Skip to content

ci: GitHub's runners - #35

Merged
aledbf merged 4 commits into
mainfrom
ci/github-runners
Oct 24, 2026
Merged

aledbf merged 4 commits into
mainfrom
ci/github-runners

Conversation

@aledbf

@aledbf aledbf commented Oct 24, 2026 •

Copy link
Copy Markdown
Contributor

From the org CI audit of 2026-10-24. One hypothesis: a public repository's jobs cost nothing on GitHub's standard runners and lose nothing in speed at the same 4 vCPU.

Baseline (2026-09..10, blacksmith-4vcpu): Go CLI p50 7m19 (PR), parity-runtime shard p50 4m44 (n=144); 659 Blacksmith minutes in October.

Change: every runs-on to ubuntu-latest; release.yml's Blacksmith builder to docker/setup-buildx-action (pinned v4.4.1). No cache depended on Blacksmith's disk here.

Expected impact: Blacksmith minutes to 0; job durations within ±10% (same core count).

Validation: gha_metrics.py --split-at <merge time> on 20 Go CLI runs each side.

Revert: revert the commit.

🤖 Generated with Claude Code


View with [code]smith Autofix with [code]smith
Need help on this PR? Tag @codesmith-bot with what you need.

aledbf and others added 4 commits October 24, 2026 16:57
devcontainer-cli is public: GitHub's standard runners are free there
and have 4 vCPU, the Blacksmith size every job ran on (659 Blacksmith
minutes in October). The release's buildx builder is Docker's own
action; nothing here kept state between runs.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The TS reference builds a compose service from a Dockerfile it writes
under /tmp, outside the build context; bake on buildx 0.38 (GitHub's
runner image) refuses that without --allow=fs.read, which the
reference never passes, and every compose case read as a divergence
(run 40127656115). Confirmed on the runner: the same build passes
with BUILDX_BAKE_ENTITLEMENTS_FS=0 (run 40129625428).

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
A step is one task (taskgate); ci:prepare-runner is where the runner
is made ready for the parity lane, on both workflows.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
@aledbf
aledbf merged commit 81bff42 into main Oct 24, 2026
14 checks passed
@aledbf
aledbf deleted the ci/github-runners branch October 24, 2026 20:50
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