Parent pipeline status "pending" when child status is "waiting for manual action"

Everyone can contribute. Help move this issue forward while earning points, leveling up and collecting rewards.

• Close this issue

I couldn't find a bug ticket for this issue, could only find it described in a comment on a related but distinct issue here: #202239 (closed)

Summary

When a parent pipeline triggers a child pipeline, and that child pipeline has a manual step in it, when waiting for that manual step the parent pipeline displays the status "Pending" instead of "Waaiting for manual action".

Steps to reproduce

Make a project with the following files

.gitlab-ci.yaml:

trigger-child-pipeline:
  trigger:
    strategy: depend
    include:
      - local : child-pipeline-with-manual-step.yml

child-pipeline-with-manual-step.yml:

stages:
  - first
  - second
  - third

auto-first-job:
  stage: first
  script:
    - echo foo

manual-second-job:
  stage: second
  script:
    - echo "manual job, can proceed after it's triggered"
  when: manual
  allow_failure: false

auto-third-job:
  stage: third
  script:
    - echo bar

Push to gitlab, and observe the pipeline in UI under the CI/CD tab on the left.

Example Project

https://gitlab.com/alexhumphreys/test-child-pipeline-manual

What is the current bug behavior?

• The parent triggers the child pipeline
• The child pipeline gets to the manual stage and stops
• The status of the parent pipeline stays as "Pending"

What is the expected correct behavior?

It is using strategy: depend. I would expect the parent pipeline status to reflect the child pipeline status, so the parent status would be "waiting for manual action" instead of "pending".

Relevant logs and/or screenshots

Output of checks

This bug happens on GitLab.com

Possible fixes

From the discussions on #202239 (closed), it sounds like the child pipeline status is only reported at the end of the run of the child pipeline. Since the manual job stops it from reaching the end, the status never gets updated. Not sure about a line of code, but that behaviour probably needs to be changed to reflect other status. This comment here seems to suggest the same, but it's from a year ago and I can't see any issue/MR where it has been addressed.