search

rapidsai / miniforge-cuda

https://hub.docker.com/r/rapidsai/miniforge-cuda/
12 stars 9 forks source link

GHA jobs fail because of too many requests #72

Closed jakirkham closed 1 week ago

jakirkham commented 1 month ago

Have started seeing GHA jobs fail because of too many request (for example:

#16 ERROR: toomanyrequests: 
------
 > pushing rapidsai/staging:miniforge-cuda-71-cuda12.0.1-base-rockylinux8-py3.9-arm64 with docker:

Should we consider some changes here to mange this?

  1. Reduce number of images built
  2. Add a shared docker pull step to minimize pulls and reuse across builds
  3. Combine all pushes into a single command at the end
  4. Minimize how many jobs are run at once
  5. Other ideas?

Note: This happens with ci-imgs too ( https://github.com/rapidsai/ci-imgs/issues/162 )

ajschmidt8 commented 1 month ago

The quickest solution is probably just to adjust the number of concurrent jobs that are allowed in the matrix workflow:

https://docs.github.com/en/actions/using-jobs/using-a-matrix-for-your-jobs#defining-the-maximum-number-of-concurrent-jobs

bdice commented 1 week ago

Closing in favor of https://github.com/rapidsai/ci-imgs/issues/162, now that miniforge-cuda is being merged into ci-imgs (https://github.com/rapidsai/ci-imgs/pull/167).