Build timeouts due to infrastructure error retries?
Over the past few days, I have seen some runs timeout:
https://github.com/ClangBuiltLinux/continuous-integration2/runs/6640129078?check_suite_focus=true
https://github.com/ClangBuiltLinux/continuous-integration2/runs/6649571438?check_suite_focus=true
https://github.com/ClangBuiltLinux/continuous-integration2/runs/6665264356?check_suite_focus=true
https://github.com/ClangBuiltLinux/continuous-integration2/runs/6641485414?check_suite_focus=true
In the logs, I see certain builds go through the "provisioning" and "running" status twice:
https://builds.tuxbuild.com/29ofn9GPH3GdQdc7MHhvwNT4uG6/
https://builds.tuxbuild.com/29sD5Ez3QA13OwBN3I9q4rnYpH1/
https://builds.tuxbuild.com/29v0Yt8AKlvDB7wGvTy5GLkhkil/
https://builds.tuxbuild.com/29pKLeGq8JPyca1xXmS84OOAcJy/
I would expect the timeout limit to be restarted when a build is restarted or for the timeout to be per-build rather than per-job? Furthermore, it would be nice if builds.json
was still created when the build timed out so that we could still get results for all the other jobs that passed.