Skip to content

Test a_stalled_pull_of_a_downloading_layer_releases_its_staging_space_once_stored fails intermittently in CI #915

Open

Description

@schmitch

Summary

a_stalled_pull_of_a_downloading_layer_releases_its_staging_space_once_stored (crates/server/tests/registry.rs, added for #763) fails now and then in CI's test job, at its first step: read_some waits 10 seconds for the first 1 MiB of the 32 MiB layer, and the wait runs out.

thread 'a_stalled_pull_of_a_downloading_layer_releases_its_staging_space_once_stored' panicked at crates/server/tests/registry.rs:3481:6:
arrived before the upstream sent the rest: Elapsed(())

Seen

Twice on 2026-11-17, on unrelated changes:

The other CI runs of that day passed, main's included.

Not reproduced

On a cloud session's machine the test passes: six runs alone (about 6.5 s each) and two runs of the whole registry binary (58 tests, about 21 s).

What is not known

Why the first megabyte takes more than 10 seconds on a CI runner: the runner's load alone (the registry binary runs its 58 tests side by side, several with layers of tens of megabytes), or something in the proxy's download path that holds the response back. A fix should find which, not only raise the bound.

Activity

  1. added this to the Backlog milestone on Nov 17, 2026
  2. added theissue type on Nov 17, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    bugSomething isn't working

    Type

    Projects

    No projects

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions