1. Forum
  2. Usenet
  3. LINUX.DEBIAN.BUGS.DIST

  • Bug#1108997: unblock: mdanalysis/2.9.0-9

    From Drew Parsons@21:1/5 to All on Wed Jul 9 11:20:01 2025
    XPost: linux.debian.devel.release

    Package: release.debian.org
    Severity: normal
    X-Debbugs-Cc: [email protected]
    Control: affects -1 + src:mdanalysis
    User: [email protected]
    Usertags: unblock

    Please unblock package mdanalysis

    [ Reason ]

    mdanalysis tests are generally flakey, randomly timing out.
    In response to Bug#1108309, mdanalysis/2.9.0-9 adds
    Restrictions: flaky to debian/tests and skips a few more tests in
    order to reduce the level of nuisance from the flakey tests in
    trixie-stable.

    Tests are currently halted on riscv64 so 2.9.0-9 will not
    automatically migrate. Hence filing this unblock request to consider
    if we're happy to provide mdanalysis/2.9.0-9 in trixie.

    [ Impact ]

    Without this update, random timeouts in debci will be registed as
    "Failed" rather than "Flakey", making the state of mdanalysis seem
    worse than its actual "known flakey" state.

    [ Tests ]
    debian/tests runs the upstream test-suite, which is actually the cause
    of the problem. Would it be better to stop running mdanalysis tests altogether?

    [ Risks ]
    No risk in 2.9.0-9 compared to 2.9.0-8. It skips more tests so in
    principle reduces the risk of test failure.

    [ Checklist ]
    [x ] all changes are documented in the d/changelog
    [x ] I reviewed all changes and I approve them
    [x ] attach debdiff against the package in testing

    [ Other info ]
    This update does not fix the random timeouts, but rather marks them as
    flaky in debci to keep them from blocking the package. Santiago Vila
    has requested to keep Bug#1108309 open for communication with upstream
    to find a more fundamental fix. We are suggesting this bug be tagged not-trixie, to let mdanalysis/2.9.0-9 into trixie.

    unblock mdanalysis/2.9.0-9

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Drew Parsons@21:1/5 to All on Wed Jul 9 14:00:02 2025
    XPost: linux.debian.devel.release

    This is a multi-part MIME message sent by reportbug.


    Package: release.debian.org
    Followup-For: Bug #1108997

    Forgot to attach the debdiff

    diff -Nru mdanalysis-2.9.0/debian/changelog mdanalysis-2.9.0/debian/changelog --- mdanalysis-2.9.0/debian/changelog 2025-05-11 23:44:46.000000000 +0200
    +++ mdanalysis-2.9.0/debian/changelog 2025-07-04 10:23:27.000000000 +0200
    @@ -1,3 +1,17 @@
    +mdanalysis (2.9.0-9) unstable; urgency=medium
    +
    + * debian/tests/control: mark mdanalysis with Restrictions: flaky.
    + mdanalysis tests are generally unreliable and fail often and
    + randomly. Closes: #1108309
    + * debian/tests: skip tests observed failing in debci
    + migration-reference runs:
    + test_dihedral_attr_warning test_AnalysisFromFunction
    + test_villin_unfolded test_rmsd_frames test_file_guess_hydrogens
    + test_gnm_run_step test_all_backends_give_correct_results
    + test_startframe test_hbond_analysis
    +
    + -- Drew Parsons <[email protected]> Fri, 04 Jul 2025 10:23:27 +0200
    +
    mdanalysis (2.9.0-8) unstable; urgency=medium

    * skip test_distances, since NSGrid is flaky and causes distance
    diff -Nru mdanalysis-2.9.0/debian/tests/control mdanalysis-2.9.0/debian/tests/control
    --- mdanalysis-2.9.0/debia
  • From Graham Inggs@21:1/5 to Drew Parsons on Wed Jul 9 14:30:02 2025
    XPost: linux.debian.devel.release

    Control: tag -1 + moreinfo

    Hi Drew

    On Wed, 9 Jul 2025 at 11:17, Drew Parsons <[email protected]> wrote:
    This update does not fix the random timeouts, but rather marks them as
    flaky in debci to keep them from blocking the package.

    Marking them 'flaky' is really bad for the infrastructure, as when the
    tests fail, they run for the full time before timing out anyway.

    Also, this upload does not address the FTBFS in bug #1108997, which
    according to past buildd logs, seems to occur on all release
    architectures (except ppc64el and s390x).

    We won't unblock this upload. Please remove the 'moreinfo' tag when
    there is an upload fixing these issues.

    Regards
    Graham

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Drew Parsons@21:1/5 to All on Wed Jul 9 14:50:01 2025
    XPost: linux.debian.devel.release

    Package: release.debian.org
    Followup-For: Bug #1108997
    X-Debbugs-Cc: [email protected]

    It sounds like the only thing we can do is to just deactivate all
    tests altogether (build and runtime), until upstream gets back to us
    on Issue#5078 (Bug#1108309).

    Drew

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Drew Parsons@21:1/5 to All on Wed Jul 9 15:00:01 2025
    XPost: linux.debian.devel.release

    Package: release.debian.org
    Followup-For: Bug #1108997

    But before entirely deactivating all tests,
    I'll first try Santiago's patch which conditionally deactivates them
    depending on the number of cpus at hand.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Drew Parsons@21:1/5 to All on Thu Jul 10 16:10:01 2025
    XPost: linux.debian.devel.release

    This is a multi-part MIME message sent by reportbug.


    Package: release.debian.org
    Followup-For: Bug #1108997
    Control: retitle 1108997 unblock: mdanalysis/2.9.0-12
    Control: tags 1108997 - moreinfo

    Santiago clarified that skipping tests on 1 or 2 cpu won't actually
    help. Tests still randomly timeout on more than 2 cpu.

    mdanalysis/2.9.0-12 deactivates all tests altogether, both build-time
    and run-time. Can this one be unblocked? debdiff attached.

    unblock mdanalysis/2.9.0-12

    diff -Nru mdanalysis-2.9.0/debian/changelog mdanalysis-2.9.0/debian/changelog --- mdanalysis-2.9.0/debian/changelog 2025-05-11 23:44:46.000000000 +0200
    +++ mdanalysis-2.9.0/debian/changelog 2025-07-10 11:07:57.000000000 +0200
    @@ -1,3 +1,38 @@
    +mdanalysis (2.9.0-12) unstable; urgency=medium
    +
    + * skip all tests, both build-time and run-time. See Bug#1108309.
    +
    + -- Drew Parsons <[email protected]> Thu, 10 Jul 2025 11:07:57 +0200
    +
    +mdanalysis (2.9.0-11) unstable; urgency=medium
    +
    + * fix order of conditional control of build-time tests
    +
    + -- Drew Parsons <[email protected]> Wed, 09 Jul 2025 19:16:36 +0200
    +
    +mdanalysis (2.9.0-10) unstable; urgency=medium
    +
    + * apply suggestion from Bug#1108309 to skip tests (build and
    + runtime) if less than 2 CPUs are available
    + * do not run build-time tests on hurd-i386
    + or when nocheck is set in DEB_BUILD_OPTIONS
    +
    + -- Drew Parsons <[email protected]> Wed, 09 Jul 2025 15:12:21 +0200
    +
    +mdanalysis (2.9.0-9) unstable; urgency=medium
    +
    + * debian/tests/control: mark mdanalysis with Restrictions: flaky.
  • From Graham Inggs@21:1/5 to Drew Parsons on Mon Jul 14 16:20:01 2025
    XPost: linux.debian.devel.release

    On Thu, 10 Jul 2025 at 16:09, Drew Parsons <[email protected]> wrote:
    Santiago clarified that skipping tests on 1 or 2 cpu won't actually
    help. Tests still randomly timeout on more than 2 cpu.

    mdanalysis/2.9.0-12 deactivates all tests altogether, both build-time
    and run-time. Can this one be unblocked? debdiff attached.

    unblock mdanalysis/2.9.0-12

    Please confirm that the package no longer randomly FTBFS and downgrade
    the severity of #1108309 if this is the case.

    I've added a skiptest hint for now. This will be fine for trixie.
    Please let elbrus know when the autopkgtests are properly fixed and
    can be enabled again.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Drew Parsons@21:1/5 to Graham Inggs on Mon Jul 14 17:20:01 2025
    XPost: linux.debian.devel.release

    Package: release.debian.org
    Followup-For: Bug #1108997
    Control: severity 1108309 important

    Graham Inggs wrote:
    Please confirm that the package no longer randomly FTBFS and downgrade
    the severity of #1108309 if this is the case.

    That's right, mdanalysis/2.9.0-12 skips tests altogether so by
    construction they will not randomly fail during build.

    For autopkgtest I implemented the change with Restrictions: flaky
    in mind. So the skip is applied via "exit 2" in debian/tests/test-mdanalysis. I did it this way so that debci tests can continue to be undertaken in experimental, without locking up the test machines in unstable,
    testing and trixie.

    Downgrading #1108309 bug severity accordingly.


    I've added a skiptest hint for now. This will be fine for trixie.
    Please let elbrus know when the autopkgtests are properly fixed and
    can be enabled again.

    Thanks. It will be safe to restart autopkgtest once 2.9.0-12 is in
    trixie.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Santiago Vila@21:1/5 to Graham Inggs on Mon Jul 14 20:40:01 2025
    XPost: linux.debian.devel.release

    severity 1108309 normal
    thanks

    On Mon, Jul 14, 2025 at 04:16:55PM +0200, Graham Inggs wrote:
    Please confirm that the package no longer randomly FTBFS and downgrade
    the severity of #1108309 if this is the case.

    I confirm that the package builds ok now without random failures.

    I've added a skiptest hint for now. This will be fine for trixie.
    Please let elbrus know when the autopkgtests are properly fixed and
    can be enabled again.

    Ok, I hope Drew can take care of that (whenever the time come).

    Thanks.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)