[14.0]: Tracking issue - flaky tests

[deepthi]

Overview of the Issue

While waiting for a PR to finish running on the release-14.0 branch, @frouioui and I noticed the following tests being flaky

• TestSchemaChange/online_CREATE_TABLE: #10674
• TestSchemaChange/revert_revert_CREATE_TABLE: #10674
• Cluster (shardedrecovery_stress_verticalsplit_heavy) - TestVerticalSplit: Delete all legacy sharding related code #10278
• Cluster (onlineddl_ghost) mysql80:
• Cluster (vreplication_across_db_versions): VReplication: retry in WaitForPos when read of pos is killed off by deadlock detector #10621

Reproduction Steps

Links to failed runs:
https://github.com/vitessio/vitess/runs/7097331285?check_suite_focus=true
https://github.com/vitessio/vitess/runs/7097333414?check_suite_focus=true
https://github.com/vitessio/vitess/runs/7097328820?check_suite_focus=true
https://github.com/vitessio/vitess/runs/7097330419?check_suite_focus=true

Binary Version

release-14.0 commit `3319a6bf39f4547b5be01744cf53f733462ee89d`

Operating System and Environment details

CI

Log Fragments

No response

[deepthi]

Re: TestVerticalSplit we should just start on deleting the old resharding code and this will go away.

[mattlord]

The flakiness of these two are related to MySQL 8 usage:

• Cluster (onlineddl_ghost) mysql80
• Cluster (vreplication_across_db_versions)

They are otherwise identical to other workflows which use 5.7 and are not (so) flakey. I have an open issue on this: #10590

[GuptaManan100]

I don't think we need this issue anymore.