Drain the PR queue in batches, and send the token that is actually set #409
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!409
Loading…
Add table
Add a link
Reference in a new issue
No description provided.
Delete branch "ci/forgejo-auto-rebase"
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?
Scheduled Forgejo Actions job that rebases stale open pull requests onto
mainand arms squash auto-merge, so a PR lands as soon as its checks pass. Conflicts are left for a human.Why it was not working
Two defects in the version this replaces:
Authorization: token ***instead of the token from the environment, so every request was unauthenticated and the job
could not have rebased or armed anything. It now reads
$TOKENand exits non-zero if it is unset.a full CI matrix (3 Node versions + smoke + docker). A single pass would have added ~1000 jobs
to a queue that already takes hours to drain. Each run now takes the oldest
batch(default 5)mergeable PRs;
workflow_dispatchaccepts a larger batch when more runner capacity exists.Conflicted PRs are filtered on
mergeablerather than retried every ten minutes — a conflict is adecision about behaviour, not a transient error.
What lands once this is on
mainmainhas just dropped.github/workflows/(#1059). The remaining blocker for the open PRs is thattheir heads still carry those legacy workflows, whose docker job runs a bare
docker buildin anode:20-bookwormjob image and fails withexit 127. Rebasing them onto the newmaindrops thelegacy workflow and leaves the
.forgejoCI, whose docker job probes for a daemon and reports itsabsence instead of failing.
ci: add Forgejo auto-rebase workflowto Drain the PR queue in batches, and send the token that is actually set8de66f92dce6e75917be