Ability to rerun just a single job in a workflow

I know you can restart an entire workflow, but sometimes in a release process only one job in the workflow fails due to some third party system being unstable. I was wondering if there were any thoughts to allowing restarting a workflow from the job where it failed and also potentially restarting only one job in a particular workflow?

202 Likes

This would be really helpful. We won’t want to rerun all CI tests if only 1 job failed, we want rerun from where it’s failed.

9 Likes

Agreed. We have one workflow that has 9 jobs and it’s a pain to rerun the entire workflow if a single job fails.

5 Likes

Do we have any update on this thread?

I think this woudl be a very useful feature to avoid rerunning unnecessary time consuming jobs which are previously successful.

Another idea as workaround is to deouple the jobs into separate workflows (so the jobs can run more independently on the workflow level).

2 Likes

I’d love to see this feature as well.  We moved over to Gh-Actions from CircleCI and this is the one feature we really miss.

2 Likes

I see build tasks failing intermittently in one of my projects, hence breaking the flow. I really miss this feature right now, definitely a flaw not to have it.

2 Likes

Did we have a plan about this? I want to track this process.

1 Like

I also need this. I have macOS jobs that fail for no known reason. My guess is that they timeout waiting for capacity, but I don’t know. I would like to re-run only the failed job(s) in this case.

5 Likes

+1 specially for macos ios CI flows that sometimes need to re-run due to appstore timeouts and whatnot…

1 Like

Yep, macOS fails randomly in brew install mysql or other stuff, and having to wait 20 minutes (windows client + server) instead of ~5 mins (mac server) is painful

Please add support to run a single failed job. Thanks.

1 Like

One more guy here waiting for this! :raising_hand_man:

Really don’t get why this isn’t a thing considering it would reduce the total load on the job servers.

Please add support for this. Thanks!

1 Like

Please add support to run a single failed job. Thanks.

1 Like

bump cc @chrispat :pray:

1 Like

Don’t know if people realise this, but this is a pretty key functionality. Sometimes a test fails because of a 3-rd party resource not being available, or for whatever reason. Having to re-run the whole workflow is too much (+ one more chance for something to fail).

huge +1

3 Likes

One more in favor of that.

1 Like

+1 to this. It doesn’t make sense to rerun all our CI jobs when the only failure was building a docker image.

2 Likes

I vote for this too. I have ~10 jobs running and some of them are flaky. I’d like to be able to rerun only failed jobs. When I rerun the whole suite, other job may fail. I have to rerun workflow again and again until ALL jobs eventually pass simultaneously. It’s really annoying.

1 Like