Handling temporary CI failures


#1

Howdy! I’m setting up a bors instance for a private github org where we currently use gitlab CI. I have a weird problem that I’m tracking down where the gitlab CI check on github reports failure as soon as it starts, and a second or two later changes to “in progress”.

I’m hoping we can fix this gitlab CI + github check issue, but if we can’t, is there any way that we can teach bors how to handle this? I’m thinking of something like a check-timeout parameter, where if bors receives a status check failure, it’ll delay by the timeout parameter and re-check the status of that job (either by making a request to github, or just seeing if a newer status update was received in the meantime). Is this possible, and if so, how painful would it be to implement?

Thanks!


#2

We’ve fixed it on our gitlab CI hook. It just wasn’t properly handling a “pending” status, whoops!