Tie 'pending stuck' with a build timeout
Problem to solve
GitLab allows people to configure long timeouts for jobs themselves (many hours), yet it kills any pending jobs that wait for more than one hour. This is a problem for small GitLab installations with a single runner, where jobs take more than one hour to complete.
Further details
This is based on discussions in
Proposal
Creating a separate timeout configuration for pending stuck
might be a UI overload, but maybe allow it to be equal or multiple of 'job timeout' for a project. In this case, such 'multiple' can be a rough measure of the allowed pending jobs.
What does success look like, and how can we measure that?
- Jobs won't be cancelled and pipelines won't fail when they wait for more than one hour.
- I won't have to patch GitLab manually
😆 (Currently I have to do this: https://gitlab.com/gitlab-org/gitlab-ce/issues/38265#note_41592455)
Links / references
- Original bugreport gitlab-ce#38265
- Current patch https://gitlab.com/gitlab-org/gitlab-ce/issues/38265#note_41592455