search

conda-forge / numba-feedstock

A conda-smithy repository for numba.
BSD 3-Clause "New" or "Revised" License
0 stars 27 forks source link

Arch Migrator #53

Closed regro-cf-autotick-bot closed 4 years ago

regro-cf-autotick-bot commented 4 years ago

This feedstock is being rebuilt as part of the aarch64/ppc64le migration.

Feel free to merge the PR if CI is all green, but please don't close it without reaching out the the ARM migrators first at @conda-forge/arm-arch.

If this PR was opened in error or needs to be updated please add the bot-rerun label to this PR. The bot will close this PR and schedule another one. If you do not have permissions to add this label, you can use the phrase code>@<space/conda-forge-admin, please rerun bot in a PR comment to have the conda-forge-admin add it for you.

This PR was created by the regro-cf-autotick-bot. The regro-cf-autotick-bot is a service to automatically track the dependency graph, migrate packages, and propose package version updates for conda-forge. If you would like a local version of this bot, you might consider using rever. Rever is a tool for automating software releases and forms the backbone of the bot's conda-forge PRing capability. Rever is both conda (conda install -c conda-forge rever) and pip (pip install re-ver) installable. Finally, feel free to drop us a line if there are any issues! This PR was generated by https://circleci.com/gh/regro/circle_worker/27897, please use this URL for debugging

conda-forge-linter commented 4 years ago

Hi! This is the friendly automated conda-forge-linting service.

I just wanted to let you know that I linted all conda-recipes in your PR (recipe) and found it was in an excellent condition.

step21 commented 4 years ago

@conda-forge-admin, please restart ci

jakirkham commented 4 years ago

The aarch64 build seems to timeout. Trying restarting it again.

However the ppc64le build seems to have a legitimate hang. Maybe the test is oversubscribing the hardware?

step21 commented 4 years ago

@conda-forge-admin, please restart ci

step21 commented 4 years ago

@conda-forge-admin, please restart ci

step21 commented 4 years ago

The aarch64 build seems to timeout. Trying restarting it again.

I tried again. Before, it seems like restart didn't restart drone, only travis and azure. (now too)

However the ppc64le build seems to have a legitimate hang. Maybe the test is oversubscribing the hardware?

in numba docs, it is mentioned for ppc that especially for huge machines, core count can be inaccurate and that messes with parallelization and proper execution. is that what you mean by oversubscribing?

jakirkham commented 4 years ago

Manually restarted Drone. Though I'm not too hopeful here.

That may be it. FWIW it may be the same issue with aarch64.

step21 commented 4 years ago

Mmh, cannot find the docs about cores right now. Looking at numba conda recipe, they exclude a lot of stuff for aarch64 and armv7l, but probably that relates to doing defaults and llvm differently too. https://github.com/numba/numba/blob/master/buildscripts/condarecipe.local/meta.yaml

step21 commented 4 years ago

The number is already constrained for ppc64le in https://github.com/conda-forge/numba-feedstock/blob/master/recipe/run_test.sh - locally I added a restriction for aarch64 as well. But at least for ppc64le it doesn't help (it was fixed to 16 - do we know what hardware we run on? like probably if we somehow have less that would also not be good.)

jakirkham commented 4 years ago

16 sounds way too high. Would try 1 or 2. In general CI providers aren’t giving us lots of cores to work with

step21 commented 4 years ago

Drone passes now with reduced test suite. On numba gitter I was told that there is also a --random switch which can do a certain percentage of tests. I will probably add that. ppc64le still hangs, but that is known as well, from gitter too:

as a warning, for ppc64le, there's sometimes random problems in LLVM (SIGABRT :)) and also really poorly compiled ELFs that eat memory, we've added a few test skips for these tests but they shift about as it seems to depend on some internal state we've never quite managed to pin down

Some more info on their CI issues from an issue: https://github.com/numba/numba/pull/5343#issuecomment-594428436 Not sure what to do about ppc64le, even with only a really small subset of tests it still fails. I could reduce nprocs to 1 from 2 maybe.