Skip to content
This repository has been archived by the owner on Apr 17, 2019. It is now read-only.

checkBuilds() returns true even when builds are failing? #112

Closed
ghost opened this issue Sep 23, 2015 · 3 comments
Closed

checkBuilds() returns true even when builds are failing? #112

ghost opened this issue Sep 23, 2015 · 3 comments

Comments

@ghost
Copy link

ghost commented Sep 23, 2015

https://github.com/kubernetes/contrib/blob/023e2c05df5899bf9734b0d6d21fad0b0617fb45/submit-queue/e2e.go#L93

As far as I can tell by looking at this code, this function returns true even if stable==false for some or all builds.

In theory this will cause PR's to be merged even if Jenkins e2e tests are failing.

@gmarek has reported this to be the case and @lavalamp has therefore disabled the merge bot.

Need to verify that this theoretical bug is in fact manifesting. The fix is pretty straightforward.

@brendandburns FYI.

@eparis
Copy link
Contributor

eparis commented Sep 23, 2015

Isn't that why @brendandburns filed #111

@gmarek
Copy link
Contributor

gmarek commented Sep 23, 2015

Indeed it is. We can keep this issue as a reminder:)

@ghost
Copy link
Author

ghost commented Sep 23, 2015

Before I forget, we should only merge PR's if Jenkins e2e tests are stable across multiple runs. If I'm reading the code correctly, it deems a single successful test run per job to constitute a "stable" job. In theory, a single fluky success could cause all pending PR's to auto-merge, which is bad.

In practise, we have multiple test jobs being checked, so if all of them become flaky, their intermittent successes would all have to line up correctly for PR's to merge (which is statistically unlikely). But even so, I'd suggest adding a "number of consecutive successes required" config value per job. Faster, more reliable test jobs (e.g. e2e-gce-parallel, at 20 min and 90%+ reliability) might require, say, 4 consecutive successes, while slower jobs (e.g. scalability, at multiple hours per fun), might require fewer.

I'll open a separate issue...

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

No branches or pull requests

3 participants