Slack notifications for broken pipelines should should not display retried jobs as failed

Problem to solve

With GitLab 12.2, Slack messages about pipelines has more information on the message.

Our project use retry option to rerun jobs to handle some random failure (unable to write on runner disk, etc.).

image

In example above :

  • test:cypress is a job which failed in every retries
  • test:functional:ui:others is a job which failed one time but passed on retry

It is a little bit misleading on the report. We thought that the job test:functional:ui:others failed after all retry but it was not the case.

Intended users

  • Delaney (Development Team Lead)
  • Sasha (Software Developer)
  • Devon (DevOps Engineer)
  • Other ?

Further details

My goal as a user is to know what jobs did not pass on a pipeline even with retry.

I want also to know which jobs were unstable (i.e. required an auto or manual retry to pass)

Proposal

  • Move outside of Failed job(s) section the list of passed jobs which were retried.
  • Add a Retried job(s) section for the list of jobs which pass after retry
  • Add the number of manual/auto retry for each job to see how unstable it was