Skip to content

ci: retry distribution artifact uploads once - #200

Merged
kvlonge merged 1 commit into
mainfrom
ci/retry-distribution-uploads
Nov 12, 2026
Merged

kvlonge merged 1 commit into
mainfrom
ci/retry-distribution-uploads

Conversation

@kvlonge

@kvlonge kvlonge commented Nov 12, 2026

Copy link
Copy Markdown
Contributor

Why

GitHub's artifact service sometimes times out creating an artifact, even after its own internal retries. One timed-out wheel upload then fails the whole distribution build, and with it a release's publish.

Changes

  • Each of the three upload-artifact steps (wheels, sdist and the release manifest) continues on error.
  • A second upload with the same name, path and options runs only when the first one failed, with overwrite: true so it replaces anything the first attempt left behind.
  • A failed retry still fails the job, so a missing or broken artifact still blocks the publish.

Verification

  • The workflow parses, and each upload is followed by its retry step under the same condition.
  • Pull request runs leave upload off, so these steps first run in a release build.
  • make check-ci passes.

A transient artifact service timeout on one upload fails the whole
distribution build, and with it a release's publish. Each wheel,
sdist and manifest upload now continues on error and, when it failed,
uploads again under the same name, replacing anything the first
attempt left behind.
@kvlonge
kvlonge added this pull request to the merge queue Nov 12, 2026
Merged via the queue into main with commit 9b9070d Nov 12, 2026
18 checks passed
@kvlonge
kvlonge deleted the ci/retry-distribution-uploads branch November 12, 2026 10:57
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