-
Notifications
You must be signed in to change notification settings - Fork 3.8k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[ci] fix flaky Azure Pipelines jobs (2) #4097
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
oy, CI is really challenging us this month
Indeed 😕 |
* [ci] use conda-forge in CI jobs (fixes #4948) * comment out more jobs * try reverting graphviz patch, running more cuda jobs * get graphviz from PyPI and try removing some patches for r-lintr * start running appveyor again * use conda-forge if using conda * fix commands * conda install graphviz * try newer openmp * pin below openmp 11.x * focus on gpu task * trying to narrow down error * maybe gcc11 is the issue * start adding other tests back * pin openmp too * maybe need to pin to gcc less than 10.x * pin libgfortran and libstdcxx as well * pin to gcc 9.3.0 * move constraints up to initial environment * add all CI jobs back * try installing python-graphviz separately * try new lightgbm/vsts-agent image * fix typo * test if pinning gcc for linux gpu_source build is still necessary * ok yes, pinning gcc is necessary * test if Linux gpu_source works with Python 3.9.6 * no special exception for Linux gpu_source job * pin to Python 3.9.6 in Linux gpu_source * try explicitly asking for libstdcxx-ng for every linux build * swap compilers * switch compilers back * revert accidental whitespace change * comment out CI * try Linux gpu_source with different Python versions * Revert "try Linux gpu_source with different Python versions" This reverts commit f6f63cb. * Revert "comment out CI" This reverts commit ece191f. * remove libxml2 install, change CONDA path * avoid installing conda in rchk job * empty commit 1 * empty commit 2 * empty commit 3 * empty commit 4 * add more verbose logging around installation of python-graphviz * empty commit 1 * get mamba info * get more conda info * add another mamba info call * allow for other macOS environments in GHA configuration * Revert "allow for other macOS environments in GHA configuration" This reverts commit a3c7a19. * get more logs from mamba * get Build.ArtifactsStagingDirectory * get more logs and try to force re-installing everything * clean cache after every step * remove --update-all and make logs less verbose * remove more print statements and uncomment jobs * test if conda-clean issue fixes segfaults for gpu_source * pin python version for gpu_source * empty commit 1 * use miniforge instead * empty commit 1 * Apply suggestions from code review * bring workarounds back * remove duplicated graphviz system-wide installation (reverts #4095, #4097, #4238) * empty commit 1 * empty commit 2 * empty commit 3 * empty commit 4 * empty commit 5 * empty commit 6 * empty commit 7 * empty commit 8 * empty commit 9 * empty commit 10 * empty commit 10 * empty commit 10 * empty commit 10 * empty commit 11 * one more try * try to downgrade Python version for Linux GPU job * swap compilers * Revert "swap compilers" This reverts commit f04dc27. Co-authored-by: Nikita Titov <nekit94-12@hotmail.com> Co-authored-by: Nikita Titov <nekit94-08@mail.ru>
This pull request has been automatically locked since there has not been any recent activity since it was closed. To start a new related discussion, open a new issue at https://github.com/microsoft/LightGBM/issues including a reference to this. |
Continuation of #4095.
Now we get
randomly.
Also, I think our Azure agents are unhealthy because I see some weird errors: failures to clone the repo, Dask test hangs for 1h and aborted, failures during
apt-get install ...
, etc.