Stuck on an issue?

Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.

And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.

gpuCI follow-up

See original GitHub issue

In https://github.com/dask/dask/pull/7966 and https://github.com/dask/distributed/pull/5147 we added support for gpu-enabled CI builds 🎉 This issue is to track some follow-up work we may to do

For dask:

We may want to include dask-cudf as some tests are currently being skipped with could not import 'dask_cudf': No module named 'dask_cudf'
In the conda list step, I see dask is installed with pip (from the python setup.py install step in build.sh), but I also see dask-core installed from conda-forge. It’d be good to confirm that the dask package which is actually used in tests corresponds to the PR being built (cc @jakirkham)
Again from the conda list step, our environment has the latest released version of distributed installed. In general, the dev version of dask depends on the dev version of distributed. I’m not sure what the best way is to ensure we always have the latest dev version of distributed installed. Perhaps add a pip install git+https://github.com/dask/distributed line to build.sh?

For distributed:

In the conda list step, our environment has the latest released version of dask installed. In general, the dev version of distributed depends on the dev version of dask. I’m not sure what the best way is to ensure we always have the latest dev version of dask installed. Perhaps add a pip install git+https://github.com/dask/dask line to build.sh?

xref https://github.com/rapidsai/dask-build-environment/pull/1 where some of these points are being addressed

cc @charlesbluca @quasiben @raydouglass

Issue Analytics

State:
Created 2 years ago
Comments:14 (14 by maintainers)

Top GitHub Comments

2reactions

GenevieveBuckleycommented, Aug 24, 2022

Looks great, thanks all!

2reactions

charlesblucacommented, Jul 25, 2022

@GenevieveBuckley update - chatted with RAPIDS ops and turns out we are able to change the bot message 🙂 now the message should read:

Can one of the admins verify this patch?

Admins can comment `ok to test` to allow this one PR to run or `add to allowlist` to allow all PRs from the same author to run.

Unfortunately, this message is set globally across all gpuCI projects, so there wasn’t an opportunity here to include a link to Dask’s gpuCI documentation, but I think this should hopefully make it a little clearer to maintainers on how to proceed when this message pops up.

In the meantime, I can look into blockers on dask/community#184 to make it clearer who the admins are for each Dask repo.

Top Results From Across the Web

gpuCI Usage Docs - RAPIDS Docs

gpuCI Usage Docs. Our gpuCI is responsible for testing each PR and commit with GPUs and building conda packages for all RAPIDS projects....

gpuCI admin team for Dask and Dask-contrib orgs · Issue #184 ...

I think a sensible solution for now would be to add all the missing members of the dask/gpu team to the current admin...

Changelog — Dask.distributed 2022.12.1 documentation

Update ga-yaml-parser step in gpuCI updating workflow (GH#7335) Charles ... Follow-up on removing report and safe from Worker.close (GH#6423) Gabe Joseph.

dask-cuda Changelog - PyUp.io

Query UCX-Py from gpuCI versioning service ... Follow-up revision of local_cudf_merge benchmark (213) rjzamora - Add codeowners file (217) raydouglass

目录 - Gitee

Siu needs to follow up ... will be a follow-up PR to current type-inference PR ... #5123 - [DISCUSS] Run CUDA tests on...