Remove the group that has never let the rebase run start #1086
No reviewers
Labels
No milestone
No project
No assignees
1 participant
Notifications
Due date
No due date set.
Dependencies
No dependencies set.
Reference
jaylouisw/packetforge!1086
Loading…
Add table
Add a link
Reference in a new issue
No description provided.
Delete branch "ci/auto-rebase-no-concurrency"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Over the entire history of this workflow the database holds fourteen runs: ten cancelled, four waiting, and not one that ever started.
startedis epoch 0 on every single one — including the push-triggered run created the moment #1083 landed, which was supposed to be the fix.They all entered
concurrency: {group: auto-rebase}. A run that enters a group and never gets a runner blocks every later run in that group: the group drains in order, and the order never advances past the stuck one.cancel-in-progress: truedoes not rescue it — it cancels running jobs, not waiting ones, so the waiting runs just accumulate behind the one that will never move.Bringing the two fixes together, #1071 was treating a symptom: the group was the trap, and the first run to fall into it stayed there.
Two rebase passes overlapping is not a correctness problem. Each pass reads
mergeable, rebases one branch at a time, and pushes only to pull-request branches; the worst case is a pass finding a branch already current and reporting it as current.So the group goes. If the run still does not start after this, the problem is not in this file — and that will be worth knowing too.