Use proper zstd decoder pool for binlog event compression handling

[mattlord]

Description

This is a follow-up to #16328. There we added usage of sync.Pool, however when putting the object back in the pool we used the wrong pool. Instead of using the statelessDocoderPool we were using another sync.Pool instance that also existed in the package: readersPool.

This caused the statelessDocoderPool to NOT be used, but worse than that it could cause a panic when the readersPool was used here IF --mysql-server-pool-conn-read-buffers was set (defaults to false): https://github.com/planetscale/vitess/blob/4522a0ad8930f3556627ec1ebbe69103f3b0fc01/go/mysql/conn.go#L290-L292

We should backport this to v21 as that's the first release that contains #16328.

Related Issue(s)

• Fixes: Bug Report: Incorrect sync.Pool usage in compressed transaction payload handling #17044

Checklist

• "Backport to:" labels have been added if this change should be back-ported to release branches
• If this change is to be back-ported to previous releases, a justification is included in the PR description
• Tests were added or are not required
• Did the new or modified tests pass consistently locally and on CI?
• Documentation was added or is not required

[vitess-bot]

Review Checklist

Hello reviewers! 👋 Please follow this checklist when reviewing this Pull Request.

General

• Ensure that the Pull Request has a descriptive title.
• Ensure there is a link to an issue (except for internal cleanup and flaky test fixes), new features should have an RFC that documents use cases and test cases.

Tests

• Bug fixes should have at least one unit or end-to-end test, enhancement and new features should have a sufficient number of tests.

Documentation

• Apply the release notes (needs details) label if users need to know about this change.
• New features should be documented.
• There should be some code comments as to why things are implemented the way they are.
• There should be a comment at the top of each new or modified test to explain what the test does.

New flags

• Is this flag really necessary?
• Flag names must be clear and intuitive, use dashes (-), and have a clear help text.

If a workflow is added or modified:

• Each item in Jobs should be named in order to mark it as required.
• If the workflow needs to be marked as required, the maintainer team must be notified.

Backward compatibility

• Protobuf changes should be wire-compatible.
• Changes to _vt tables and RPCs need to be backward compatible.
• RPC changes should be compatible with vitess-operator
• If a flag is removed, then it should also be removed from vitess-operator and arewefastyet, if used there.
• vtctl command output order should be stable and awk-able.

[codecov]

Codecov Report

Attention: Patch coverage is 73.52941% with 9 lines in your changes missing coverage. Please review.

Project coverage is 67.15%. Comparing base (d651a42) to head (e7191f0).
Report is 7 commits behind head on main.

Files with missing lines	Patch %	Lines
go/mysql/binlog_event_compression.go	73.52%	9 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##             main   #17042   +/-   ##
=======================================
  Coverage   67.15%   67.15%           
=======================================
  Files        1571     1571           
  Lines      251836   251876   +40     
=======================================
+ Hits       169110   169146   +36     
- Misses      82726    82730    +4     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[rohit-nayak-ps]

Should we panic here and prevent the process from launching? Option is for the decompress() to return an error if statefulDecoderPool failed to initialize.

[mattlord]

It's better to panic, I think, as this is only allocated once at init() time. If we can't initialize it, we won't be able to run properly. The way the code was previously, it would just continually log errors w/o being able to do much. This should only ever fail if you're very low on memory (ENOMEM). This isn't the decoder pool here, it's the global shared stateless decoder used for smaller payloads.

[rohit-nayak-ps]

Only for compressed payloads, right? Is it documented that this failure can only happen for ENOMEM. In case there is some other issue with zstd preventing the executable from launching seems excessive and risky.

[mattlord]

Yeah, that's true.... maybe I'll just go back to logging like I was before then. It also has the benefit of limiting the changes.

[mattlord]

I went ahead and got rid of the panic and init(): e7191f0

No reason for us NOT to try initializing it again the next time we retry...

[rohit-nayak-ps]

I will let you decide on the panic.

Otherwise lgtm.