Skip to content

quest(ci): record platform queue measurements #5684

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
kixelated merged 3 commits into from
Oct 18, 2026
Merged

quest(ci): record platform queue measurements #5684

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
764eece
quest: claim m0/ci/platform-queue
kixelated Oct 16, 2026
c8337a5
quest: finish platform-queue; split the release ruleset wait
kixelated Oct 16, 2026
ca8acbf
Merge main; record platform queue measurements in the quest
kixelated Oct 16, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
22 changes: 14 additions & 8 deletions quest/m0/ci/platform-queue.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -19,19 +19,25 @@ the implementation PR.
execute them before merging. A release PR runs the workflows merged with
`release`, so wait until the next cut carries this workflow there; earlier,
backports would wait forever on contexts the old workflow never reports.
- Raise the main queue's check timeout from 60 to 90 minutes with the
contexts: the full macOS suite took 43 minutes in one serial job, and the
jobs allow 90.
- Raise the main queue's check timeout from 60 to 120 minutes with the
contexts. The timeout covers runner wait plus the jobs' own 90-minute
limit: of 100 queue macOS jobs (2026-10-14 to 10-16), runner wait was p50
0.2, p90 33, max 83 minutes, and wait plus run exceeded 90 minutes for 3
and 120 for none.
- Require `Platform scope` too: a failed selector skips its OS dependents,
and skipped jobs alone satisfy required checks.
- Confirm a platform-affecting main merge group runs at most one macOS job
and one Windows job, while an unrelated diff skips both. Confirm a
release-targeting PR runs the selected platform checks.
- Verified 2026-10-16: no merge group ran more than one macOS and one Windows
job, unrelated diffs skipped both, and main PRs report all three contexts
as skipped. Still confirm a release-targeting PR runs the selected platform
checks after the cut.
- Check and Test keep running on every PR and again in the merge queue; only
the native jobs move off main PRs (maintainer, 2026-10-14). Linux-only
platform jobs remain on PRs, and release packaging remains nightly.
- Nightly and releases share the five-job macOS cap. Measure overlap with
queue runs before reserving capacity or reducing the queue build limit.
- Nightly and releases share the five-job macOS cap. Measured 2026-10-14 to
10-16, Nightly never overlapped a queue wait: the long waits came from the
queue's own five concurrent builds plus release-PR and manually dispatched
Platform runs. Reserve no Nightly capacity; revisit the queue build limit
if waits recur once the gates are required.

## Required

Expand Down
Loading