Skip to content

fix(molecule): repair destroy playbook paths so package-testing teardown runs#1144

Open
nogueiraanderson wants to merge 2 commits into
Percona-QA:masterfrom
nogueiraanderson:fix-molecule-teardown-fullleak
Open

fix(molecule): repair destroy playbook paths so package-testing teardown runs#1144
nogueiraanderson wants to merge 2 commits into
Percona-QA:masterfrom
nogueiraanderson:fix-molecule-teardown-fullleak

Conversation

@nogueiraanderson
Copy link
Copy Markdown

@nogueiraanderson nogueiraanderson commented Jun 5, 2026

Bug

  • 13 package-testing EC2 molecule scenarios (pxc, pxb-package-testing) reference their destroy playbook at the wrong relative depth, so molecule destroy resolves to a missing file and silently no-ops, leaking the run's instances on failure. ps/debian-13 has destroy commented out entirely, with the same effect.

Fix

  • Correct each destroy: path to resolve to the suite's existing destroy playbook, computed per scenario rather than blanket-edited.
  • Re-enable destroy in ps/debian-13 to match its working siblings.
  • Remove the orphaned pxc rhel-10/rhel-9 nested scenario (a rhel-10 dir holding a second rhel-9 that Molecule never discovers, colliding with the real rhel-9 sibling) instead of path-patching a dead scenario.

Tickets

…own runs

- 14 EC2 molecule scenarios (pxc, pxb-package-testing) referenced their destroy
  playbook at the wrong relative depth, so `molecule destroy` resolved to a
  missing file and silently no-oped, leaking instances on every failed run.
  Each path is corrected to resolve to the suite's existing destroy playbook
  (including the doubly-nested pxc rhel-10/rhel-9, which needed one more level).
- ps/debian-13 had destroy commented out of the provisioner and sequences;
  re-enabled to match its working siblings.
- ps80-binary-tarball is fixed separately in PR Percona-QA#1138 and is left untouched.
- molecule/pxc/pxc84-common-min_upgrade/molecule/rhel-10/rhel-9 is an
  accidental double-nest: a rhel-10 scenario dir holding a second
  rhel-9 scenario whose name collides with the real rhel-9 sibling.
- Molecule discovers scenarios one level deep (molecule/<name>/), so the
  nested dir is never run; the valid rhel-9 sibling already covers RHEL 9.
- Supersedes the destroy-path patch this PR applied to that file.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant