Skip to content

Intermittent: library_route_tests crumb-root test failed once under load #1001

Closed

Description

@GLinnik21

app::bridge::library_route_tests::the_crumb_root_pops_to_the_library_and_up_from_it_is_the_movies_pill (rust-modules/src/app/library_route_tests.rs) failed once locally. It was one of the three unit-suite runs in a make build-bench on the lifecycle transport branch (head ccbf99a), at a host load of 10 to 25 on 10 cores:

RED: Unit suite run (default features): 1 unit test(s) failed: app::bridge::library_route_tests::the_crumb_root_pops_to_the_library_and_up_from_it_is_the_movies_pill

The bench prints only the test's name, so the assertion message is not known.

It did not fail again:

  • the test alone passed 25 of 25 runs;
  • the full plxnative_modules suite passed 4 of 4 runs at load 17 to 37;
  • both make check passes on the same tree passed it, default and hostsim;
  • two later benchmark runs, three suite runs each, passed it.

The branch changes three re-export lines and one path in the app crate. The moved code is in plx_media, a different test binary. The test file reads no clock, sleep or TX.

So the cause is not determined. The candidates are shared process-global state between tests in the same binary under parallel scheduling, or a timing dependence that the file does not show directly.

This is filed so that a recurrence in CI or locally can be matched to it. Done when a run captures the assertion message and the cause is named with evidence, or the test is shown to depend on no shared state.

Activity

  1. added
    bugSomething isn't working
    sharingShared servers and multi-server behaviour
    on Oct 24, 2026
  2. added
    internalTooling, CI, tests and code health; no user-facing change
    and removed
    sharingShared servers and multi-server behaviour
    on Oct 24, 2026
  3. Bardioc1977 commented on Nov 2, 2026

    @Bardioc1977
    Contributor

    A status note: #1189 made World::settle wait for the store workers before it counts frames, which closes the interleaving I traced here (the Collections page graded while its page worker was still out), so I closed my PR (#1182) as superseded. I have not re-run the pair recipe against current main, so I am leaving this open for you to close if you are satisfied; if the crumb or rail rows ever fail again I will reopen with the recipe (the crumb test against each other test, --test-threads=2).

  4. GLinnik21 commented on Nov 12, 2026

    @GLinnik21
    CollaboratorAuthor

    The race behind this was fixed in #1189, so I'm closing it. It's listed in #1344; reopen if it recurs.

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 workinginternalTooling, CI, tests and code health; no user-facing change

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions