ast).await;
assert_eq!(res, Ok(PluginOutput::Allow));
}
}

View File

@@ -1,37 +1,158 @@
/// Implementation of the PostgreSQL server (database) protocol.
/// Here we are pretending to the a Postgres client.
use bytes::{Buf, BufMut, BytesMut};
use fallible_iterator::FallibleIterator;
use log::{debug, error, info, trace, warn};
use parking_lot::{Mutex, RwLock};
use postgres_protocol::message;
use std::collections::{BTreeSet, HashMap};
use std::io::Read;
use std::net::IpAddr;
use std::sync::Arc;
use std::time::SystemTime;
use tokio::io::{AsyncReadExt, BufReader};
use tokio::net::{
tcp::{OwnedReadHalf, OwnedWriteHalf},
TcpStream,
};
use tokio::io::{AsyncRead, AsyncReadExt, AsyncWrite, BufStream};
use tokio::net::TcpStream;
use tokio_rustls::rustls::{OwnedTrustAnchor, RootCertStore};
use tokio_rustls::{client::TlsStream, TlsConnector};
use crate::config::{Address, User};
use crate::config::{get_config, Address, User};
use crate::constants::*;
use crate::errors::Error;
use crate::dns_cache::{AddrSet, CACHED_RESOLVER};
use crate::errors::{Error, ServerIdentifier};
use crate::messages::*;
use crate::mirrors::MirroringManager;
use crate::pool::ClientServerMap;
use crate::scram::ScramSha256;
use crate::stats::Reporter;
use crate::stats::ServerStats;
use std::io::Write;
use pin_project::pin_project;
#[pin_project(project = SteamInnerProj)]
pub enum StreamInner {
Plain {
#[pin]
stream: TcpStream,
},
Tls {
#[pin]
stream: TlsStream<TcpStream>,
},
}
impl AsyncWrite for StreamInner {
fn poll_write(
self: std::pin::Pin<&mut Self>,
cx: &mut std::task::Context<'_>,
buf: &[u8],
) -> std::task::Poll<Result<usize, std::io::Error>> {
let this = self.project();
match this {
SteamInnerProj::Tls { stream } => stream.poll_write(cx, buf),
SteamInnerProj::Plain { stream } => stream.poll_write(cx, buf),
}
}
fn poll_flush(
self: std::pin::Pin<&mut Self>,
cx: &mut std::task::Context<'_>,
) -> std::task::Poll<Result<(), std::io::Error>> {
let this = self.project();
match this {
SteamInnerProj::Tls { stream } => stream.poll_flush(cx),
SteamInnerProj::Plain { stream } => stream.poll_flush(cx),
}
}
fn poll_shutdown(
self: std::pin::Pin<&mut Self>,
cx: &mut std::task::Context<'_>,
) -> std::task::Poll<Result<(), std::io::Error>> {
let this = self.project();
match this {
SteamInnerProj::Tls { stream } => stream.poll_shutdown(cx),
SteamInnerProj::Plain { stream } => stream.poll_shutdown(cx),
}
}
}
impl AsyncRead for StreamInner {
fn poll_read(
self: std::pin::Pin<&mut Self>,
cx: &mut std::task::Context<'_>,
buf: &mut tokio::io::ReadBuf<'_>,
) -> std::task::Poll<std::io::Result<()>> {
let this = self.project();
match this {
SteamInnerProj::Tls { stream } => stream.poll_read(cx, buf),
SteamInnerProj::Plain { stream } => stream.poll_read(cx, buf),
}
}
}
impl StreamInner {
pub fn try_write(&mut self, buf: &[u8]) -> std::io::Result<usize> {
match self {
StreamInner::Tls { stream } => {
let r = stream.get_mut();
let mut w = r.1.writer();
w.write(buf)
}
StreamInner::Plain { stream } => stream.try_write(buf),
}
}
}
#[derive(Copy, Clone)]
struct CleanupState {
/// If server connection requires DISCARD ALL before checkin because of set statement
needs_cleanup_set: bool,
/// If server connection requires DISCARD ALL before checkin because of prepare statement
needs_cleanup_prepare: bool,
}
impl CleanupState {
fn new() -> Self {
CleanupState {
needs_cleanup_set: false,
needs_cleanup_prepare: false,
}
}
fn needs_cleanup(&self) -> bool {
self.needs_cleanup_set || self.needs_cleanup_prepare
}
fn set_true(&mut self) {
self.needs_cleanup_set = true;
self.needs_cleanup_prepare = true;
}
fn reset(&mut self) {
self.needs_cleanup_set = false;
self.needs_cleanup_prepare = false;
}
}
impl std::fmt::Display for CleanupState {
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
write!(
f,
"SET: {}, PREPARE: {}",
self.needs_cleanup_set, self.needs_cleanup_prepare
)
}
}
/// Server state.
pub struct Server {
server_id: i32,
/// Server host, e.g. localhost,
/// port, e.g. 5432, and role, e.g. primary or replica.
address: Address,
/// Buffered read socket.
read: BufReader<OwnedReadHalf>,
/// Unbuffered write socket (our client code buffers).
write: OwnedWriteHalf,
/// Server TCP connection.
stream: BufStream<StreamInner>,
/// Our server response buffer. We buffer data before we give it to the client.
buffer: BytesMut,
@@ -52,8 +173,8 @@ pub struct Server {
/// Is the server broken? We'll remote it from the pool if so.
bad: bool,
/// If server connection requires a DISCARD ALL before checkin
needs_cleanup: bool,
/// If server connection requires DISCARD ALL before checkin
cleanup_state: CleanupState,
/// Mapping of clients and servers used for query cancellation.
client_server_map: ClientServerMap,
@@ -62,28 +183,56 @@ pub struct Server {
connected_at: chrono::naive::NaiveDateTime,
/// Reports various metrics, e.g. data sent & received.
stats: Reporter,
stats: Arc<ServerStats>,
/// Application name using the server at the moment.
application_name: String,
// Last time that a successful server send or response happened
/// Last time that a successful server send or response happened
last_activity: SystemTime,
mirror_manager: Option<MirroringManager>,
/// Associated addresses used
addr_set: Option<AddrSet>,
/// Should clean up dirty connections?
cleanup_connections: bool,
/// Prepared statements
prepared_statements: BTreeSet<String>,
}
impl Server {
/// Pretend to be the Postgres client and connect to the server given host, port and credentials.
/// Perform the authentication and return the server in a ready for query state.
pub async fn startup(
server_id: i32,
address: &Address,
user: &User,
database: &str,
client_server_map: ClientServerMap,
stats: Reporter,
stats: Arc<ServerStats>,
auth_hash: Arc<RwLock<Option<String>>>,
cleanup_connections: bool,
) -> Result<Server, Error> {
let cached_resolver = CACHED_RESOLVER.load();
let mut addr_set: Option<AddrSet> = None;
// If we are caching addresses and hostname is not an IP
if cached_resolver.enabled() && address.host.parse::<IpAddr>().is_err() {
debug!("Resolving {}", &address.host);
addr_set = match cached_resolver.lookup_ip(&address.host).await {
Ok(ok) => {
debug!("Obtained: {:?}", ok);
Some(ok)
}
Err(err) => {
warn!("Error trying to resolve {}, ({:?})", &address.host, err);
None
}
}
};
let mut stream =
match TcpStream::connect(&format!("{}:{}", &address.host, address.port)).await {
Ok(stream) => stream,
@@ -95,30 +244,137 @@ impl Server {
)));
}
};
// TCP timeouts.
configure_socket(&stream);
let config = get_config();
let mut stream = if config.general.server_tls {
// Request a TLS connection
ssl_request(&mut stream).await?;
let response = match stream.read_u8().await {
Ok(response) => response as char,
Err(err) => {
return Err(Error::SocketError(format!(
"Server socket error: {:?}",
err
)))
}
};
match response {
// Server supports TLS
'S' => {
debug!("Connecting to server using TLS");
let mut root_store = RootCertStore::empty();
root_store.add_server_trust_anchors(
webpki_roots::TLS_SERVER_ROOTS.0.iter().map(|ta| {
OwnedTrustAnchor::from_subject_spki_name_constraints(
ta.subject,
ta.spki,
ta.name_constraints,
)
}),
);
let mut tls_config = rustls::ClientConfig::builder()
.with_safe_defaults()
.with_root_certificates(root_store)
.with_no_client_auth();
// Equivalent to sslmode=prefer which is fine most places.
// If you want verify-full, change `verify_server_certificate` to true.
if !config.general.verify_server_certificate {
let mut dangerous = tls_config.dangerous();
dangerous.set_certificate_verifier(Arc::new(
crate::tls::NoCertificateVerification {},
));
}
let connector = TlsConnector::from(Arc::new(tls_config));
let stream = match connector
.connect(address.host.as_str().try_into().unwrap(), stream)
.await
{
Ok(stream) => stream,
Err(err) => {
return Err(Error::SocketError(format!("Server TLS error: {:?}", err)))
}
};
StreamInner::Tls { stream }
}
// Server does not support TLS
'N' => StreamInner::Plain { stream },
// Something else?
m => {
return Err(Error::SocketError(format!(
"Unknown message: {}",
m as char
)));
}
}
} else {
StreamInner::Plain { stream }
};
// let (read, write) = split(stream);
// let (mut read, mut write) = (ReadInner::Plain { stream: read }, WriteInner::Plain { stream: write });
trace!("Sending StartupMessage");
// StartupMessage
startup(&mut stream, &user.username, database).await?;
let username = match user.server_username {
Some(ref server_username) => server_username,
None => &user.username,
};
let password = match user.server_password {
Some(ref server_password) => Some(server_password),
None => match user.password {
Some(ref password) => Some(password),
None => None,
},
};
startup(&mut stream, username, database).await?;
let mut server_info = BytesMut::new();
let mut process_id: i32 = 0;
let mut secret_key: i32 = 0;
let server_identifier = ServerIdentifier::new(username, &database);
// We'll be handling multiple packets, but they will all be structured the same.
// We'll loop here until this exchange is complete.
let mut scram = ScramSha256::new(&user.password);
let mut scram: Option<ScramSha256> = match password {
Some(password) => Some(ScramSha256::new(password)),
None => None,
};
loop {
let code = match stream.read_u8().await {
Ok(code) => code as char,
Err(_) => return Err(Error::SocketError(format!("Error reading message code on server startup {{ username: {:?}, database: {:?} }}", user.username, database))),
Err(_) => {
return Err(Error::ServerStartupError(
"message code".into(),
server_identifier,
))
}
};
let len = match stream.read_i32().await {
Ok(len) => len,
Err(_) => return Err(Error::SocketError(format!("Error reading message len on server startup {{ username: {:?}, database: {:?} }}", user.username, database))),
Err(_) => {
return Err(Error::ServerStartupError(
"message len".into(),
server_identifier,
))
}
};
trace!("Message: {}", code);
@@ -129,7 +385,12 @@ impl Server {
// Determine which kind of authentication is required, if any.
let auth_code = match stream.read_i32().await {
Ok(auth_code) => auth_code,
Err(_) => return Err(Error::SocketError(format!("Error reading auth code on server startup {{ username: {:?}, database: {:?} }}", user.username, database))),
Err(_) => {
return Err(Error::ServerStartupError(
"auth code".into(),
server_identifier,
))
}
};
trace!("Auth: {}", auth_code);
@@ -142,32 +403,79 @@ impl Server {
match stream.read_exact(&mut salt).await {
Ok(_) => (),
Err(_) => return Err(Error::SocketError(format!("Error reading salt on server startup {{ username: {:?}, database: {:?} }}", user.username, database))),
Err(_) => {
return Err(Error::ServerStartupError(
"salt".into(),
server_identifier,
))
}
};
md5_password(&mut stream, &user.username, &user.password, &salt[..])
.await?;
match password {
// Using plaintext password
Some(password) => {
md5_password(&mut stream, username, password, &salt[..]).await?
}
// Using auth passthrough, in this case we should already have a
// hash obtained when the pool was validated. If we reach this point
// and don't have a hash, we return an error.
None => {
let option_hash = (*auth_hash.read()).clone();
match option_hash {
Some(hash) =>
md5_password_with_hash(
&mut stream,
&hash,
&salt[..],
)
.await?,
None => return Err(
Error::ServerAuthError(
"Auth passthrough (auth_query) failed and no user password is set in cleartext".into(),
server_identifier
)
),
}
}
}
}
AUTHENTICATION_SUCCESSFUL => (),
SASL => {
if scram.is_none() {
return Err(Error::ServerAuthError(
"SASL auth required and no password specified. \
Auth passthrough (auth_query) method is currently \
unsupported for SASL auth"
.into(),
server_identifier,
));
}
debug!("Starting SASL authentication");
let sasl_len = (len - 8) as usize;
let mut sasl_auth = vec![0u8; sasl_len];
match stream.read_exact(&mut sasl_auth).await {
Ok(_) => (),
Err(_) => return Err(Error::SocketError(format!("Error reading sasl message on server startup {{ username: {:?}, database: {:?} }}", user.username, database))),
Err(_) => {
return Err(Error::ServerStartupError(
"sasl message".into(),
server_identifier,
))
}
};
let sasl_type = String::from_utf8_lossy(&sasl_auth[..sasl_len - 2]);
if sasl_type == SCRAM_SHA_256 {
if sasl_type.contains(SCRAM_SHA_256) {
debug!("Using {}", SCRAM_SHA_256);
// Generate client message.
let sasl_response = scram.message();
let sasl_response = scram.as_mut().unwrap().message();
// SASLInitialResponse (F)
let mut res = BytesMut::new();
@@ -186,7 +494,7 @@ impl Server {
res.put_i32(sasl_response.len() as i32);
res.put(sasl_response);
write_all(&mut stream, res).await?;
write_all_flush(&mut stream, &res).await?;
} else {
error!("Unsupported SCRAM version: {}", sasl_type);
return Err(Error::ServerError);
@@ -200,11 +508,16 @@ impl Server {
match stream.read_exact(&mut sasl_data).await {
Ok(_) => (),
Err(_) => return Err(Error::SocketError(format!("Error reading sasl cont message on server startup {{ username: {:?}, database: {:?} }}", user.username, database))),
Err(_) => {
return Err(Error::ServerStartupError(
"sasl cont message".into(),
server_identifier,
))
}
};
let msg = BytesMut::from(&sasl_data[..]);
let sasl_response = scram.update(&msg)?;
let sasl_response = scram.as_mut().unwrap().update(&msg)?;
// SASLResponse
let mut res = BytesMut::new();
@@ -212,7 +525,7 @@ impl Server {
res.put_i32(4 + sasl_response.len() as i32);
res.put(sasl_response);
write_all(&mut stream, res).await?;
write_all_flush(&mut stream, &res).await?;
}
SASL_FINAL => {
@@ -221,10 +534,19 @@ impl Server {
let mut sasl_final = vec![0u8; len as usize - 8];
match stream.read_exact(&mut sasl_final).await {
Ok(_) => (),
Err(_) => return Err(Error::SocketError(format!("Error reading sasl final message on server startup {{ username: {:?}, database: {:?} }}", user.username, database))),
Err(_) => {
return Err(Error::ServerStartupError(
"sasl final message".into(),
server_identifier,
))
}
};
match scram.finish(&BytesMut::from(&sasl_final[..])) {
match scram
.as_mut()
.unwrap()
.finish(&BytesMut::from(&sasl_final[..]))
{
Ok(_) => {
debug!("SASL authentication successful");
}
@@ -247,7 +569,12 @@ impl Server {
'E' => {
let error_code = match stream.read_u8().await {
Ok(error_code) => error_code,
Err(_) => return Err(Error::SocketError(format!("Error reading error code message on server startup {{ username: {:?}, database: {:?} }}", user.username, database))),
Err(_) => {
return Err(Error::ServerStartupError(
"error code message".into(),
server_identifier,
))
}
};
trace!("Error: {}", error_code);
@@ -263,7 +590,12 @@ impl Server {
match stream.read_exact(&mut error).await {
Ok(_) => (),
Err(_) => return Err(Error::SocketError(format!("Error reading error message on server startup {{ username: {:?}, database: {:?} }}", user.username, database))),
Err(_) => {
return Err(Error::ServerStartupError(
"error message".into(),
server_identifier,
))
}
};
// TODO: the error message contains multiple fields; we can decode them and
@@ -282,7 +614,12 @@ impl Server {
match stream.read_exact(&mut param).await {
Ok(_) => (),
Err(_) => return Err(Error::SocketError(format!("Error reading parameter status message on server startup {{ username: {:?}, database: {:?} }}", user.username, database))),
Err(_) => {
return Err(Error::ServerStartupError(
"parameter status message".into(),
server_identifier,
))
}
};
// Save the parameter so we can pass it to the client later.
@@ -299,12 +636,22 @@ impl Server {
// See: <https://www.postgresql.org/docs/12/protocol-message-formats.html>.
process_id = match stream.read_i32().await {
Ok(id) => id,
Err(_) => return Err(Error::SocketError(format!("Error reading process id message on server startup {{ username: {:?}, database: {:?} }}", user.username, database))),
Err(_) => {
return Err(Error::ServerStartupError(
"process id message".into(),
server_identifier,
))
}
};
secret_key = match stream.read_i32().await {
Ok(id) => id,
Err(_) => return Err(Error::SocketError(format!("Error reading secret key message on server startup {{ username: {:?}, database: {:?} }}", user.username, database))),
Err(_) => {
return Err(Error::ServerStartupError(
"secret key message".into(),
server_identifier,
))
}
};
}
@@ -314,25 +661,27 @@ impl Server {
match stream.read_exact(&mut idle).await {
Ok(_) => (),
Err(_) => return Err(Error::SocketError(format!("Error reading transaction status message on server startup {{ username: {:?}, database: {:?} }}", user.username, database))),
Err(_) => {
return Err(Error::ServerStartupError(
"transaction status message".into(),
server_identifier,
))
}
};
let (read, write) = stream.into_split();
let mut server = Server {
address: address.clone(),
read: BufReader::new(read),
write,
stream: BufStream::new(stream),
buffer: BytesMut::with_capacity(8196),
server_info,
server_id,
process_id,
secret_key,
in_transaction: false,
data_available: false,
bad: false,
needs_cleanup: false,
cleanup_state: CleanupState::new(),
client_server_map,
addr_set,
connected_at: chrono::offset::Utc::now().naive_utc(),
stats,
application_name: String::new(),
@@ -345,6 +694,8 @@ impl Server {
address.mirrors.clone(),
)),
},
cleanup_connections,
prepared_statements: BTreeSet::new(),
};
server.set_name("pgcat").await?;
@@ -377,7 +728,7 @@ impl Server {
Ok(stream) => stream,
Err(err) => {
error!("Could not connect to server: {}", err);
return Err(Error::SocketError(format!("Error reading cancel message")));
return Err(Error::SocketError("Error reading cancel message".into()));
}
};
configure_socket(&stream);
@@ -390,22 +741,25 @@ impl Server {
bytes.put_i32(process_id);
bytes.put_i32(secret_key);
write_all(&mut stream, bytes).await
write_all_flush(&mut stream, &bytes).await
}
/// Send messages to the server from the client.
pub async fn send(&mut self, messages: &BytesMut) -> Result<(), Error> {
self.mirror_send(messages);
self.stats.data_sent(messages.len(), self.server_id);
self.stats().data_sent(messages.len());
match write_all_half(&mut self.write, messages).await {
match write_all_flush(&mut self.stream, &messages).await {
Ok(_) => {
// Successfully sent to server
self.last_activity = SystemTime::now();
Ok(())
}
Err(err) => {
error!("Terminating server because of: {:?}", err);
error!(
"Terminating server {:?} because of: {:?}",
self.address, err
);
self.bad = true;
Err(err)
}
@@ -417,10 +771,13 @@ impl Server {
/// in order to receive all data the server has to offer.
pub async fn recv(&mut self) -> Result<BytesMut, Error> {
loop {
let mut message = match read_message(&mut self.read).await {
let mut message = match read_message(&mut self.stream).await {
Ok(message) => message,
Err(err) => {
error!("Terminating server because of: {:?}", err);
error!(
"Terminating server {:?} because of: {:?}",
self.address, err
);
self.bad = true;
return Err(err);
}
@@ -487,12 +844,12 @@ impl Server {
// This will reduce amount of discard statements sent
if !self.in_transaction {
debug!("Server connection marked for clean up");
self.needs_cleanup = true;
self.cleanup_state.needs_cleanup_set = true;
}
}
"PREPARE\0" => {
debug!("Server connection marked for clean up");
self.needs_cleanup = true;
self.cleanup_state.needs_cleanup_prepare = true;
}
_ => (),
}
@@ -545,7 +902,7 @@ impl Server {
let bytes = self.buffer.clone();
// Keep track of how much data we got from the server for stats.
self.stats.data_received(bytes.len(), self.server_id);
self.stats().data_received(bytes.len());
// Clear the buffer for next query.
self.buffer.clear();
@@ -557,6 +914,43 @@ impl Server {
Ok(bytes)
}
pub fn will_prepare(&mut self, name: &str) {
debug!("Will prepare `{}`", name);
self.prepared_statements.insert(name.to_string());
}
pub fn should_prepare(&self, name: &str) -> bool {
let should_prepare = !self.prepared_statements.contains(name);
debug!("Should prepare `{}`: {}", name, should_prepare);
if should_prepare {
self.stats.prepared_cache_miss();
} else {
self.stats.prepared_cache_hit();
}
should_prepare
}
pub async fn prepare(&mut self, parse: &Parse) -> Result<(), Error> {
debug!("Preparing `{}`", parse.name);
let bytes: BytesMut = parse.try_into()?;
self.send(&bytes).await?;
self.send(&flush()).await?;
// Read and discard ParseComplete (B)
let _ = read_message(&mut self.stream).await?;
self.prepared_statements.insert(parse.name.to_string());
debug!("Prepared `{}`", parse.name);
Ok(())
}
/// If the server is still inside a transaction.
/// If the client disconnects while the server is in a transaction, we will clean it up.
pub fn in_transaction(&self) -> bool {
@@ -573,7 +967,23 @@ impl Server {
/// Server & client are out of sync, we must discard this connection.
/// This happens with clients that misbehave.
pub fn is_bad(&self) -> bool {
self.bad
if self.bad {
return self.bad;
};
let cached_resolver = CACHED_RESOLVER.load();
if cached_resolver.enabled() {
if let Some(addr_set) = &self.addr_set {
if cached_resolver.has_changed(self.address.host.as_str(), addr_set) {
warn!(
"DNS changed for {}, it was {:?}. Dropping server connection.",
self.address.host.as_str(),
addr_set
);
return true;
}
}
}
false
}
/// Get server startup information to forward it to the client.
@@ -606,6 +1016,8 @@ impl Server {
/// It will use the simple query protocol.
/// Result will not be returned, so this is useful for things like `SET` or `ROLLBACK`.
pub async fn query(&mut self, query: &str) -> Result<(), Error> {
debug!("Running `{}` on server {:?}", query, self.address);
let query = simple_query(query);
self.send(&query).await?;
@@ -638,10 +1050,11 @@ impl Server {
// to avoid leaking state between clients. For performance reasons we only
// send `DISCARD ALL` if we think the session is altered instead of just sending
// it before each checkin.
if self.needs_cleanup {
warn!("Server returned with session state altered, discarding state");
if self.cleanup_state.needs_cleanup() && self.cleanup_connections {
warn!("Server returned with session state altered, discarding state ({}) for application {}", self.cleanup_state, self.application_name);
self.query("DISCARD ALL").await?;
self.needs_cleanup = false;
self.query("RESET ROLE").await?;
self.cleanup_state.reset();
}
Ok(())
@@ -653,30 +1066,29 @@ impl Server {
self.application_name = name.to_string();
// We don't want `SET application_name` to mark the server connection
// as needing cleanup
let needs_cleanup_before = self.needs_cleanup;
let needs_cleanup_before = self.cleanup_state;
let result = Ok(self
.query(&format!("SET application_name = '{}'", name))
.await?);
self.needs_cleanup = needs_cleanup_before;
self.cleanup_state = needs_cleanup_before;
result
} else {
Ok(())
}
}
/// get Server stats
pub fn stats(&self) -> Arc<ServerStats> {
self.stats.clone()
}
/// Get the servers address.
#[allow(dead_code)]
pub fn address(&self) -> Address {
self.address.clone()
}
/// Get the server connection identifier
/// Used to uniquely identify connection in statistics
pub fn server_id(&self) -> i32 {
self.server_id
}
// Get server's latest response timestamp
pub fn last_activity(&self) -> SystemTime {
self.last_activity
@@ -684,7 +1096,7 @@ impl Server {
// Marks a connection as needing DISCARD ALL at checkin
pub fn mark_dirty(&mut self) {
self.needs_cleanup = true;
self.cleanup_state.set_true();
}
pub fn mirror_send(&mut self, bytes: &BytesMut) {
@@ -700,6 +1112,106 @@ impl Server {
None => (),
}
}
// This is so we can execute out of band queries to the server.
// The connection will be opened, the query executed and closed.
pub async fn exec_simple_query(
address: &Address,
user: &User,
query: &str,
) -> Result<Vec<String>, Error> {
let client_server_map: ClientServerMap = Arc::new(Mutex::new(HashMap::new()));
debug!("Connecting to server to obtain auth hashes.");
let mut server = Server::startup(
address,
user,
&address.database,
client_server_map,
Arc::new(ServerStats::default()),
Arc::new(RwLock::new(None)),
true,
)
.await?;
debug!("Connected!, sending query.");
server.send(&simple_query(query)).await?;
let mut message = server.recv().await?;
Ok(parse_query_message(&mut message).await?)
}
}
async fn parse_query_message(message: &mut BytesMut) -> Result<Vec<String>, Error> {
let mut pair = Vec::<String>::new();
match message::backend::Message::parse(message) {
Ok(Some(message::backend::Message::RowDescription(_description))) => {}
Ok(Some(message::backend::Message::ErrorResponse(err))) => {
return Err(Error::ProtocolSyncError(format!(
"Protocol error parsing response. Err: {:?}",
err.fields()
.iterator()
.fold(String::default(), |acc, element| acc
+ element.unwrap().value())
)))
}
Ok(_) => {
return Err(Error::ProtocolSyncError(
"Protocol error, expected Row Description.".to_string(),
))
}
Err(err) => {
return Err(Error::ProtocolSyncError(format!(
"Protocol error parsing response. Err: {:?}",
err
)))
}
}
while !message.is_empty() {
match message::backend::Message::parse(message) {
Ok(postgres_message) => {
match postgres_message {
Some(message::backend::Message::DataRow(data)) => {
let buf = data.buffer();
trace!("Data: {:?}", buf);
for item in data.ranges().iterator() {
match item.as_ref() {
Ok(range) => match range {
Some(range) => {
pair.push(String::from_utf8_lossy(&buf[range.clone()]).to_string());
}
None => return Err(Error::ProtocolSyncError(String::from(
"Data expected while receiving query auth data, found nothing.",
))),
},
Err(err) => {
return Err(Error::ProtocolSyncError(format!(
"Data error, err: {:?}",
err
)))
}
}
}
}
Some(message::backend::Message::CommandComplete(_)) => {}
Some(message::backend::Message::ReadyForQuery(_)) => {}
_ => {
return Err(Error::ProtocolSyncError(
"Unexpected message while receiving auth query data.".to_string(),
))
}
}
}
Err(err) => {
return Err(Error::ProtocolSyncError(format!(
"Parse error, err: {:?}",
err
)))
}
};
}
Ok(pair)
}
impl Drop for Server {
@@ -708,25 +1220,31 @@ impl Drop for Server {
/// for a write.
fn drop(&mut self) {
self.mirror_disconnect();
self.stats.server_disconnecting(self.server_id);
let mut bytes = BytesMut::with_capacity(4);
// Update statistics
self.stats.disconnect();
let mut bytes = BytesMut::with_capacity(5);
bytes.put_u8(b'X');
bytes.put_i32(4);
match self.write.try_write(&bytes) {
Ok(_) => (),
Err(_) => debug!("Dirty shutdown"),
match self.stream.get_mut().try_write(&bytes) {
Ok(5) => (),
_ => debug!("Dirty shutdown"),
};
// Should not matter.
self.bad = true;
let now = chrono::offset::Utc::now().naive_utc();
let duration = now - self.connected_at;
let message = if self.bad {
"Server connection terminated"
} else {
"Server connection closed"
};
info!(
"Server connection closed {:?}, session duration: {}",
"{} {:?}, session duration: {}",
message,
self.address,
crate::format_duration(&duration)
);

File diff suppressed because it is too large Load Diff

226
src/stats/address.rs Normal file
View File

@@ -0,0 +1,226 @@
use std::sync::atomic::*;
use std::sync::Arc;
#[derive(Debug, Clone, Default)]
struct AddressStatFields {
xact_count: Arc<AtomicU64>,
query_count: Arc<AtomicU64>,
bytes_received: Arc<AtomicU64>,
bytes_sent: Arc<AtomicU64>,
xact_time: Arc<AtomicU64>,
query_time: Arc<AtomicU64>,
wait_time: Arc<AtomicU64>,
errors: Arc<AtomicU64>,
}
/// Internal address stats
#[derive(Debug, Clone, Default)]
pub struct AddressStats {
total: AddressStatFields,
current: AddressStatFields,
averages: AddressStatFields,
// Determines if the averages have been updated since the last time they were reported
pub averages_updated: Arc<AtomicBool>,
}
impl IntoIterator for AddressStats {
type Item = (String, u64);
type IntoIter = std::vec::IntoIter<Self::Item>;
fn into_iter(self) -> Self::IntoIter {
vec![
(
"total_xact_count".to_string(),
self.total.xact_count.load(Ordering::Relaxed),
),
(
"total_query_count".to_string(),
self.total.query_count.load(Ordering::Relaxed),
),
(
"total_received".to_string(),
self.total.bytes_received.load(Ordering::Relaxed),
),
(
"total_sent".to_string(),
self.total.bytes_sent.load(Ordering::Relaxed),
),
(
"total_xact_time".to_string(),
self.total.xact_time.load(Ordering::Relaxed),
),
(
"total_query_time".to_string(),
self.total.query_time.load(Ordering::Relaxed),
),
(
"total_wait_time".to_string(),
self.total.wait_time.load(Ordering::Relaxed),
),
(
"total_errors".to_string(),
self.total.errors.load(Ordering::Relaxed),
),
(
"avg_xact_count".to_string(),
self.averages.xact_count.load(Ordering::Relaxed),
),
(
"avg_query_count".to_string(),
self.averages.query_count.load(Ordering::Relaxed),
),
(
"avg_recv".to_string(),
self.averages.bytes_received.load(Ordering::Relaxed),
),
(
"avg_sent".to_string(),
self.averages.bytes_sent.load(Ordering::Relaxed),
),
(
"avg_errors".to_string(),
self.averages.errors.load(Ordering::Relaxed),
),
(
"avg_xact_time".to_string(),
self.averages.xact_time.load(Ordering::Relaxed),
),
(
"avg_query_time".to_string(),
self.averages.query_time.load(Ordering::Relaxed),
),
(
"avg_wait_time".to_string(),
self.averages.wait_time.load(Ordering::Relaxed),
),
]
.into_iter()
}
}
impl AddressStats {
pub fn xact_count_add(&self) {
self.total.xact_count.fetch_add(1, Ordering::Relaxed);
self.current.xact_count.fetch_add(1, Ordering::Relaxed);
}
pub fn query_count_add(&self) {
self.total.query_count.fetch_add(1, Ordering::Relaxed);
self.current.query_count.fetch_add(1, Ordering::Relaxed);
}
pub fn bytes_received_add(&self, bytes: u64) {
self.total
.bytes_received
.fetch_add(bytes, Ordering::Relaxed);
self.current
.bytes_received
.fetch_add(bytes, Ordering::Relaxed);
}
pub fn bytes_sent_add(&self, bytes: u64) {
self.total.bytes_sent.fetch_add(bytes, Ordering::Relaxed);
self.current.bytes_sent.fetch_add(bytes, Ordering::Relaxed);
}
pub fn xact_time_add(&self, time: u64) {
self.total.xact_time.fetch_add(time, Ordering::Relaxed);
self.current.xact_time.fetch_add(time, Ordering::Relaxed);
}
pub fn query_time_add(&self, time: u64) {
self.total.query_time.fetch_add(time, Ordering::Relaxed);
self.current.query_time.fetch_add(time, Ordering::Relaxed);
}
pub fn wait_time_add(&self, time: u64) {
self.total.wait_time.fetch_add(time, Ordering::Relaxed);
self.current.wait_time.fetch_add(time, Ordering::Relaxed);
}
pub fn error(&self) {
self.total.errors.fetch_add(1, Ordering::Relaxed);
self.current.errors.fetch_add(1, Ordering::Relaxed);
}
pub fn update_averages(&self) {
let stat_period_per_second = crate::stats::STAT_PERIOD / 1_000;
// xact_count
let current_xact_count = self.current.xact_count.load(Ordering::Relaxed);
let current_xact_time = self.current.xact_time.load(Ordering::Relaxed);
self.averages.xact_count.store(
current_xact_count / stat_period_per_second,
Ordering::Relaxed,
);
if current_xact_count == 0 {
self.averages.xact_time.store(0, Ordering::Relaxed);
} else {
self.averages
.xact_time
.store(current_xact_time / current_xact_count, Ordering::Relaxed);
}
// query_count
let current_query_count = self.current.query_count.load(Ordering::Relaxed);
let current_query_time = self.current.query_time.load(Ordering::Relaxed);
self.averages.query_count.store(
current_query_count / stat_period_per_second,
Ordering::Relaxed,
);
if current_query_count == 0 {
self.averages.query_time.store(0, Ordering::Relaxed);
} else {
self.averages
.query_time
.store(current_query_time / current_query_count, Ordering::Relaxed);
}
// bytes_received
let current_bytes_received = self.current.bytes_received.load(Ordering::Relaxed);
self.averages.bytes_received.store(
current_bytes_received / stat_period_per_second,
Ordering::Relaxed,
);
// bytes_sent
let current_bytes_sent = self.current.bytes_sent.load(Ordering::Relaxed);
self.averages.bytes_sent.store(
current_bytes_sent / stat_period_per_second,
Ordering::Relaxed,
);
// wait_time
let current_wait_time = self.current.wait_time.load(Ordering::Relaxed);
self.averages.wait_time.store(
current_wait_time / stat_period_per_second,
Ordering::Relaxed,
);
// errors
let current_errors = self.current.errors.load(Ordering::Relaxed);
self.averages
.errors
.store(current_errors / stat_period_per_second, Ordering::Relaxed);
}
pub fn reset_current_counts(&self) {
self.current.xact_count.store(0, Ordering::Relaxed);
self.current.xact_time.store(0, Ordering::Relaxed);
self.current.query_count.store(0, Ordering::Relaxed);
self.current.query_time.store(0, Ordering::Relaxed);
self.current.bytes_received.store(0, Ordering::Relaxed);
self.current.bytes_sent.store(0, Ordering::Relaxed);
self.current.wait_time.store(0, Ordering::Relaxed);
self.current.errors.store(0, Ordering::Relaxed);
}
pub fn populate_row(&self, row: &mut Vec<String>) {
for (_key, value) in self.clone() {
row.push(value.to_string());
}
}
}

