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

feat: add metrics=timings to prefer header #3507

Open
wants to merge 1 commit into
base: main
Choose a base branch
from

Conversation

taimoorzaeem
Copy link
Collaborator

@taimoorzaeem taimoorzaeem commented May 11, 2024

Closes #3410.

@steve-chavez @wolfgangwalther What exactly do I need to do to deprecate the server-timings-enabled config? Should I just remove it from the codebase?

@taimoorzaeem taimoorzaeem force-pushed the prefer/metrics branch 2 times, most recently from 953a808 to 0da1e8c Compare May 12, 2024 20:21
@laurenceisla
Copy link
Member

laurenceisla commented May 20, 2024

What exactly do I need to do to deprecate the server-timings-enabled config? Should I just remove it from the codebase?

Deprecating the config does not mean to remove it from the code (right now), only that it's not recommended anymore (could be removed in the future). It should be mentioned in comments and in the docs that it's deprecated in favor of the new feature. See, for example: 6c3d7a9 and https://postgrest.org/en/latest/references/api/preferences.html#single-json-object-as-function-parameter

Comment on lines +147 to +149
-- the preference metrics=timings cant be used before it is parsed, hence
-- parseTime will be calculated for all requests
(parseTime, apiReq@ApiRequest{..}) <- withTiming True $ liftEither . mapLeft Error.ApiRequestError $ ApiRequest.userApiRequest conf req body sCache
Copy link
Member

@laurenceisla laurenceisla May 20, 2024

Choose a reason for hiding this comment

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

Ah, so it's not possible to disable ALL timings with Prefer: metrics=timings, without parsing the headers first. So it's not a perfect replacement for server-timings-enabled = false, since some timings will be calculated even if the header is not sent: the "jwt" and "parse" timings as far as I can see.

Not sure if it's OK to always calculate these timings, I believe it will still affect the performance as mentioned in the issue. cc: @steve-chavez

Copy link
Member

Choose a reason for hiding this comment

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

Hm, so here we're forcing to always calculate the time for parsing? That would defeat the whole purpose of #3410

But I understand that we first need to parse to get the header. I guess we would need to do this in a step separate from userApiRequest, but that's messy.

Copy link
Member

Choose a reason for hiding this comment

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

So far the parse and response timings (ref) have not been that useful because they're always fast.

Another option could be to omit the parse timing when doing Prefer: metrics=timings or we could just remove support for the above mentioned. Then we wouldn't have this problem.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Development

Successfully merging this pull request may close these issues.

On-demand server-timing
3 participants