Loading
ci: disable Azure storage driver jobs
Disables the azure storage driver CI jobs, which have been failing on master
and on any merge request that touches the Azure driver.
Why
IT removed the team's Azure subscription as part of a migration to a new one,
taking the test storage account and the azure-managed_identity_auth custom
runner with it. Per Pawel, the migration was "remove everything and re-register
in the new subscription", and it was never followed up on.
The runner behind these jobs, 45220409, last contacted GitLab on
2026-04-17 — four months ago. Its two siblings are healthy, which is why
only Azure is affected:
| Runner | Tag | Status | Last contact |
|---|---|---|---|
| 45220409 | azure-managed_identity_auth |
stale | 2026-04-17 |
| 46266590 | s3-managed_identity_auth |
online | today |
| 47236873 | gcs-managed_identity_auth |
online | today |
The six azure: matrix jobs are allow_failure: false, so they sit pending for
roughly two hours before GitLab fails them with
stuck_pending_no_matching_runners, taking the whole pipeline down with them.
What this does
Comments the azure: job out rather than deleting it, so it can be restored
once we are re-provisioned and the runner is rebuilt per docs/custom_runners.md.
A comment above the block records why it is disabled and what needs to happen to
bring it back.
No other job is affected — the block defines no YAML anchors, it only consumes
*storage-driver-test and *storage-driver-variables, which the s3 and gcs jobs
continue to use.
Follow-up
Re-provisioning is tracked in #2374, linked from a TODO(jaime) above the
disabled block in .gitlab/ci/integration.yml.
Merging this also unblocks !2925 (merged), whose pipeline is currently red only because it
touches registry/storage/driver/azure/v2/ and therefore triggers these jobs.
Edited by Jaime Martinez