vtorc: add support for dynamic enable/disable of ERS by keyspace/shard

[timvaillancourt]

Description

This PR adds support for ERS to be enabled/disabled in VTOrc dynamically by keyspace or keyspace/shard using the topo

Summary:

• New vtctld client/server RPCs:
  • SetVtorcEmergencyReparent [--enable|-e] [--disable|-d] <keyspace> <shard>
  • Existing GetKeyspace and GetShard can be used to read the written record
• New VTOrc stat (EmergencyReparentShardDisabled) to track ERS-disabled shards
• VTOrc now:
  • Fetch/stores the disable_emergency_reparent tinyint field in the vitess_keyspace and vitess_shard tables
  • Considers the keyspace or shard-level ERS-disabled state at recovery time

tim@Tims-MacBook-Pro vitess % ./bin/vtctldclient SetVtorcEmergencyReparent --help 2>&1 | head -12 
Enable/disables the use of EmergencyReparentShard in VTOrc recoveries for a given keyspace or keyspace/shard.

Usage:
  vtctldclient SetVtorcEmergencyReparent [--enable|-e] [--disable|-d] <keyspace> <shard>

Aliases:
  SetVtorcEmergencyReparent, setvtorcemergencyreparent

Flags:
  -e, --enable    Enable the use of EmergencyReparentShard in recoveries.
  -d, --disable   Disable the use of EmergencyReparentShard in recoveries.
  -h, --help      help for SetVtorcEmergencyReparent

Related Issue(s)

#17988

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

❌ Patch coverage is 34.80176% with 148 lines in your changes missing coverage. Please review.
✅ Project coverage is 67.50%. Comparing base (1baf02b) to head (4ea4fc4).
⚠️ Report is 8 commits behind head on main.

Files with missing lines	Patch %	Lines
go/vt/vtorc/logic/topology_recovery.go	0.00%	52 Missing ⚠️
go/vt/vtctl/grpcvtctldserver/server.go	0.00%	49 Missing ⚠️
go/cmd/vtctldclient/command/topology.go	14.28%	24 Missing ⚠️
go/vt/vtorc/logic/tablet_discovery.go	39.13%	14 Missing ⚠️
go/vt/vtctl/grpcvtctldclient/client_gen.go	0.00%	5 Missing ⚠️
go/vt/topotools/keyspace.go	50.00%	2 Missing ⚠️
go/vt/vtctl/localvtctldclient/client_gen.go	0.00%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #17985      +/-   ##
==========================================
- Coverage   67.52%   67.50%   -0.02%     
==========================================
  Files        1607     1607              
  Lines      263338   263518     +180     
==========================================
+ Hits       177812   177892      +80     
- Misses      85526    85626     +100     

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[timvaillancourt]

@shlomi-noach thanks for the review! I believe the code and changelog suggestions are now addressed 👍

[shlomi-noach]

@timvaillancourt I see more traffic on this PR, please let us know when it is ready for final review.

[timvaillancourt]

@timvaillancourt I see more traffic on this PR, please let us know when it is ready for final review.

@shlomi-noach 👋 yes, this should be ready for another pass of reviews now. Thanks!

[shlomi-noach]

This is tricky. We wait for something to not happen and thus we provide 0 for expected count. But if you look at the code for utils.WaitForSuccessfulRecoveryCount and utils.WaitForSuccessfulERSCount, these functions just loop with timeout until they reach requested value, which is 0, which means they will exit immediately with success code.

Point being, if a failover was to happen, we'd still exit immediately, and this is wrong. I think in this case we need to add an imposed Sleep of a few seconds (you call the number) before calling these functions.

Or, otherwise, find a more formal way of knowing for sure that the recovery was ignored. But Sleep should be fine.

[timvaillancourt]

@shlomi-noach great point, I missed that! 👍. Yes, a sleep would work

Another way to solve this would be to wait for another metric: I'm thinking the "# of recoveries" counter, to increase for the DeadPrimary recovery AND only then should the logic check if the desired value was reached

An alternative idea, which I pondered adding in this PR, was a "# of ERS-recoveries skipped" metric

I think I'll try this 2-dimensional metrics check and see where it goes! If a sleep approach is necessary, that's an option too

[shlomi-noach]

I'm thinking the "# of recoveries"

Wouldn't that have the same behavior? You want that number to not incrase so the only way to check that would be to wait/sleep?

"# of ERS-recoveries skipped" metric

That would be good indeed.

[timvaillancourt]

@shlomi-noach I've updated the e2e test to wait for the RecoverDeadPrimary recovery to be skipped, using a new metric: SkippedRecoveries

Once we know RecoverDeadPrimary was skipped, it should be safe to check for 0 for the stats you originally pointed out. Let me know how it looks! 🙇