Skip to content

flaky test-net-throttle on daily master #33135

Description

@sam-github

Failed last three nightlies:

Error Message
fail (1)
Stacktrace
build big string
server started on port 38697
assert.js:103
  throw new AssertionError(obj);
  ^

AssertionError [ERR_ASSERTION]: Expected values to be strictly equal:

true !== false

    at Server.<anonymous> (/home/iojs/build/workspace/node-test-commit-custom-suites-freestyle/test/pummel/test-net-throttle.js:38:10)
    at Server.emit (events.js:315:20)
    at TCP.onconnection (net.js:1537:8) {
  generatedMessage: true,
  code: 'ERR_ASSERTION',
  actual: true,
  expected: false,
  operator: 'strictEqual'
}

Activity

  1. added
    flaky-testIssues and PRs involving tests that fail intermittently in CI.
    on Apr 28, 2020
  2. sam-github commented on Apr 28, 2020

    @sam-github
    ContributorAuthor

    We were almost green last night... albeit by diligent skipping of tests.

  3. Trott commented on May 9, 2020

    @Trott
    Member

    First failed daily run was for f8d5474 and last successful daily run was for 24a4e61. That's a span of only 6 commits. If this reproduces locally, a bisect should find the problematic commit pretty quickly. And if it doesn't reproduce locally, using CI to find the problematic commit should be feasible.

  4. Trott commented on May 9, 2020

    @Trott
    Member

    Starting the bisect on CI since that will probably finish before the local compilation I'm doing for this will finish.

    https://ci.nodejs.org/job/node-test-commit-custom-suites-freestyle/14177/ will be f8d5474 and should fail.

    https://ci.nodejs.org/job/node-test-commit-custom-suites-freestyle/14178/ will be 24a4e61 and should pass.

    https://ci.nodejs.org/job/node-test-commit-custom-suites-freestyle/14179/ will be 658cae0, which is between the two.

  5. Trott commented on May 9, 2020

    @Trott
    Member

    The test passes locally for me, and all three CI jobs failed, so that suggests it may have been a change on the CI host and not a change in the code...

  6. Trott commented on May 9, 2020

    @Trott
    Member

    It looks to me like the test makes assumptions about how much data it will take before the kernel buffer fills and data gets queued in memory, and that this assumption may no longer be true for the CI host. Is that your assessment too, @sam-github? Or should all systems queue data with this test?

    @nodejs/build Any chance anyone beefed up test-rackspace-ubuntu1604-x64-1 around two weeks ago?

  7. Trott commented on May 9, 2020

    @Trott
    Member

    #33329 seems to fix it.

  8. added a commit that references this issue on May 11, 2020
  9. added a commit that references this issue on May 15, 2020
  10. added a commit that references this issue on Jun 7, 2020
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

    flaky-testIssues and PRs involving tests that fail intermittently in CI.

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions