srvtopo: Setup metrics in init() function

[dbussink]

We currently create a single srvTopoServer inside NewTabletServer to avoid panics on duplicate metrics.

This is wrong though, since the topo server created is tied to the context given, and if that is cancelled, it will be stopped. We will then never start a new one.

The reason this was set up like that is because of metrics. So instead of what we did before, we need to create metrics once. All other metrics are created in init() functions and we need to do the same here.

Issue

Fixes #15319

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 41.66667% with 21 lines in your changes are missing coverage. Please review.

Project coverage is 67.64%. Comparing base (696fe0e) to head (2973dc4).
Report is 49 commits behind head on main.

Files	Patch %	Lines
go/cmd/vttablet/cli/cli.go	0.00%	5 Missing ⚠️
go/vt/vtcombo/tablet_map.go	0.00%	5 Missing ⚠️
go/cmd/vtcombo/cli/main.go	0.00%	4 Missing ⚠️
go/cmd/vtgate/cli/cli.go	0.00%	3 Missing ⚠️
go/cmd/vtexplain/cli/vtexplain.go	0.00%	2 Missing ⚠️
go/vt/vttablet/tabletserver/tabletserver.go	50.00%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #15304      +/-   ##
==========================================
+ Coverage   67.41%   67.64%   +0.22%     
==========================================
  Files        1560     1561       +1     
  Lines      192752   193537     +785     
==========================================
+ Hits       129952   130919     +967     
+ Misses      62800    62618     -182     

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

[GrahamCampbell]

This sounds like a bug rather than an internal cleanup, and should be backported?

[dbussink]

This sounds like a bug rather than an internal cleanup, and should be backported?

There's no user facing bug here at the moment. Creating multiple of these is something that only happens in tests, and not in for example spinning up a vttablet. That's why I didn't mark it as a bug, but I think having this is enough of a footgun to make sure we fix it so there's not a new user facing problem introduced accidentally in the future.

[deepthi]

Can you create an issue describing the problems that this leads to? It's unclear from the description what bad things can happen with the current state.

[dbussink]

Can you create an issue describing the problems that this leads to? It's unclear from the description what bad things can happen with the current state.

Since it was an internal refactor to avoid potential problems in the future I didn't. But I guess I failed to describe in the PR properly what this is avoiding? What exactly is unclear there that I should expand on?

[dbussink]

counterPrefix was never empty, so we never created counters with no name anyway.

[dbussink]

The dynamic naming here is what is core to the problem and if the same name is (accidentally) used twice, it would panic. Instead we now always pass in the counter and build it in init() functions in the appropriate places beforehand.

[deepthi]

Callers have to do a little more work, but I do think this is a good change overall. As a bonus we get rid of the random hack we had to introduce in tests to avoid panics.