Following on from the previous couple of closed tickets, we would find it very handy to have a feature that will automatically retry failed builds, failed build steps or failed workflows in the event of transient network issues.
We experience a high number of build failures due to temporary strangeness in network conditions that are remedied by a simple Workflow click on ‘retry from failed’. Sometimes I do this ten times a day.
This is a feature in GitLab, Travis and other competitors, so it would be nice to know it’s being considered.
I got an answer from support that I am hoping to try.
You could set something like this up by using the when attribute combined with the api to retry a build. For the when attribute, you can set a step after the test job that only runs on_fail and calls the api to automatically trigger a rerun, using the built in environment variables.