Example of failed task: http://koji.stg.fedoraproject.org/koji/taskinfo?taskID=90155816 We can't really build unless the initial repo is created.
Lubos
workaround is to call it manually ... but we really need to process that message and re-trigger
I investigated this more and I think that in this case, our current code just marks the module build as failed in this case, because Koji itself marks the component build task as failed in this case. I think this is expected behaviour here.
We might try to resubmit the failed component build, but this would need MBS to examine the reason of failure and have a list of failures from which it can recover by simple build resubmission. We do not do that right now for anything else and it is probably out of scope for this bug.
@ralph Do you think we could close this as fixed? We clearly do not block module build in the mentioned situation.
:+1: yeah, closing as fixed works for me. :)
@ralph changed the status to Closed
Closed
Login to comment on this ticket.