VDiff/OnlineDDL: add support for running VDiffs for OnlineDDL migrations

[rohit-nayak-ps]

Description

Currently VDiff errors out if they are run for a vreplication workflow initiated by OnlineDDL. This PR adds support for that.

Caveat

Note that only OnlineDDLs created as vitess with the --postpone-completions option qualify to be validated by a VDiff. This is because the VReplication workflow needs to be in a Running state. VDiff needs to synchronize source and target streams so that it can expect to find identical snapshots on the source and the target. For this it will Stop/Start workflows.

If a VDiff is already complete then the vreplication workflow is not expected to restart since the cutover has already happened.

Related Issue(s)

• Fixes: Vdiff Support for Online DDL Workflows #12965

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 0% with 23 lines in your changes are missing coverage. Please review.

Project coverage is 68.37%. Comparing base (84dfb30) to head (c3fcd73).
Report is 2 commits behind head on main.

Files	Patch %	Lines
go/vt/vtctl/workflow/server.go	0.00%	22 Missing ⚠️
...vt/vttablet/tabletmanager/vdiff/workflow_differ.go	0.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #15546      +/-   ##
==========================================
- Coverage   68.38%   68.37%   -0.02%     
==========================================
  Files        1556     1556              
  Lines      195028   195051      +23     
==========================================
- Hits       133374   133370       -4     
- Misses      61654    61681      +27     

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

[shlomi-noach]

The main test TestOnlineDDLVDiff is very clear and looks correct. There is a manual step where you decide that "it's now a good time to test":

		waitForAdditionalRows(t, keyspace, "temp", 200)
 		want := &expectedVDiff2Result{
 			state:               "completed",
 			minimumRowsCompared: 200,
 			hasMismatch:         false,
 			shards:              []string{"0"},
 		}

I wonder what a user would actually use in production. I want to suggest a more automated approach for production, which is to wait until ready_to_complete column is 1.
In fact, I suggest adding that to TestOnlineDDLVDiff right after

waitForAdditionalRows(t, keyspace, "temp", 200)

Or am I misunderstanding and VDiff should start running as soon as the workflow does? In which case we can stop VDiff no earlier than ready_to_complete=1 (it should keep running while ready_to_complete=0).

[mattlord]

LGTM! I only had minor nits that you can ignore or address as you prefer.

[mattlord]

Side note, I think that adding this info somewhere on the VDiff reference page would be nice:

Note that only OnlineDDLs created as vitess with the --postpone-completions option qualify to be validated by a VDiff. This is because the VReplication workflow needs to be in a Running state. VDiff needs to synchronize source and target streams so that it can expect to find identical snapshots on the source and the target. For this it will Stop/Start workflows.

If a VDiff is already complete then the vreplication workflow is not expected to restart since the cutover has already happened.

[rohit-nayak-ps]

Or am I misunderstanding and VDiff should start running as soon as the workflow does? In which case we can stop VDiff no earlier than ready_to_complete=1 (it should keep running while ready_to_complete=0).

Thanks for the suggestion: I added this check.

[shlomi-noach]

I added this check.

The broader question is whether this should also be the behavior in production.

[rohit-nayak-ps]

I added this check.

The broader question is whether this should also be the behavior in production.

You mean as an option as part of an Online DDL migration, or, when the user invokes VDiff separately for an ongoing Online DDL migration?

Re: VDiff: users usually call VDiff at any time, mostly close to switching I guess, but also one or more times after the copy phase is done, while they let the workflow run for a period of time, to get confidence before they are ready to switch.

[shlomi-noach]

or, when the user invokes VDiff separately for an ongoing Online DDL migration?

That's the one I meant. Thanks for clarifying!