vtorc: fetch shard names only every --topo-information-refresh-duration

[timvaillancourt]

Description

This PR causes VTOrc to fetch+cache the list of shard names it is watching when --clusters_to_watch is defined, at an interval of --topo-information-refresh-duration (default 15s). When --clusters_to_watch is undef this change has no impact. The goal of this is to reduce calls to the topo and to respect the topo refresh duration flag

This was achieved by storing the shard names when they're already fetched at the desired interval by the existent RefreshAllKeyspacesAndShards(...) func, and allowing it to be read under lock. While I was there I added a gauge for the keyspace/shard combos watched by VTOrc, as previously discussed in #feat-vtorc in the Vitess Slack

Today a list of shard names is fetched for every "refresh" action in go/vt/vtorc/logic/tablet_discovery.go, which is potentially called concurrently and more frequently than the --topo-information-refresh-duration

Related Issue(s)

Tracking issue: #17330

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.

[timvaillancourt]

This func is called by a ticker-loop in vtorc.go with a duration == to --topo-information-refresh-duration

It is also called before the ticker loop begins

[deepthi]

You might want to create a blanket "vtorc performance improvements" issue so that you can link multiple PRs to that same issue.

[codecov]

Codecov Report

Attention: Patch coverage is 61.22449% with 38 lines in your changes missing coverage. Please review.

Project coverage is 67.53%. Comparing base (45192d2) to head (0221904).
Report is 2 commits behind head on main.

Files with missing lines	Patch %	Lines
go/vt/vtorc/logic/keyspace_shard_discovery.go	0.00%	36 Missing ⚠️
go/vt/vtorc/logic/tablet_discovery.go	0.00%	2 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##             main   #17319   +/-   ##
=======================================
  Coverage   67.52%   67.53%           
=======================================
  Files        1581     1581           
  Lines      253948   253991   +43     
=======================================
+ Hits       171480   171522   +42     
- Misses      82468    82469    +1     

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

[timvaillancourt]

You might want to create a blanket "vtorc performance improvements" issue so that you can link multiple PRs to that same issue.

@deepthi good idea, I've summarized the efforts here: #17330 👍

[GuptaManan100]

I like the change, but I would implement it slightly differently. VTOrc maintains its state in a database, and this change feels like an anti-pattern. I would rather store this new information about shards to watch in a database table too. What do you think @timvaillancourt @deepthi?

[deepthi]

I like the change, but I would implement it slightly differently. VTOrc maintains its state in a database, and this change feels like an anti-pattern. I would rather store this new information about shards to watch in a database table too. What do you think @timvaillancourt @deepthi?

That's a good point. @timvaillancourt can you make that change?

[timvaillancourt]

@GuptaManan100 / @deepthi i see your point, yes I can make this a query to a sqlite3 table 👍

[timvaillancourt]

Updated without mutex because this isn't called concurrently