Ensure we send all Stream messages in a single go routine

[GuptaManan100]

Description

This PR fixes the issue pointed out in #14760 (comment) after following up on #14939 (review).

The proposed fix is to create a new struct that ensures that all the stream messages are sent from a single go routine. Tests have also been added to ensure that everywith works as intended.

Related Issue(s)

• Fixes Bug Report: Demote Primary stuck #14760

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

All modified and coverable lines are covered by tests ✅

Comparison is base (f82fb7c) 67.45% compared to head (7fecaea) 67.47%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #15266      +/-   ##
==========================================
+ Coverage   67.45%   67.47%   +0.01%     
==========================================
  Files        1560     1560              
  Lines      193203   193224      +21     
==========================================
+ Hits       130324   130373      +49     
+ Misses      62879    62851      -28     

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

[vmg]

I don't quite like this fix. The bug we're seeing is clearly a corner case, because it's been very hard to diagnose, so it's not happening consistently. It's not a generalized issue because the scatter code in the planner tries to use locking where necessary to ensure we're not returning results concurrently. Hence, changing all response writes to go through a channel is just a fixed amount of unnecessary overhead which we should not pay for.

The right fix is figuring out exactly which of the plan primitives is sending this data concurrently and adding synchronization there, without forcing serialization on all the other primitives which are already serialized. @harshit-gangal and @systay have already gone down this path before because we found a few routes that had concurrent races and we fixed them. I'm sure they have thoughts here.

[vitess-bot]

Hello! 👋

This Pull Request is now handled by arewefastyet. The current HEAD and future commits will be benchmarked.

You can find the performance comparison on the arewefastyet website.

[harshit-gangal]

vtgate always sends a single stream of responses. It is protected by the mutex in executor which is above any engine primitive. So, we do not need this streaming protection here.