174
src/stats/client.rs Normal file
View File

@@ -0,0 +1,174 @@
use super::{get_reporter, Reporter};
use atomic_enum::atomic_enum;
use std::sync::atomic::*;
use std::sync::Arc;
use tokio::time::Instant;
/// The various states that a client can be in
#[atomic_enum]
#[derive(PartialEq)]
pub enum ClientState {
Idle = 0,
Waiting,
Active,
}
impl std::fmt::Display for ClientState {
fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
match *self {
ClientState::Idle => write!(f, "idle"),
ClientState::Waiting => write!(f, "waiting"),
ClientState::Active => write!(f, "active"),
}
}
}
#[derive(Debug, Clone)]
/// Information we keep track of which can be queried by SHOW CLIENTS
pub struct ClientStats {
/// A random integer assigned to the client and used by stats to track the client
client_id: i32,
/// Data associated with the client, not writable, only set when we construct the ClientStat
application_name: String,
username: String,
pool_name: String,
connect_time: Instant,
reporter: Reporter,
/// Total time spent waiting for a connection from pool, measures in microseconds
pub total_wait_time: Arc<AtomicU64>,
/// Maximum time spent waiting for a connection from pool, measures in microseconds
pub max_wait_time: Arc<AtomicU64>,
/// Current state of the client
pub state: Arc<AtomicClientState>,
/// Number of transactions executed by this client
pub transaction_count: Arc<AtomicU64>,
/// Number of queries executed by this client
pub query_count: Arc<AtomicU64>,
/// Number of errors made by this client
pub error_count: Arc<AtomicU64>,
}
impl Default for ClientStats {
fn default() -> Self {
ClientStats {
client_id: 0,
connect_time: Instant::now(),
application_name: String::new(),
username: String::new(),
pool_name: String::new(),
total_wait_time: Arc::new(AtomicU64::new(0)),
max_wait_time: Arc::new(AtomicU64::new(0)),
state: Arc::new(AtomicClientState::new(ClientState::Idle)),
transaction_count: Arc::new(AtomicU64::new(0)),
query_count: Arc::new(AtomicU64::new(0)),
error_count: Arc::new(AtomicU64::new(0)),
reporter: get_reporter(),
}
}
}
impl ClientStats {
pub fn new(
client_id: i32,
application_name: &str,
username: &str,
pool_name: &str,
connect_time: Instant,
) -> Self {
Self {
client_id,
connect_time,
application_name: application_name.to_string(),
username: username.to_string(),
pool_name: pool_name.to_string(),
..Default::default()
}
}
/// Reports a client is disconnecting from the pooler and
/// update metrics on the corresponding pool.
pub fn disconnect(&self) {
self.reporter.client_disconnecting(self.client_id);
}
/// Register a client with the stats system. The stats system uses client_id
/// to track and aggregate statistics from all source that relate to that client
pub fn register(&self, stats: Arc<ClientStats>) {
self.reporter.client_register(self.client_id, stats);
self.state.store(ClientState::Idle, Ordering::Relaxed);
}
/// Reports a client is done querying the server and is no longer assigned a server connection
pub fn idle(&self) {
self.state.store(ClientState::Idle, Ordering::Relaxed);
}
/// Reports a client is waiting for a connection
pub fn waiting(&self) {
self.state.store(ClientState::Waiting, Ordering::Relaxed);
}
/// Reports a client is done waiting for a connection and is about to query the server.
pub fn active(&self) {
self.state.store(ClientState::Active, Ordering::Relaxed);
}
/// Reports a client has failed to obtain a connection from a connection pool
pub fn checkout_error(&self) {
self.state.store(ClientState::Idle, Ordering::Relaxed);
}
/// Reports a client has had the server assigned to it be banned
pub fn ban_error(&self) {
self.state.store(ClientState::Idle, Ordering::Relaxed);
self.error_count.fetch_add(1, Ordering::Relaxed);
}
/// Reporters the time spent by a client waiting to get a healthy connection from the pool
pub fn checkout_time(&self, microseconds: u64) {
self.total_wait_time
.fetch_add(microseconds, Ordering::Relaxed);
self.max_wait_time
.fetch_max(microseconds, Ordering::Relaxed);
}
/// Report a query executed by a client against a server
pub fn query(&self) {
self.query_count.fetch_add(1, Ordering::Relaxed);
}
/// Report a transaction executed by a client a server
/// we report each individual queries outside a transaction as a transaction
/// We only count the initial BEGIN as a transaction, all queries within do not
/// count as transactions
pub fn transaction(&self) {
self.transaction_count.fetch_add(1, Ordering::Relaxed);
}
// Helper methods for show clients
pub fn connect_time(&self) -> Instant {
self.connect_time
}
pub fn client_id(&self) -> i32 {
self.client_id
}
pub fn application_name(&self) -> String {
self.application_name.clone()
}
pub fn username(&self) -> String {
self.username.clone()
}
pub fn pool_name(&self) -> String {
self.pool_name.clone()
}
}

151
src/stats/pool.rs Normal file
View File

@@ -0,0 +1,151 @@
use log::debug;
use super::{ClientState, ServerState};
use crate::{config::PoolMode, messages::DataType, pool::PoolIdentifier};
use std::collections::HashMap;
use std::sync::atomic::*;
use crate::pool::get_all_pools;
#[derive(Debug, Clone)]
/// A struct that holds information about a Pool .
pub struct PoolStats {
pub identifier: PoolIdentifier,
pub mode: PoolMode,
pub cl_idle: u64,
pub cl_active: u64,
pub cl_waiting: u64,
pub cl_cancel_req: u64,
pub sv_active: u64,
pub sv_idle: u64,
pub sv_used: u64,
pub sv_tested: u64,
pub sv_login: u64,
pub maxwait: u64,
}
impl PoolStats {
pub fn new(identifier: PoolIdentifier, mode: PoolMode) -> Self {
PoolStats {
identifier,
mode,
cl_idle: 0,
cl_active: 0,
cl_waiting: 0,
cl_cancel_req: 0,
sv_active: 0,
sv_idle: 0,
sv_used: 0,
sv_tested: 0,
sv_login: 0,
maxwait: 0,
}
}
pub fn construct_pool_lookup() -> HashMap<PoolIdentifier, PoolStats> {
let mut map: HashMap<PoolIdentifier, PoolStats> = HashMap::new();
let client_map = super::get_client_stats();
let server_map = super::get_server_stats();
for (identifier, pool) in get_all_pools() {
map.insert(
identifier.clone(),
PoolStats::new(identifier, pool.settings.pool_mode),
);
}
for client in client_map.values() {
match map.get_mut(&PoolIdentifier {
db: client.pool_name(),
user: client.username(),
}) {
Some(pool_stats) => {
match client.state.load(Ordering::Relaxed) {
ClientState::Active => pool_stats.cl_active += 1,
ClientState::Idle => pool_stats.cl_idle += 1,
ClientState::Waiting => pool_stats.cl_waiting += 1,
}
let max_wait = client.max_wait_time.load(Ordering::Relaxed);
pool_stats.maxwait = std::cmp::max(pool_stats.maxwait, max_wait);
}
None => debug!("Client from an obselete pool"),
}
}
for server in server_map.values() {
match map.get_mut(&PoolIdentifier {
db: server.pool_name(),
user: server.username(),
}) {
Some(pool_stats) => match server.state.load(Ordering::Relaxed) {
ServerState::Active => pool_stats.sv_active += 1,
ServerState::Idle => pool_stats.sv_idle += 1,
ServerState::Login => pool_stats.sv_login += 1,
ServerState::Tested => pool_stats.sv_tested += 1,
},
None => debug!("Server from an obselete pool"),
}
}
return map;
}
pub fn generate_header() -> Vec<(&'static str, DataType)> {
return vec![
("database", DataType::Text),
("user", DataType::Text),
("pool_mode", DataType::Text),
("cl_idle", DataType::Numeric),
("cl_active", DataType::Numeric),
("cl_waiting", DataType::Numeric),
("cl_cancel_req", DataType::Numeric),
("sv_active", DataType::Numeric),
("sv_idle", DataType::Numeric),
("sv_used", DataType::Numeric),
("sv_tested", DataType::Numeric),
("sv_login", DataType::Numeric),
("maxwait", DataType::Numeric),
("maxwait_us", DataType::Numeric),
];
}
pub fn generate_row(&self) -> Vec<String> {
return vec![
self.identifier.db.clone(),
self.identifier.user.clone(),
self.mode.to_string(),
self.cl_idle.to_string(),
self.cl_active.to_string(),
self.cl_waiting.to_string(),
self.cl_cancel_req.to_string(),
self.sv_active.to_string(),
self.sv_idle.to_string(),
self.sv_used.to_string(),
self.sv_tested.to_string(),
self.sv_login.to_string(),
(self.maxwait / 1_000_000).to_string(),
(self.maxwait % 1_000_000).to_string(),
];
}
}
impl IntoIterator for PoolStats {
type Item = (String, u64);
type IntoIter = std::vec::IntoIter<Self::Item>;
fn into_iter(self) -> Self::IntoIter {
vec![
("cl_idle".to_string(), self.cl_idle),
("cl_active".to_string(), self.cl_active),
("cl_waiting".to_string(), self.cl_waiting),
("cl_cancel_req".to_string(), self.cl_cancel_req),
("sv_active".to_string(), self.sv_active),
("sv_idle".to_string(), self.sv_idle),
("sv_used".to_string(), self.sv_used),
("sv_tested".to_string(), self.sv_tested),
("sv_login".to_string(), self.sv_login),
("maxwait".to_string(), self.maxwait / 1_000_000),
("maxwait_us".to_string(), self.maxwait % 1_000_000),
]
.into_iter()
}
}

216
src/stats/server.rs Normal file
View File

@@ -0,0 +1,216 @@
use super::AddressStats;
use super::{get_reporter, Reporter};
use crate::config::Address;
use atomic_enum::atomic_enum;
use parking_lot::RwLock;
use std::sync::atomic::*;
use std::sync::Arc;
use tokio::time::Instant;
/// The various states that a server can be in
#[atomic_enum]
#[derive(PartialEq)]
pub enum ServerState {
Login = 0,
Active,
Tested,
Idle,
}
impl std::fmt::Display for ServerState {
fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
match *self {
ServerState::Login => write!(f, "login"),
ServerState::Active => write!(f, "active"),
ServerState::Tested => write!(f, "tested"),
ServerState::Idle => write!(f, "idle"),
}
}
}
/// Information we keep track of which can be queried by SHOW SERVERS
#[derive(Debug, Clone)]
pub struct ServerStats {
/// A random integer assigned to the server and used by stats to track the server
server_id: i32,
/// Context information, only to be read
address: Address,
connect_time: Instant,
reporter: Reporter,
/// Data
pub application_name: Arc<RwLock<String>>,
pub state: Arc<AtomicServerState>,
pub bytes_sent: Arc<AtomicU64>,
pub bytes_received: Arc<AtomicU64>,
pub transaction_count: Arc<AtomicU64>,
pub query_count: Arc<AtomicU64>,
pub error_count: Arc<AtomicU64>,
pub prepared_hit_count: Arc<AtomicU64>,
pub prepared_miss_count: Arc<AtomicU64>,
}
impl Default for ServerStats {
fn default() -> Self {
ServerStats {
server_id: 0,
application_name: Arc::new(RwLock::new(String::new())),
address: Address::default(),
connect_time: Instant::now(),
state: Arc::new(AtomicServerState::new(ServerState::Login)),
bytes_sent: Arc::new(AtomicU64::new(0)),
bytes_received: Arc::new(AtomicU64::new(0)),
transaction_count: Arc::new(AtomicU64::new(0)),
query_count: Arc::new(AtomicU64::new(0)),
error_count: Arc::new(AtomicU64::new(0)),
reporter: get_reporter(),
prepared_hit_count: Arc::new(AtomicU64::new(0)),
prepared_miss_count: Arc::new(AtomicU64::new(0)),
}
}
}
impl ServerStats {
pub fn new(address: Address, connect_time: Instant) -> Self {
Self {
address,
connect_time,
server_id: rand::random::<i32>(),
..Default::default()
}
}
pub fn server_id(&self) -> i32 {
self.server_id
}
/// Register a server connection with the stats system. The stats system uses server_id
/// to track and aggregate statistics from all source that relate to that server
// Delegates to reporter
pub fn register(&self, stats: Arc<ServerStats>) {
self.reporter.server_register(self.server_id, stats);
self.login();
}
/// Reports a server connection is no longer assigned to a client
/// and is available for the next client to pick it up
pub fn idle(&self) {
self.state.store(ServerState::Idle, Ordering::Relaxed);
}
/// Reports a server connection is disconnecting from the pooler.
/// Also updates metrics on the pool regarding server usage.
pub fn disconnect(&self) {
self.reporter.server_disconnecting(self.server_id);
}
/// Reports a server connection is being tested before being given to a client.
pub fn tested(&self) {
self.set_undefined_application();
self.state.store(ServerState::Tested, Ordering::Relaxed);
}
/// Reports a server connection is attempting to login.
pub fn login(&self) {
self.state.store(ServerState::Login, Ordering::Relaxed);
self.set_undefined_application();
}
/// Reports a server connection has been assigned to a client that
/// is about to query the server
pub fn active(&self, application_name: String) {
self.state.store(ServerState::Active, Ordering::Relaxed);
self.set_application(application_name);
}
pub fn address_stats(&self) -> Arc<AddressStats> {
self.address.stats.clone()
}
pub fn check_address_stat_average_is_updated_status(&self) -> bool {
self.address.stats.averages_updated.load(Ordering::Relaxed)
}
pub fn set_address_stat_average_is_updated_status(&self, is_checked: bool) {
self.address
.stats
.averages_updated
.store(is_checked, Ordering::Relaxed);
}
// Helper methods for show_servers
pub fn pool_name(&self) -> String {
self.address.pool_name.clone()
}
pub fn username(&self) -> String {
self.address.username.clone()
}
pub fn address_name(&self) -> String {
self.address.name()
}
pub fn connect_time(&self) -> Instant {
self.connect_time
}
fn set_application(&self, name: String) {
let mut application_name = self.application_name.write();
*application_name = name;
}
fn set_undefined_application(&self) {
self.set_application(String::from("Undefined"))
}
pub fn checkout_time(&self, microseconds: u64, application_name: String) {
// Update server stats and address aggregation stats
self.set_application(application_name);
self.address.stats.wait_time_add(microseconds);
}
/// Report a query executed by a client against a server
pub fn query(&self, milliseconds: u64, application_name: &str) {
self.set_application(application_name.to_string());
self.address.stats.query_count_add();
self.address.stats.query_time_add(milliseconds);
self.query_count.fetch_add(1, Ordering::Relaxed);
}
/// Report a transaction executed by a client a server
/// we report each individual queries outside a transaction as a transaction
/// We only count the initial BEGIN as a transaction, all queries within do not
/// count as transactions
pub fn transaction(&self, application_name: &str) {
self.set_application(application_name.to_string());
self.transaction_count.fetch_add(1, Ordering::Relaxed);
self.address.stats.xact_count_add();
}
/// Report data sent to a server
pub fn data_sent(&self, amount_bytes: usize) {
self.bytes_sent
.fetch_add(amount_bytes as u64, Ordering::Relaxed);
self.address.stats.bytes_sent_add(amount_bytes as u64);
}
/// Report data received from a server
pub fn data_received(&self, amount_bytes: usize) {
self.bytes_received
.fetch_add(amount_bytes as u64, Ordering::Relaxed);
self.address.stats.bytes_received_add(amount_bytes as u64);
}
/// Report a prepared statement that already exists on the server.
pub fn prepared_cache_hit(&self) {
self.prepared_hit_count.fetch_add(1, Ordering::Relaxed);
}
/// Report a prepared statement that does not exist on the server yet.
pub fn prepared_cache_miss(&self) {
self.prepared_miss_count.fetch_add(1, Ordering::Relaxed);
}
}

View File

@@ -4,7 +4,12 @@ use rustls_pemfile::{certs, read_one, Item};
use std::iter;
use std::path::Path;
use std::sync::Arc;
use tokio_rustls::rustls::{self, Certificate, PrivateKey};
use std::time::SystemTime;
use tokio_rustls::rustls::{
self,
client::{ServerCertVerified, ServerCertVerifier},
Certificate, PrivateKey, ServerName,
};
use tokio_rustls::TlsAcceptor;
use crate::config::get_config;
@@ -64,3 +69,19 @@ impl Tls {
})
}
}
pub struct NoCertificateVerification;
impl ServerCertVerifier for NoCertificateVerification {
fn verify_server_cert(
&self,
_end_entity: &Certificate,
_intermediates: &[Certificate],
_server_name: &ServerName,
_scts: &mut dyn Iterator<Item = &[u8]>,
_ocsp_response: &[u8],
_now: SystemTime,
) -> Result<ServerCertVerified, rustls::Error> {
Ok(ServerCertVerified::assertion())
}
}

View File

@@ -36,6 +36,15 @@ services:
POSTGRES_PASSWORD: postgres
POSTGRES_INITDB_ARGS: --auth-local=scram-sha-256 --auth-host=scram-sha-256 --auth=scram-sha-256
command: ["postgres", "-p", "9432", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-c", "pg_stat_statements.max=100000"]
pg5:
image: postgres:14
network_mode: "service:main"
environment:
POSTGRES_USER: postgres
POSTGRES_DB: postgres
POSTGRES_PASSWORD: postgres
POSTGRES_INITDB_ARGS: --auth-local=md5 --auth-host=md5 --auth=md5
command: ["postgres", "-c", "shared_preload_libraries=pg_stat_statements", "-c", "pg_stat_statements.track=all", "-p", "10432"]
main:
build: .
command: ["bash", "/app/tests/docker/run.sh"]

View File

@@ -63,6 +63,7 @@ def cleanup_conn(conn: psycopg2.extensions.connection, cur: psycopg2.extensions.
def test_normal_db_access():
pgcat_start()
conn, cur = connect_db(autocommit=False)
cur.execute("SELECT 1")
res = cur.fetchall()

View File

@@ -11,282 +11,6 @@ describe "Admin" do
processes.pgcat.shutdown
end
describe "SHOW STATS" do
context "clients connect and make one query" do
it "updates *_query_time and *_wait_time" do
connection = PG::connect("#{pgcat_conn_str}?application_name=one_query")
connection.async_exec("SELECT pg_sleep(0.25)")
connection.async_exec("SELECT pg_sleep(0.25)")
connection.async_exec("SELECT pg_sleep(0.25)")
connection.close
# wait for averages to be calculated, we shouldn't do this too often
sleep(15.5)
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
results = admin_conn.async_exec("SHOW STATS")[0]
admin_conn.close
expect(results["total_query_time"].to_i).to be_within(200).of(750)
expect(results["avg_query_time"].to_i).to_not eq(0)
expect(results["total_wait_time"].to_i).to_not eq(0)
expect(results["avg_wait_time"].to_i).to_not eq(0)
end
end
end
describe "SHOW POOLS" do
context "bad credentials" do
it "does not change any stats" do
bad_passsword_url = URI(pgcat_conn_str)
bad_passsword_url.password = "wrong"
expect { PG::connect("#{bad_passsword_url.to_s}?application_name=bad_password") }.to raise_error(PG::ConnectionBad)
sleep(1)
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
results = admin_conn.async_exec("SHOW POOLS")[0]
%w[cl_idle cl_active cl_waiting cl_cancel_req sv_active sv_used sv_tested sv_login maxwait].each do |s|
raise StandardError, "Field #{s} was expected to be 0 but found to be #{results[s]}" if results[s] != "0"
end
expect(results["sv_idle"]).to eq("1")
end
end
context "bad database name" do
it "does not change any stats" do
bad_db_url = URI(pgcat_conn_str)
bad_db_url.path = "/wrong_db"
expect { PG::connect("#{bad_db_url.to_s}?application_name=bad_db") }.to raise_error(PG::ConnectionBad)
sleep(1)
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
results = admin_conn.async_exec("SHOW POOLS")[0]
%w[cl_idle cl_active cl_waiting cl_cancel_req sv_active sv_used sv_tested sv_login maxwait].each do |s|
raise StandardError, "Field #{s} was expected to be 0 but found to be #{results[s]}" if results[s] != "0"
end
expect(results["sv_idle"]).to eq("1")
end
end
context "client connects but issues no queries" do
it "only affects cl_idle stats" do
connections = Array.new(20) { PG::connect(pgcat_conn_str) }
sleep(1)
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
results = admin_conn.async_exec("SHOW POOLS")[0]
%w[cl_active cl_waiting cl_cancel_req sv_active sv_used sv_tested sv_login maxwait].each do |s|
raise StandardError, "Field #{s} was expected to be 0 but found to be #{results[s]}" if results[s] != "0"
end
expect(results["cl_idle"]).to eq("20")
expect(results["sv_idle"]).to eq("1")
connections.map(&:close)
sleep(1.1)
results = admin_conn.async_exec("SHOW POOLS")[0]
%w[cl_active cl_idle cl_waiting cl_cancel_req sv_active sv_used sv_tested sv_login maxwait].each do |s|
raise StandardError, "Field #{s} was expected to be 0 but found to be #{results[s]}" if results[s] != "0"
end
expect(results["sv_idle"]).to eq("1")
end
end
context "clients connect and make one query" do
it "only affects cl_idle, sv_idle stats" do
connections = Array.new(5) { PG::connect("#{pgcat_conn_str}?application_name=one_query") }
connections.each do |c|
Thread.new { c.async_exec("SELECT pg_sleep(2.5)") }
end
sleep(1.1)
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
results = admin_conn.async_exec("SHOW POOLS")[0]
%w[cl_idle cl_waiting cl_cancel_req sv_idle sv_used sv_tested sv_login maxwait].each do |s|
raise StandardError, "Field #{s} was expected to be 0 but found to be #{results[s]}" if results[s] != "0"
end
expect(results["cl_active"]).to eq("5")
expect(results["sv_active"]).to eq("5")
sleep(3)
results = admin_conn.async_exec("SHOW POOLS")[0]
%w[cl_active cl_waiting cl_cancel_req sv_active sv_used sv_tested sv_login maxwait].each do |s|
raise StandardError, "Field #{s} was expected to be 0 but found to be #{results[s]}" if results[s] != "0"
end
expect(results["cl_idle"]).to eq("5")
expect(results["sv_idle"]).to eq("5")
connections.map(&:close)
sleep(1)
results = admin_conn.async_exec("SHOW POOLS")[0]
%w[cl_idle cl_active cl_waiting cl_cancel_req sv_active sv_used sv_tested sv_login maxwait].each do |s|
raise StandardError, "Field #{s} was expected to be 0 but found to be #{results[s]}" if results[s] != "0"
end
expect(results["sv_idle"]).to eq("5")
end
end
context "client connects and opens a transaction and closes connection uncleanly" do
it "produces correct statistics" do
connections = Array.new(5) { PG::connect("#{pgcat_conn_str}?application_name=one_query") }
connections.each do |c|
Thread.new do
c.async_exec("BEGIN")
c.async_exec("SELECT pg_sleep(0.01)")
c.close
end
end
sleep(1.1)
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
results = admin_conn.async_exec("SHOW POOLS")[0]
%w[cl_idle cl_active cl_waiting cl_cancel_req sv_active sv_used sv_tested sv_login maxwait].each do |s|
raise StandardError, "Field #{s} was expected to be 0 but found to be #{results[s]}" if results[s] != "0"
end
expect(results["sv_idle"]).to eq("5")
end
end
context "client fail to checkout connection from the pool" do
it "counts clients as idle" do
new_configs = processes.pgcat.current_config
new_configs["general"]["connect_timeout"] = 500
new_configs["general"]["ban_time"] = 1
new_configs["general"]["shutdown_timeout"] = 1
new_configs["pools"]["sharded_db"]["users"]["0"]["pool_size"] = 1
processes.pgcat.update_config(new_configs)
processes.pgcat.reload_config
threads = []
connections = Array.new(5) { PG::connect("#{pgcat_conn_str}?application_name=one_query") }
connections.each do |c|
threads << Thread.new { c.async_exec("SELECT pg_sleep(1)") rescue PG::SystemError }
end
sleep(2)
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
results = admin_conn.async_exec("SHOW POOLS")[0]
%w[cl_active cl_waiting cl_cancel_req sv_active sv_used sv_tested sv_login maxwait].each do |s|
raise StandardError, "Field #{s} was expected to be 0 but found to be #{results[s]}" if results[s] != "0"
end
expect(results["cl_idle"]).to eq("5")
expect(results["sv_idle"]).to eq("1")
threads.map(&:join)
connections.map(&:close)
end
end
context "clients overwhelm server pools" do
let(:processes) { Helpers::Pgcat.single_instance_setup("sharded_db", 2) }
it "cl_waiting is updated to show it" do
threads = []
connections = Array.new(4) { PG::connect("#{pgcat_conn_str}?application_name=one_query") }
connections.each do |c|
threads << Thread.new { c.async_exec("SELECT pg_sleep(1.5)") }
end
sleep(1.1) # Allow time for stats to update
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
results = admin_conn.async_exec("SHOW POOLS")[0]
%w[cl_idle cl_cancel_req sv_idle sv_used sv_tested sv_login maxwait].each do |s|
raise StandardError, "Field #{s} was expected to be 0 but found to be #{results[s]}" if results[s] != "0"
end
expect(results["cl_waiting"]).to eq("2")
expect(results["cl_active"]).to eq("2")
expect(results["sv_active"]).to eq("2")
sleep(2.5) # Allow time for stats to update
results = admin_conn.async_exec("SHOW POOLS")[0]
%w[cl_active cl_waiting cl_cancel_req sv_active sv_used sv_tested sv_login maxwait].each do |s|
raise StandardError, "Field #{s} was expected to be 0 but found to be #{results[s]}" if results[s] != "0"
end
expect(results["cl_idle"]).to eq("4")
expect(results["sv_idle"]).to eq("2")
threads.map(&:join)
connections.map(&:close)
end
it "show correct max_wait" do
threads = []
connections = Array.new(4) { PG::connect("#{pgcat_conn_str}?application_name=one_query") }
connections.each do |c|
threads << Thread.new { c.async_exec("SELECT pg_sleep(1.5)") }
end
sleep(2.5) # Allow time for stats to update
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
results = admin_conn.async_exec("SHOW POOLS")[0]
expect(results["maxwait"]).to eq("1")
expect(results["maxwait_us"].to_i).to be_within(200_000).of(500_000)
sleep(4.5) # Allow time for stats to update
results = admin_conn.async_exec("SHOW POOLS")[0]
expect(results["maxwait"]).to eq("0")
threads.map(&:join)
connections.map(&:close)
end
end
end
describe "SHOW CLIENTS" do
it "reports correct number and application names" do
conn_str = processes.pgcat.connection_string("sharded_db", "sharding_user")
connections = Array.new(20) { |i| PG::connect("#{conn_str}?application_name=app#{i % 5}") }
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
sleep(1) # Wait for stats to be updated
results = admin_conn.async_exec("SHOW CLIENTS")
expect(results.count).to eq(21) # count admin clients
expect(results.select { |c| c["application_name"] == "app3" || c["application_name"] == "app4" }.count).to eq(8)
expect(results.select { |c| c["database"] == "pgcat" }.count).to eq(1)
connections[0..5].map(&:close)
sleep(1) # Wait for stats to be updated
results = admin_conn.async_exec("SHOW CLIENTS")
expect(results.count).to eq(15)
connections[6..].map(&:close)
sleep(1) # Wait for stats to be updated
expect(admin_conn.async_exec("SHOW CLIENTS").count).to eq(1)
admin_conn.close
end
it "reports correct number of queries and transactions" do
conn_str = processes.pgcat.connection_string("sharded_db", "sharding_user")
connections = Array.new(2) { |i| PG::connect("#{conn_str}?application_name=app#{i}") }
connections.each do |c|
c.async_exec("SELECT 1")
c.async_exec("SELECT 2")
c.async_exec("SELECT 3")
c.async_exec("BEGIN")
c.async_exec("SELECT 4")
c.async_exec("SELECT 5")
c.async_exec("COMMIT")
end
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
sleep(1) # Wait for stats to be updated
results = admin_conn.async_exec("SHOW CLIENTS")
expect(results.count).to eq(3)
normal_client_results = results.reject { |r| r["database"] == "pgcat" }
expect(normal_client_results[0]["transaction_count"]).to eq("4")
expect(normal_client_results[1]["transaction_count"]).to eq("4")
expect(normal_client_results[0]["query_count"]).to eq("7")
expect(normal_client_results[1]["query_count"]).to eq("7")
admin_conn.close
connections.map(&:close)
end
end
describe "Manual Banning" do
let(:processes) { Helpers::Pgcat.single_shard_setup("sharded_db", 10) }
before do
@@ -357,7 +81,7 @@ describe "Admin" do
end
end
describe "SHOW users" do
describe "SHOW USERS" do
it "returns the right users" do
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
results = admin_conn.async_exec("SHOW USERS")[0]

View File

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

BIN
tests/ruby/capture Normal file

Binary file not shown.

View File

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

View File

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

View File

@@ -2,6 +2,14 @@ require 'json'
require 'ostruct'
require_relative 'pgcat_process'
require_relative 'pg_instance'
require_relative 'pg_socket'
class ::Hash
def deep_merge(second)
merger = proc { |key, v1, v2| Hash === v1 && Hash === v2 ? v1.merge(v2, &merger) : v2 }
self.merge(second, &merger)
end
end
module Helpers
module Pgcat
@@ -33,7 +41,24 @@ module Helpers
"1" => { "database" => "shard1", "servers" => [["localhost", primary1.port.to_s, "primary"]] },
"2" => { "database" => "shard2", "servers" => [["localhost", primary2.port.to_s, "primary"]] },
},
"users" => { "0" => user }
"users" => { "0" => user },
"plugins" => {
"intercept" => {
"enabled" => true,
"queries" => {
"0" => {
"query" => "select current_database() as a, current_schemas(false) as b",
"schema" => [
["a", "text"],
["b", "text"],
],
"result" => [
["${DATABASE}", "{public}"],
]
}
}
}
}
}
}
pgcat.update_config(pgcat_cfg)
@@ -93,7 +118,7 @@ module Helpers
end
end
def self.single_shard_setup(pool_name, pool_size, pool_mode="transaction", lb_mode="random", log_level="info")
def self.single_shard_setup(pool_name, pool_size, pool_mode="transaction", lb_mode="random", log_level="info", pool_settings={})
user = {
"password" => "sharding_user",
"pool_size" => pool_size,
@@ -109,28 +134,32 @@ module Helpers
replica1 = PgInstance.new(8432, user["username"], user["password"], "shard0")
replica2 = PgInstance.new(9432, user["username"], user["password"], "shard0")
pool_config = {
"default_role" => "any",
"pool_mode" => pool_mode,
"load_balancing_mode" => lb_mode,
"primary_reads_enabled" => false,
"query_parser_enabled" => false,
"sharding_function" => "pg_bigint_hash",
"shards" => {
"0" => {
"database" => "shard0",
"servers" => [
["localhost", primary.port.to_s, "primary"],
["localhost", replica0.port.to_s, "replica"],
["localhost", replica1.port.to_s, "replica"],
["localhost", replica2.port.to_s, "replica"]
]
},
},
"users" => { "0" => user }
}
pool_config = pool_config.merge(pool_settings)
# Main proxy configs
pgcat_cfg["pools"] = {
"#{pool_name}" => {
"default_role" => "any",
"pool_mode" => pool_mode,
"load_balancing_mode" => lb_mode,
"primary_reads_enabled" => false,
"query_parser_enabled" => false,
"sharding_function" => "pg_bigint_hash",
"shards" => {
"0" => {
"database" => "shard0",
"servers" => [
["localhost", primary.port.to_s, "primary"],
["localhost", replica0.port.to_s, "replica"],
["localhost", replica1.port.to_s, "replica"],
["localhost", replica2.port.to_s, "replica"]
]
},
},
"users" => { "0" => user }
}
"#{pool_name}" => pool_config,
}
pgcat_cfg["general"]["port"] = pgcat.port
pgcat.update_config(pgcat_cfg)

View File

@@ -67,17 +67,21 @@ class PgcatProcess
def start
raise StandardError, "Process is already started" unless @pid.nil?
@pid = Process.spawn(@env, @command, err: @log_filename, out: @log_filename)
Process.detach(@pid)
ObjectSpace.define_finalizer(@log_filename, proc { PgcatProcess.finalize(@pid, @log_filename, @config_filename) })
return self
end
def wait_until_ready
def wait_until_ready(connection_string = nil)
exc = nil
10.times do
PG::connect(example_connection_string).close
Process.kill 0, @pid
PG::connect(connection_string || example_connection_string).close
return self
rescue Errno::ESRCH
raise StandardError, "Process #{@pid} died. #{logs}"
rescue => e
exc = e
sleep(0.5)
@@ -108,13 +112,10 @@ class PgcatProcess
"postgresql://#{username}:#{password}@0.0.0.0:#{@port}/pgcat"
end
def connection_string(pool_name, username)
def connection_string(pool_name, username, password = nil)
cfg = current_config
user_idx, user_obj = cfg["pools"][pool_name]["users"].detect { |k, user| user["username"] == username }
password = user_obj["password"]
"postgresql://#{username}:#{password}@0.0.0.0:#{@port}/#{pool_name}"
"postgresql://#{username}:#{password || user_obj["password"]}@0.0.0.0:#{@port}/#{pool_name}"
end
def example_connection_string

View File

@@ -65,7 +65,7 @@ describe "Least Outstanding Queries Load Balancing" do
processes.pgcat.shutdown
end
context "under homogenous load" do
context "under homogeneous load" do
it "balances query volume between all instances" do
conn = PG.connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))

View File

@@ -25,7 +25,7 @@ describe "Query Mirroing" do
processes.pgcat.shutdown
end
it "can mirror a query" do
xit "can mirror a query" do
conn = PG.connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
runs = 15
runs.times { conn.async_exec("SELECT 1 + 2") }

View File

@@ -241,6 +241,18 @@ describe "Miscellaneous" do
expect(processes.primary.count_query("DISCARD ALL")).to eq(10)
end
it "Resets server roles correctly" do
10.times do
conn = PG::connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
conn.async_exec("SET SERVER ROLE to 'primary'")
conn.async_exec("SELECT 1")
conn.async_exec("SET statement_timeout to 5000")
conn.close
end
expect(processes.primary.count_query("RESET ROLE")).to eq(10)
end
end
context "transaction mode" do
@@ -308,6 +320,31 @@ describe "Miscellaneous" do
expect(processes.primary.count_query("DISCARD ALL")).to eq(0)
end
end
context "server cleanup disabled" do
let(:processes) { Helpers::Pgcat.single_shard_setup("sharded_db", 1, "transaction", "random", "info", { "cleanup_server_connections" => false }) }
it "will not clean up connection state" do
conn = PG::connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
processes.primary.reset_stats
conn.async_exec("SET statement_timeout TO 1000")
conn.close
puts processes.pgcat.logs
expect(processes.primary.count_query("DISCARD ALL")).to eq(0)
end
it "will not clean up prepared statements" do
conn = PG::connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
processes.primary.reset_stats
conn.async_exec("PREPARE prepared_q (int) AS SELECT $1")
conn.close
puts processes.pgcat.logs
expect(processes.primary.count_query("DISCARD ALL")).to eq(0)
end
end
end
describe "Idle client timeout" do

View File

@@ -0,0 +1,14 @@
require_relative 'spec_helper'
describe "Plugins" do
let(:processes) { Helpers::Pgcat.three_shard_setup("sharded_db", 5) }
context "intercept" do
it "will intercept an intellij query" do
conn = PG.connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
res = conn.exec("select current_database() as a, current_schemas(false) as b")
expect(res.values).to eq([["sharded_db", "{public}"]])
end
end
end

155
tests/ruby/protocol_spec.rb Normal file
View File

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

View File

@@ -27,7 +27,7 @@ describe "Sharding" do
processes.pgcat.shutdown
end
describe "automatic routing of extended procotol" do
describe "automatic routing of extended protocol" do
it "can do it" do
conn = PG.connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
conn.exec("SET SERVER ROLE TO 'auto'")

View File

@@ -19,3 +19,10 @@ ensure
STDOUT.reopen(sout)
STDERR.reopen(serr)
end
def clients_connected_to_pool(pool_index: 0, processes:)
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
results = admin_conn.async_exec("SHOW POOLS")[pool_index]
admin_conn.close
results['cl_idle'].to_i + results['cl_active'].to_i + results['cl_waiting'].to_i
end

369
tests/ruby/stats_spec.rb Normal file
View File

@@ -0,0 +1,369 @@
# frozen_string_literal: true
require 'open3'
require_relative 'spec_helper'
describe "Stats" do
let(:processes) { Helpers::Pgcat.single_instance_setup("sharded_db", 10) }
let(:pgcat_conn_str) { processes.pgcat.connection_string("sharded_db", "sharding_user") }
after do
processes.all_databases.map(&:reset)
processes.pgcat.shutdown
end
describe "SHOW STATS" do
context "clients connect and make one query" do
it "updates *_query_time and *_wait_time" do
connections = Array.new(3) { PG::connect("#{pgcat_conn_str}?application_name=one_query") }
connections.each do |c|
Thread.new { c.async_exec("SELECT pg_sleep(0.25)") }
end
sleep(1)
connections.map(&:close)
# wait for averages to be calculated, we shouldn't do this too often
sleep(15.5)
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
results = admin_conn.async_exec("SHOW STATS")[0]
admin_conn.close
expect(results["total_query_time"].to_i).to be_within(200).of(750)
expect(results["avg_query_time"].to_i).to be_within(50).of(250)
expect(results["total_wait_time"].to_i).to_not eq(0)
expect(results["avg_wait_time"].to_i).to_not eq(0)
end
end
end
describe "SHOW POOLS" do
context "bad credentials" do
it "does not change any stats" do
bad_password_url = URI(pgcat_conn_str)
bad_password_url.password = "wrong"
expect { PG::connect("#{bad_password_url.to_s}?application_name=bad_password") }.to raise_error(PG::ConnectionBad)
sleep(1)
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
results = admin_conn.async_exec("SHOW POOLS")[0]
%w[cl_idle cl_active cl_waiting cl_cancel_req sv_active sv_used sv_tested sv_login maxwait].each do |s|
raise StandardError, "Field #{s} was expected to be 0 but found to be #{results[s]}" if results[s] != "0"
end
expect(results["sv_idle"]).to eq("1")
end
end
context "bad database name" do
it "does not change any stats" do
bad_db_url = URI(pgcat_conn_str)
bad_db_url.path = "/wrong_db"
expect { PG::connect("#{bad_db_url.to_s}?application_name=bad_db") }.to raise_error(PG::ConnectionBad)
sleep(1)
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
results = admin_conn.async_exec("SHOW POOLS")[0]
%w[cl_idle cl_active cl_waiting cl_cancel_req sv_active sv_used sv_tested sv_login maxwait].each do |s|
raise StandardError, "Field #{s} was expected to be 0 but found to be #{results[s]}" if results[s] != "0"
end
expect(results["sv_idle"]).to eq("1")
end
end
context "client connects but issues no queries" do
it "only affects cl_idle stats" do
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
before_test = admin_conn.async_exec("SHOW POOLS")[0]["sv_idle"]
connections = Array.new(20) { PG::connect(pgcat_conn_str) }
sleep(1)
results = admin_conn.async_exec("SHOW POOLS")[0]
%w[cl_active cl_waiting cl_cancel_req sv_active sv_used sv_tested sv_login maxwait].each do |s|
raise StandardError, "Field #{s} was expected to be 0 but found to be #{results[s]}" if results[s] != "0"
end
expect(results["cl_idle"]).to eq("20")
expect(results["sv_idle"]).to eq(before_test)
connections.map(&:close)
sleep(1.1)
results = admin_conn.async_exec("SHOW POOLS")[0]
%w[cl_active cl_idle cl_waiting cl_cancel_req sv_active sv_used sv_tested sv_login maxwait].each do |s|
raise StandardError, "Field #{s} was expected to be 0 but found to be #{results[s]}" if results[s] != "0"
end
expect(results["sv_idle"]).to eq(before_test)
end
end
context "clients connect and make one query" do
it "only affects cl_idle, sv_idle stats" do
connections = Array.new(5) { PG::connect("#{pgcat_conn_str}?application_name=one_query") }
connections.each do |c|
Thread.new { c.async_exec("SELECT pg_sleep(2.5)") }
end
sleep(1.1)
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
results = admin_conn.async_exec("SHOW POOLS")[0]
%w[cl_idle cl_waiting cl_cancel_req sv_idle sv_used sv_tested sv_login maxwait].each do |s|
raise StandardError, "Field #{s} was expected to be 0 but found to be #{results[s]}" if results[s] != "0"
end
expect(results["cl_active"]).to eq("5")
expect(results["sv_active"]).to eq("5")
sleep(3)
results = admin_conn.async_exec("SHOW POOLS")[0]
%w[cl_active cl_waiting cl_cancel_req sv_active sv_used sv_tested sv_login maxwait].each do |s|
raise StandardError, "Field #{s} was expected to be 0 but found to be #{results[s]}" if results[s] != "0"
end
expect(results["cl_idle"]).to eq("5")
expect(results["sv_idle"]).to eq("5")
connections.map(&:close)
sleep(1)
results = admin_conn.async_exec("SHOW POOLS")[0]
%w[cl_idle cl_active cl_waiting cl_cancel_req sv_active sv_used sv_tested sv_login maxwait].each do |s|
raise StandardError, "Field #{s} was expected to be 0 but found to be #{results[s]}" if results[s] != "0"
end
expect(results["sv_idle"]).to eq("5")
end
end
context "client connects and opens a transaction and closes connection uncleanly" do
it "produces correct statistics" do
connections = Array.new(5) { PG::connect("#{pgcat_conn_str}?application_name=one_query") }
connections.each do |c|
Thread.new do
c.async_exec("BEGIN")
c.async_exec("SELECT pg_sleep(0.01)")
c.close
end
end
sleep(1.1)
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
results = admin_conn.async_exec("SHOW POOLS")[0]
%w[cl_idle cl_active cl_waiting cl_cancel_req sv_active sv_used sv_tested sv_login maxwait].each do |s|
raise StandardError, "Field #{s} was expected to be 0 but found to be #{results[s]}" if results[s] != "0"
end
expect(results["sv_idle"]).to eq("5")
end
end
context "client fail to checkout connection from the pool" do
it "counts clients as idle" do
new_configs = processes.pgcat.current_config
new_configs["general"]["connect_timeout"] = 500
new_configs["general"]["ban_time"] = 1
new_configs["general"]["shutdown_timeout"] = 1
new_configs["pools"]["sharded_db"]["users"]["0"]["pool_size"] = 1
processes.pgcat.update_config(new_configs)
processes.pgcat.reload_config
threads = []
connections = Array.new(5) { PG::connect("#{pgcat_conn_str}?application_name=one_query") }
connections.each do |c|
threads << Thread.new { c.async_exec("SELECT pg_sleep(1)") rescue PG::SystemError }
end
sleep(2)
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
results = admin_conn.async_exec("SHOW POOLS")[0]
%w[cl_active cl_waiting cl_cancel_req sv_active sv_used sv_tested sv_login maxwait].each do |s|
raise StandardError, "Field #{s} was expected to be 0 but found to be #{results[s]}" if results[s] != "0"
end
expect(results["cl_idle"]).to eq("5")
expect(results["sv_idle"]).to eq("1")
threads.map(&:join)
connections.map(&:close)
end
end
context "clients connects and disconnect normally" do
let(:processes) { Helpers::Pgcat.single_instance_setup("sharded_db", 2) }
it 'shows the same number of clients before and after' do
clients_before = clients_connected_to_pool(processes: processes)
threads = []
connections = Array.new(4) { PG::connect("#{pgcat_conn_str}?application_name=one_query") }
connections.each do |c|
threads << Thread.new { c.async_exec("SELECT 1") rescue nil }
end
clients_between = clients_connected_to_pool(processes: processes)
expect(clients_before).not_to eq(clients_between)
connections.each(&:close)
clients_after = clients_connected_to_pool(processes: processes)
expect(clients_before).to eq(clients_after)
end
end
context "clients connects and disconnect abruptly" do
let(:processes) { Helpers::Pgcat.single_instance_setup("sharded_db", 10) }
it 'shows the same number of clients before and after' do
threads = []
connections = Array.new(2) { PG::connect("#{pgcat_conn_str}?application_name=one_query") }
connections.each do |c|
threads << Thread.new { c.async_exec("SELECT 1") }
end
clients_before = clients_connected_to_pool(processes: processes)
random_string = (0...8).map { (65 + rand(26)).chr }.join
connection_string = "#{pgcat_conn_str}?application_name=#{random_string}"
faulty_client = Process.spawn("psql -Atx #{connection_string} >/dev/null")
sleep(1)
# psql starts two processes, we only know the pid of the parent, this
# ensure both are killed
`pkill -9 -f '#{random_string}'`
Process.wait(faulty_client)
clients_after = clients_connected_to_pool(processes: processes)
expect(clients_before).to eq(clients_after)
end
end
context "clients overwhelm server pools" do
let(:processes) { Helpers::Pgcat.single_instance_setup("sharded_db", 2) }
it "cl_waiting is updated to show it" do
threads = []
connections = Array.new(4) { PG::connect("#{pgcat_conn_str}?application_name=one_query") }
connections.each do |c|
threads << Thread.new { c.async_exec("SELECT pg_sleep(1.5)") }
end
sleep(1.1) # Allow time for stats to update
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
results = admin_conn.async_exec("SHOW POOLS")[0]
%w[cl_idle cl_cancel_req sv_idle sv_used sv_tested sv_login maxwait].each do |s|
raise StandardError, "Field #{s} was expected to be 0 but found to be #{results[s]}" if results[s] != "0"
end
expect(results["cl_waiting"]).to eq("2")
expect(results["cl_active"]).to eq("2")
expect(results["sv_active"]).to eq("2")
sleep(2.5) # Allow time for stats to update
results = admin_conn.async_exec("SHOW POOLS")[0]
%w[cl_active cl_waiting cl_cancel_req sv_active sv_used sv_tested sv_login].each do |s|
raise StandardError, "Field #{s} was expected to be 0 but found to be #{results[s]}" if results[s] != "0"
end
expect(results["cl_idle"]).to eq("4")
expect(results["sv_idle"]).to eq("2")
threads.map(&:join)
connections.map(&:close)
end
it "show correct max_wait" do
threads = []
connections = Array.new(4) { PG::connect("#{pgcat_conn_str}?application_name=one_query") }
connections.each do |c|
threads << Thread.new { c.async_exec("SELECT pg_sleep(1.5)") rescue nil }
end
sleep(2.5) # Allow time for stats to update
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
results = admin_conn.async_exec("SHOW POOLS")[0]
expect(results["maxwait"]).to eq("1")
expect(results["maxwait_us"].to_i).to be_within(200_000).of(500_000)
connections.map(&:close)
sleep(4.5) # Allow time for stats to update
results = admin_conn.async_exec("SHOW POOLS")[0]
expect(results["maxwait"]).to eq("0")
threads.map(&:join)
end
end
end
describe "SHOW CLIENTS" do
it "reports correct number and application names" do
conn_str = processes.pgcat.connection_string("sharded_db", "sharding_user")
connections = Array.new(20) { |i| PG::connect("#{conn_str}?application_name=app#{i % 5}") }
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
sleep(1) # Wait for stats to be updated
results = admin_conn.async_exec("SHOW CLIENTS")
expect(results.count).to eq(21) # count admin clients
expect(results.select { |c| c["application_name"] == "app3" || c["application_name"] == "app4" }.count).to eq(8)
expect(results.select { |c| c["database"] == "pgcat" }.count).to eq(1)
connections[0..5].map(&:close)
sleep(1) # Wait for stats to be updated
results = admin_conn.async_exec("SHOW CLIENTS")
expect(results.count).to eq(15)
connections[6..].map(&:close)
sleep(1) # Wait for stats to be updated
expect(admin_conn.async_exec("SHOW CLIENTS").count).to eq(1)
admin_conn.close
end
it "reports correct number of queries and transactions" do
conn_str = processes.pgcat.connection_string("sharded_db", "sharding_user")
connections = Array.new(2) { |i| PG::connect("#{conn_str}?application_name=app#{i}") }
connections.each do |c|
c.async_exec("SELECT 1")
c.async_exec("SELECT 2")
c.async_exec("SELECT 3")
c.async_exec("BEGIN")
c.async_exec("SELECT 4")
c.async_exec("SELECT 5")
c.async_exec("COMMIT")
end
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
sleep(1) # Wait for stats to be updated
results = admin_conn.async_exec("SHOW CLIENTS")
expect(results.count).to eq(3)
normal_client_results = results.reject { |r| r["database"] == "pgcat" }
expect(normal_client_results[0]["transaction_count"]).to eq("4")
expect(normal_client_results[1]["transaction_count"]).to eq("4")
expect(normal_client_results[0]["query_count"]).to eq("7")
expect(normal_client_results[1]["query_count"]).to eq("7")
admin_conn.close
connections.map(&:close)
end
end
describe "Query Storm" do
context "when the proxy receives overwhelmingly large number of short quick queries" do
it "should not have lingering clients or active servers" do
new_configs = processes.pgcat.current_config
new_configs["general"]["connect_timeout"] = 500
new_configs["general"]["ban_time"] = 1
new_configs["general"]["shutdown_timeout"] = 1
new_configs["pools"]["sharded_db"]["users"]["0"]["pool_size"] = 1
processes.pgcat.update_config(new_configs)
processes.pgcat.reload_config
Array.new(40) do
Thread.new do
conn = PG.connect(processes.pgcat.connection_string("sharded_db", "sharding_user"))
conn.async_exec("SELECT pg_sleep(0.1)")
rescue PG::SystemError
ensure
conn.close
end
end.each(&:join)
sleep 1
admin_conn = PG::connect(processes.pgcat.admin_connection_string)
results = admin_conn.async_exec("SHOW POOLS")[0]
%w[cl_idle cl_waiting cl_cancel_req sv_used sv_tested sv_login].each do |s|
raise StandardError, "Field #{s} was expected to be 0 but found to be #{results[s]}" if results[s] != "0"
end
admin_conn.close
end
end
end
end

1
tests/rust/.gitignore vendored Normal file
View File

@@ -0,0 +1 @@
target/

1322
tests/rust/Cargo.lock generated Normal file

File diff suppressed because it is too large Load Diff

10
tests/rust/Cargo.toml Normal file
View File

@@ -0,0 +1,10 @@
[package]
name = "rust"
version = "0.1.0"
edition = "2021"
# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
[dependencies]
sqlx = { version = "0.6.2", features = [ "runtime-tokio-rustls", "postgres", "json", "tls", "migrate", "time", "uuid", "ipnetwork"] }
tokio = { version = "1", features = ["full"] }

29
tests/rust/src/main.rs Normal file
View File

@@ -0,0 +1,29 @@
#[tokio::main]
async fn main() {
test_prepared_statements().await;
}
async fn test_prepared_statements() {
let pool = sqlx::postgres::PgPoolOptions::new()
.max_connections(5)
.connect("postgres://sharding_user:sharding_user@127.0.0.1:6432/sharded_db")
.await
.unwrap();
let mut handles = Vec::new();
for _ in 0..5 {
let pool = pool.clone();
let handle = tokio::task::spawn(async move {
for _ in 0..1000 {
sqlx::query("SELECT 1").fetch_all(&pool).await.unwrap();
}
});
handles.push(handle);
}
for handle in handles {
handle.await.unwrap();
}
}

View File

@@ -0,0 +1 @@
tomli