Fix for FullStatus gRPC connection pooling

[GuptaManan100]

Description

Fixes a bug in TabletManagerClient's pool dialer. In:

vitess/go/vt/vttablet/grpctmclient/client.go

Lines 163 to 170 in 8833092

	if client.rpcClientMap == nil {
	client.rpcClientMap = make(map[string]chan *tmc)
	}
	c, ok := client.rpcClientMap[addr]
	if !ok {
	c = make(chan *tmc, concurrency)
	client.rpcClientMap[addr] = c
	client.mu.Unlock()

rpcClientMap caches tablet connections indefinitely based on address. If a tablet goes down, it does not get evicted from the map. If a new tablet then bootstraps with the same address, then rpcClientMap makes us continuously try connecting to that tablet, even if it's on another cluster.

The solution is to evict a tablet/connection from the map upon connection/gRPC error.

However, since said error is seen long after we've returned the cached client / dialer, the user of the client needs to proactively invalidate the cached connection. To that effect, we now return an invalidator() function that can invalidate a cached connection based on the returned error.

Related Issue(s)

• Bug Report: leaking tablet client connections in poolDialer #15563

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

Deployment Notes

[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 81.81818% with 4 lines in your changes are missing coverage. Please review.

Project coverage is 65.72%. Comparing base (696fe0e) to head (7da6c01).
Report is 139 commits behind head on main.

Files	Patch %	Lines
go/vt/vttablet/grpctmclient/client.go	81.81%	4 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #15520      +/-   ##
==========================================
- Coverage   67.41%   65.72%   -1.70%     
==========================================
  Files        1560     1560              
  Lines      192752   194594    +1842     
==========================================
- Hits       129952   127893    -2059     
- Misses      62800    66701    +3901     

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

[GuptaManan100]

Rest LGTM

[GuptaManan100]

@harshit-gangal Would some error codes from MySQL also pass this check and end up causing us to invalidate the grpc connection?

[shlomi-noach]

@GuptaManan100 as @dbussink points out, adding the Close() change seems to break many tests.

[harshit-gangal]

nit: can remove passing error to invalidator

[shlomi-noach]

Converted this PR to draft, as we explore alternative approaches: #15541, #15559

[shlomi-noach]

A different implementation in #15562 .