circleci-orb icon indicating copy to clipboard operation
circleci-orb copied to clipboard

Need to add a timeout to wait_on command

Open bahmutov opened this issue 5 years ago • 7 comments

v1.26.0

Without a timeout, the wait_on waits for the default 10 minute timeout before failing without any output. If the server is exiting right away, this means 10 minute CI run just to fail. I think we could set the timeout of 1 minute and it would be enough.

bahmutov avatar Oct 26 '20 19:10 bahmutov

It should not be a timeout; if the concurrent process crashes, wait-on should terminate immediately.

haf avatar Oct 29 '20 15:10 haf

Well since the background process might crash without circle stopping it is hard to say how to detect this in any other way

Sent from my iPhone

On Oct 29, 2020, at 11:15, Henrik Feldt [email protected] wrote:

 It should not be a timeout; if the concurrent process crashes, wait-on should terminate immediately.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.

bahmutov avatar Oct 29 '20 15:10 bahmutov

Is it?

https://app.circleci.com/pipelines/github/logary/logary-js/548/workflows/3fe43dc2-fded-4f48-8cad-77684f5e58a4/jobs/700

This is how.

haf avatar Oct 29 '20 15:10 haf

https://github.com/kimmobrunfeldt/concurrently/blob/master/README.md#why

haf avatar Oct 29 '20 15:10 haf

I am not sure what I should see here. If we start a process like using concurrently or start-server-and-test then yes. But here we are using CircleCI step and put the server in the background. How do we know if it crashed?

bahmutov avatar Oct 29 '20 15:10 bahmutov

I guess I'm saying that wasting 10 minutes x all the instances using this "orb", is a huge waste, when IRL it might crash after 5 seconds: so it's better not to use a separate step, and instead call through some process manager.

EDIT: ...or even one minute, like you suggest.

Also, I know that concurrently might not work super-great on Windows, if that's a target platform, so start-server-and-test might be better then.

haf avatar Oct 29 '20 15:10 haf

Sure these are two different ways of doing it

Sent from my iPhone

On Oct 29, 2020, at 11:44, Henrik Feldt [email protected] wrote:

 I guess I'm saying that wasting 10 minutes x all the instances using this "orb", is a huge waste, when IRL it might crash after 5 seconds: so it's better not to use a separate step, and instead call through some process manager.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.

bahmutov avatar Oct 29 '20 15:10 bahmutov

The wait-on command has been removed from the orb in 3.0.0. Closing this issue as no longer relevant. Please open a new issue if you find yourself needing wait-on or encounter any other issues after upgrade.

jennifer-shehane avatar Mar 01 '23 15:03 jennifer-shehane