VReplication: Undo vschema changes on LookupVindex workflow creation fail

[mattlord]

Description

This restores the original vschema on the LookupVindex create command's --keyspace, where the vindex is created, when the workflow creation fails.

It also adds a check that when using a matching table that already exists in the --table-keyspace, the table does not already have any data.

Related Issue(s)

• Fixes: LookupVindex create does not properly cleanup / undo state changes made when it fails to create the workflow #15984

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 81.57895% with 14 lines in your changes missing coverage. Please review.

Project coverage is 69.46%. Comparing base (95f2e3e) to head (345115e).
Report is 16 commits behind head on main.

Files with missing lines	Patch %	Lines
go/vt/vtctl/workflow/server.go	80.82%	14 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #16810      +/-   ##
==========================================
- Coverage   69.51%   69.46%   -0.06%     
==========================================
  Files        1569     1571       +2     
  Lines      202517   203042     +525     
==========================================
+ Hits       140780   141033     +253     
- Misses      61737    62009     +272     

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

[deepthi]

👍

[mattlord]

The changes in this file are not relevant here, but I wanted to add the same new capabilities to the test framework for future use (soon).

[rohit-nayak-ps]

RowCount is from information_schema.tables which may not always be updated. We have anecdotally found those stats not getting updated for significant periods of time: showing 0 even after several copy phases have run. On this branch I started a second workflow sharing tables with the first one and it passed this validation, failing with aDuplicate Key error.

Which was why we were going for the approach in this PR: https://github.com/vitessio/vitess/pull/16826/files#diff-d3a320c7b03791f5d24189e3ae6d7fcac814f4fa1b3d7c02d496b3d8f0adf588R1040-R1043.

[mattlord]

Yeah, the row count is an estimate. It can remain 0 for some time if you have not inserted/updated enough rows to go beyond the initial 16KiB page in the tablespace. I would guess that in your testing you only had a few very small rows in the table (e.g. corder). Doing multiple copy phases would not help or matter in this case either. I don't think that this is a very likely scenario in production — since InnoDB updates the stats in information_schema when 1/16th of the table has changed, so now we're talking about 1KiB, and that normally happens pretty quickly on a brand new table that starts out as 16KiB in size — but it's certainly possible (then you'd have to wait for one of the other things that trigger an update).

I can remove the "table has existing data" code here in this PR then if you prefer the other approach. That was not the primary focus in this PR anyway. Sound good?

Thanks!

[rohit-nayak-ps]

My tests were for a trivial number of rows, yes. But this was also reported by others recently for larger tables where the copy phase itself took minutes. They were using Progress and not seeing any changes there without analyze table ... Not sure why the stats update didn't happen in that case.

In any case, yes, let's handle the check for data in that other PR. Rest looks good.

[mattlord]

OK, I did that here: 345115e

[rohit-nayak-ps]

My tests were for a trivial number of rows, yes. But this was also reported by others recently for larger tables where the copy phase itself took minutes. They were using Progress and not seeing any changes there without analyze table ... Not sure why the stats update didn't happen in that case.

In any case, yes, let's handle the check for data in that other PR. Rest looks good.