Skip to content

CI runs once, and never cancels a commit on main - #83

Merged
aledbf merged 1 commit into
mainfrom
ci/run-once-never-cancel-main
Sep 8, 2026
Merged

CI runs once, and never cancels a commit on main#83
aledbf merged 1 commit into
mainfrom
ci/run-once-never-cancel-main

Conversation

@aledbf

@aledbf aledbf commented Sep 8, 2026

Copy link
Copy Markdown
Collaborator

The same change storage and spin-machine took. Here it also fixes a failure that has been read as flakiness.

An unfiltered push: fires alongside pull_request: for every push to a branch with one open, so every push started two identical runs. The concurrency group killed one — after both had already started, which on a self-hosted runner is not merely a wasted runner: the two integration jobs install the same release and restart the same containerd on the same machine, so the loser fails during setup with something unrelated-looking. ctr: error stating device path: stat /dev/disk/by-label was the one that led here. One of the pair passed and the other went red, which reads as a flaky test and is not.

And nothing cancels a run on main any more: two merges landing close together used to leave the first with no verdict.

What is given up: a branch with no pull request open is not tested. That is what opening one is for.

🤖 Generated with Claude Code

https://claude.ai/code/session_019GHRRFvbnZi5q3KSTvyW5a

The same change storage and spin-machine took, and here it fixes a failure that
has been read as flakiness.

An unfiltered `push:` fires alongside `pull_request:` for every push to a branch
with one open, so every push started two identical runs. The concurrency group
killed one, after both had already started — and on a self-hosted runner that is
not merely a wasted runner: the two integration jobs install the same release and
restart the same containerd on the same machine, so the loser fails during setup
with something unrelated-looking. "ctr: error stating device path: stat
/dev/disk/by-label" was the one that led here. One of the pair passed and the
other went red, which reads as a flaky test and is not.

And nothing cancels a run on main any more: two merges landing close together
used to leave the first with no verdict.

What is given up: a branch with no pull request open is not tested. That is what
opening one is for.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_019GHRRFvbnZi5q3KSTvyW5a
@aledbf
aledbf merged commit 6be13dd into main Sep 8, 2026
1 check passed
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