Follow-up from "Refactor artifact locking"
The following discussions from !30741 (merged) should be addressed:
-
@fabiopitino started a discussion: (+1 comment) @matteeyah @dosuken123 @ayufan I think it makes sense to eventually unlock pipelines when the
Ci::Ref
changes status or latest_pipeline_id. This should also be valid for parent-child pipelines because- only the parent pipeline would contribute to the
Ci::Ref
status - when a child pipeline finishes we should do nothing
- when a parent pipeline finishes successfully it would unlock all previous pipelines (this automatically excludes its own child pipelines)
The only part of the ref life cycle that is out of our
Ci::Ref
control is the branch/tag removal which causes the remaining pipelines to be unlocked.I think it would be worth refactoring this in a follow-up MR.
- only the parent pipeline would contribute to the
-
@ayufan started a discussion: This seems not to be used.