meta/autoid: make autoid client ResetConn operation concurrency-safe (#50522)

[ti-chi-bot]

This is an automated cherry-pick of #50522

What problem does this PR solve?

Issue Number: close #50519

Problem Summary:

What changed and how does it work?

The old logic is not concurrency-safe. Although some of the fields are protected by mutex, the logic is not correct.

Imagine that there are 7K concurrent Alloc() operation, and one of them meet rpc error.
Then one ResetConn() is called, which invokes grpcConn.Close().
Note, there are many ongoing calling of Alloc() and still using the conn, so close the grpcConn cause this error:

rpc error: code = Canceled desc = grpc: the client connection is closing

This error in turn cause ResetConn() calling.
Even though GetClient() get a new client, the new client might be reset again by the ResetConn() mistakenly!
So the client can not recover from the error automatically some times (when the concurrent contention is high enough).

To summarize, there are too things we should avoid:

1. Calling grpcConn.Close() in ResetConn() when the grpcConn might still been using by some other sessions.
2. ResetConn() could be called multiple times and mistakenly reset the new connection.

Check List

Tests

• Unit test
• Integration test
• Manual test (add detailed scripts or steps below)

Rename this to cmd/autoid/main.go and go run it (modify tidb/pkg/autoid_service/autoid.go and mock 1/1000 percent error rate also).

main.txt

Use the test described in #50519, after the fix, the QPS become stable:

go test -run TestXXX
==== qps ==== 120864
==== qps ==== 143747
==== qps ==== 143514
==== qps ==== 149449
==== qps ==== 144163
==== qps ==== 147491
==== qps ==== 139042
==== qps ==== 144311
==== qps ==== 146520
==== qps ==== 139134
==== qps ==== 144419
==== qps ==== 145943
==== qps ==== 146202
==== qps ==== 137703
==== qps ==== 142836
==== qps ==== 144951
==== qps ==== 142361
==== qps ==== 142149
==== qps ==== 148664
==== qps ==== 143449
==== qps ==== 146752
==== qps ==== 142496

And the error message like this is gone:

rpc error: code = Canceled desc = grpc: the client connection is closing

• No need to test

  • I checked and no code files have been changed.

Side effects

• Performance regression: Consumes more CPU
• Performance regression: Consumes more Memory
• Breaking backward compatibility

Documentation

• Affects user behaviors
• Contains syntax changes
• Contains variable changes
• Contains experimental features
• Changes MySQL compatibility

Release note

Please refer to Release Notes Language Style Guide to write a quality release note.

Make autoid client ResetConn() operation concurrency-safe, get rid of error message like "rpc error: code = Canceled desc = grpc: the client connection is closing" when using AUTO_ID_CACHE=1

[codecov]

Codecov Report

❗ No coverage uploaded for pull request base (release-6.5@c157388). Click here to learn what that means.
The diff coverage is n/a.

Additional details and impacted files

@@               Coverage Diff                @@
##             release-6.5     #50605   +/-   ##
================================================
  Coverage               ?   73.6356%           
================================================
  Files                  ?       1094           
  Lines                  ?     350489           
  Branches               ?          0           
================================================
  Hits                   ?     258085           
  Misses                 ?      75835           
  Partials               ?      16569           

[bb7133]

LGTM

[ti-chi-bot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: bb7133, xhebox

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• OWNERS [bb7133,xhebox]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[ti-chi-bot]

[LGTM Timeline notifier]

Timeline:

• 2024-01-21 01:22:51.126734116 +0000 UTC m=+664612.691031820: ☑️ agreed by xhebox.
• 2024-01-21 01:49:19.689250754 +0000 UTC m=+666201.253548458: ☑️ agreed by bb7133.