Skip to content

Erratic build failure without any error #8920

Closed

Description

@DarthGandalf

https://travis-ci.org/znc/znc/jobs/316693823

The log looks perfectly fine, except for the last line "Done. Your build exited with 1."
Other jobs in the same build succeeded correctly.

While this could be because of trusty update, I don't see a direct connection.

Activity

  1. DarthGandalf commented on Dec 15, 2017

    @DarthGandalf
    Author

    Update: it's not related to the update, because we use group: deprecated-2017Q3

  2. BanzaiMan commented on Dec 15, 2017

    @BanzaiMan
    Contributor

    oh, you specified it.

  3. BanzaiMan commented on Dec 15, 2017

    @BanzaiMan
    Contributor

    Could you try setting the verbose bash option to see if it yields any interesting information? set -v. Thanks.

  4. DarthGandalf commented on Dec 15, 2017

    @DarthGandalf
    Author

    Thanks, I can add set -v for future, but I suspect that rerun will fix it... I hoped you have some idea what could cause this error without rerunning the job.

  5. added a commit that references this issue on Dec 16, 2017
  6. DarthGandalf commented on Dec 17, 2017

    @DarthGandalf
    Author

    Hi, seeing new errors, which may or may not be related to the one above.

    https://travis-ci.org/znc/znc/jobs/317632628 reports that a command failed without even trying to run that command, there is no $ env ... string above it.

    travis_time:end:16263fba:start=1513505651387101343,finish=1513505651584247805,duration=197146462
    �[0K
    �[31;1mThe command "env LLVM_PROFILE_FILE="$PWD/unittest.profraw" make VERBOSE=1 $UNITTEST" exited with 1.�[0m
    

    https://travis-ci.org/znc/znc/jobs/317632629 and https://travis-ci.org/znc/znc/jobs/317632637 both say make: write error after successfully doing what they were supposed to do.

  7. matthew-brett commented on Dec 18, 2017

    @matthew-brett

    I'm having this problem - here's an example where two different runs on the same code have different build failures, with no error message, just a stopped log:

    These two builds run the exact same code a few 10s of minutes apart - just an empty commit difference between them - but have different matrix entries failing:

    For example, in this job for Python 2.7, 32-bit, there is no error message in the log but the build failed. The same matrix entry passed with the same code, just adding an empty commit to trigger the build - see ok build.

    This seems may well be happening for other projects too - there are new unexpected build failures, without matching errors, at:

  8. matthew-brett commented on Dec 18, 2017

    @matthew-brett
  9. avagin commented on Dec 18, 2017

    @avagin
  10. avagin commented on Dec 18, 2017

    @avagin

    I tried to debug this issue and run make under strace, but make is completed successfully in this case.
    https://travis-ci.org/avagin/criu/builds/317936454

  11. avagin commented on Dec 19, 2017

    @avagin

    I did a few experiments and I found that sometime programs fail to write on stdout.

    https://travis-ci.org/avagin/linux/jobs/318414661

    + cat .config
    ....
    # CONFIG_SENSORS_G760A is cat: write error: Resource temporarily unavailable
    
  12. avagin commented on Dec 19, 2017

    @avagin

    A script is executed with stdout, for which O_NONBLOCK is set. It is obviously a bug.

    + cat /proc/self/fdinfo/1
    pos:	0
    flags:	0104001
    mnt_id:	10
    
  13. avagin commented on Dec 19, 2017

    @avagin

    As a workaround, I added the following line in my test script:

    python -c "import fcntl; fcntl.fcntl(1, fcntl.F_SETFL, 0)"
    
  14. 38 remaining items

  15. added a commit that references this issue on Oct 1, 2018
  16. retorquere commented on Apr 18, 2019

    @retorquere

    I'm still seeing this problem; all my builds fail with "write could not complete without blocking"

  17. TheoLechemia commented on Jul 16, 2019

    @TheoLechemia

    Same probleme here. On Ubuntu 16.04

  18. added a commit that references this issue on Sep 5, 2019
  19. bsipocz commented on Mar 2, 2020

    @bsipocz

    I think this should not be closed, the bug is still at large, and popped up for us in the past few months.
    Finding a workaround in a closed bug report is not what I would expect, maybe the stale bot shouldn't close the unresolved reports?

  20. added a commit that references this issue on Mar 5, 2020
  21. added a commit that references this issue on Mar 30, 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

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions