Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

servenv: Move away from using default HTTP muxer #12987

Merged
merged 3 commits into from
Apr 27, 2023

Conversation

dbussink
Copy link
Contributor

Up until now we've been using the implicit default HTTP mux. The problem of using this mux is that it works implicitly also for profiling endpoints. So the moment that net/http/pprof is imported, those endpoints get added.

We want to be able to make enabling the profiling endpoints optional, but there's no way to do that with the default mux except by recompiling and removing the import.

By moving everything away from the default mux to an explicit mux for the servenv, we can make this configurable in the near future. It still gets added to the default mux, but we don't use that anymore so it doesn't get exposed.

The changes here now explicitly add the profiling endpoints so that we can make that optional in a follow up.

Checklist

  • "Backport to:" labels have been added if this change should be back-ported
  • Tests were added or are not required
  • Did the new or modified tests pass consistently locally and on the CI
  • Documentation was added or is not required

@dbussink dbussink added Type: Internal Cleanup Component: General Changes throughout the code base labels Apr 26, 2023
@vitess-bot vitess-bot bot added NeedsDescriptionUpdate The description is not clear or comprehensive enough, and needs work NeedsWebsiteDocsUpdate What it says labels Apr 26, 2023
@vitess-bot
Copy link
Contributor

vitess-bot bot commented Apr 26, 2023

Review Checklist

Hello reviewers! 👋 Please follow this checklist when reviewing this Pull Request.

General

  • Ensure that the Pull Request has a descriptive title.
  • If this is a change that users need to know about, please apply the release notes (needs details) label so that merging is blocked unless the summary release notes document is included.
  • If a test is added or modified, there should be a documentation on top of the test to explain what the expected behavior is what the test does.

If a new flag is being introduced:

  • Is it really necessary to add this flag?
  • Flag names should be clear and intuitive (as far as possible)
  • Help text should be descriptive.
  • Flag names should use dashes (-) as word separators rather than underscores (_).

If a workflow is added or modified:

  • Each item in Jobs should be named in order to mark it as required.
  • If the workflow should be required, the maintainer team should be notified.

Bug fixes

  • There should be at least one unit or end-to-end test.
  • The Pull Request description should include a link to an issue that describes the bug.

Non-trivial changes

  • There should be some code comments as to why things are implemented the way they are.

New/Existing features

  • Should be documented, either by modifying the existing documentation or creating new documentation.
  • New features should have a link to a feature request issue or an RFC that documents the use cases, corner cases and test cases.

Backward compatibility

  • Protobuf changes should be wire-compatible.
  • Changes to _vt tables and RPCs need to be backward compatible.
  • vtctl command output order should be stable and awk-able.
  • RPC changes should be compatible with vitess-operator
  • If a flag is removed, then it should also be removed from VTop, if used there.

@dbussink dbussink removed NeedsDescriptionUpdate The description is not clear or comprehensive enough, and needs work NeedsWebsiteDocsUpdate What it says labels Apr 26, 2023
@github-actions github-actions bot added this to the v17.0.0 milestone Apr 26, 2023
@dbussink dbussink force-pushed the remove-default-mux-usage branch 3 times, most recently from c7bdb92 to 5480dde Compare April 26, 2023 20:56
Up until now we've been using the implicit default HTTP mux. The problem
of using this mux is that it works implicitly also for profiling
endpoints. So the moment that `net/http/pprof` is imported, those
endpoints get added.

We want to be able to make enabling the profiling endpoints optional,
but there's no way to do that with the default mux except by recompiling
and removing the import.

By moving everything away from the default mux to an explicit mux for
the servenv, we can make this configurable in the near future. It still
gets added to the default mux, but we don't use that anymore so it
doesn't get exposed.

The changes here now explicitly add the profiling endpoints so that we
can make that optional in a follow up.

Signed-off-by: Dirkjan Bussink <[email protected]>
Copy link
Member

@frouioui frouioui left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@dbussink looks good to me besides two comments.

What are we using the profiling endpoints for? I remember us having go/vt/servenv/pprof.go which allows us to profile Vitess, will the pprof feature no longer work?

go/vt/servenv/http.go Show resolved Hide resolved
go/vt/servenv/http.go Outdated Show resolved Hide resolved
Copy link
Contributor

@ajm188 ajm188 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

this is awesome! i have one suggestion around structuring, but otherwise i'm extremely excited about this

go/vt/servenv/http.go Outdated Show resolved Hide resolved
Comment on lines -32 to -33
// register the HTTP handlers for profiling
_ "net/http/pprof"
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

🎉

@dbussink
Copy link
Contributor Author

What are we using the profiling endpoints for? I remember us having go/vt/servenv/pprof.go which allows us to profile Vitess, will the pprof feature no longer work?

Because of the http pprof import that exists today and our usage of the default mux, we always end up with the http profiling endpoints today.

It’s not possible to turn those off and on because of that.

Anyone using Vitess could be using those endpoints to enable / disable specific profilers at runtime through those pprof http endpoints. We use these for example in PlanetScale internally when we need to debug issues with Vitess to do things like dump stack traces for goroutines.

That’s separate from the profiling flags which end up always enabling the profiling depending on how those flags are given. So they are separate things today.

The net change of the refactor here is that for the end user everything still works the same.

Signed-off-by: Dirkjan Bussink <[email protected]>
@dbussink dbussink mentioned this pull request Apr 27, 2023
4 tasks
@dbussink dbussink merged commit 8c3e37c into vitessio:main Apr 27, 2023
@dbussink dbussink deleted the remove-default-mux-usage branch April 27, 2023 17:20
dbussink added a commit to planetscale/vitess that referenced this pull request Jul 27, 2023
Since vitessio#12987 profiling is not
loaded implicitly through an import but only explicitly. This means we
need to make sure that `servenv.Init` is called.

It's not called for all binaries which is an oversight. It also explains
why this wasn't found with vitessio#12987
since that was tested with binaries like `vttablet` and `vtgate` and it
was assumed that this loading was done everywhere properly, which it
sadly was not.

Signed-off-by: Dirkjan Bussink <[email protected]>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Component: General Changes throughout the code base Type: Internal Cleanup
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants