Skip to content

Unit test sporadically fails when run in parallel #302

@panicgh

Description

@panicgh

v2.4.0 introduced a new unit tests, which requires tests to be run sequentially. It would be nice if unit tests were executable in parallel to save time (e.g. during packaging, rebuilding).

The "offending" test is spy_topics_tests.no_callback_called_if_no_write:
https://github.com/eProsima/Fast-DDS-statistics-backend/blob/v2.4.0/test/unittest/StatisticsBackend/SpyTopicsTests.cpp#L229

The test creates a monitor without publisher and so it checks that it indeed receives no samples. However, when tests run in parallel, another test may publish on the same DDS domain and interfere with the test. For example tests in my setup are run via /usr/bin/ctest --test-dir redhat-linux-build --output-on-failure --force-new-ctest-process -j12. Maybe the test can be made to use a different (quiet) DDS domain or there is a different solution.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions