While allow_failure: true makes the pipeline succeed, I have another problem with that:
We use the when:manual in a first job of a row of deployment jobs where the following ones use needs:.
Now when setting the first deployment job to allow_failure, the following ones all run, even if the job was failed.
That means that I have no means of creating a pipeline that
runs some build and test jobs (and enters success state after they succeed)
has a manual job followed by some more jobs needing that one
does not run the following jobs if the manual job fails
This is due to the fact that the allow_failure here seems to have two meanings:
ignore the job when calculating success
decide on whether to continue
Thus I have no means of creating the pipeline I’m after here.
Why do I need this? Because we have renovate-bot pipelines I want to automerge if tests succeed, but I still want to be able to deploy them if need arises (otherwise I’d just remove the deployment jobs from these branches)