Hi
For our e2e testing, we have a child pipeline that contains build, deploy and testing jobs. This child pipeline is triggered by a bridge job (see definition below) with a resource_group
, so as to avoid concurrency.
However, if we run the pipeline twice (parent-pipeline-1
and parent-pipeline-2
), the bridge job in parent-pipeline-1
becomes “cancelled” even though the child pipeline continues running, and the bridge job in parent-pipeline-2
becomes active. This is wrong for two reason:
- The bridge job in
parent-pipeline-2
shouldn’t have started running since theresource_group
is locked. - The bridge job in
parent-pipeline-1
shouldn’t have been cancelled.
We also tried this without a resource_group
defined and the same happened. Point number 2 shouldn’t have happened in that case either.
Gitlab version: 15.3.3
Bridge job definition:
ci-e2e:
stage: e2e
rules:
- if: '$CI_COMMIT_MESSAGE =~ /run-ci-e2e/'
trigger:
include: e2e_testing/.gitlab-ci.child.yml
strategy: depend
resource_group: e2e-ci