Skip to content

Conversation

@seemethere
Copy link
Member

The nvidia apt repositories seem to be left over on the amd nodes so
let's just go ahead and remove them explicitly if we're not testing for
CUDA

Example: https://app.circleci.com/pipelines/github/pytorch/pytorch/190222/workflows/8f75b5cd-1afd-43dc-9fa7-f7b058f07b46/jobs/6223743/steps

Signed-off-by: Eli Uriegas eliuriegas@fb.com

The nvidia apt repositories seem to be left over on the amd nodes so
let's just go ahead and remove them explicitly if we're not testing for
CUDA

Signed-off-by: Eli Uriegas <eliuriegas@fb.com>
@seemethere seemethere added the module: ci Related to continuous integration label Jul 13, 2020
@seemethere seemethere requested review from ezyang and malfet July 13, 2020 21:17
Copy link
Contributor

@ezyang ezyang left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is so janky. I have a recommendation for a follow up: let's make the ROCm workers STOP using setup_ci_environment.sh, and give them a dedicated bash script that very LOUDLY says that "these changes are permanent to the workers and not reset across runs"

@seemethere
Copy link
Member Author

This is so janky. I have a recommendation for a follow up: let's make the ROCm workers STOP using setup_ci_environment.sh, and give them a dedicated bash script that very LOUDLY says that "these changes are permanent to the workers and not reset across runs"

Agreed, I also think general we should try to move away from setup_ci_environment entirely and just have these be declared in the Dockerfiles

Copy link
Contributor

@facebook-github-bot facebook-github-bot left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@seemethere has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.

@ezyang
Copy link
Contributor

ezyang commented Jul 13, 2020

Agreed, I also think general we should try to move away from setup_ci_environment entirely and just have these be declared in the Dockerfiles

Can't do that, this environment setup is for installing nvidia-docker so we can run the docker image in the first place ;)

@facebook-github-bot
Copy link
Contributor

@seemethere merged this pull request in b6e1944.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Merged module: ci Related to continuous integration

Projects

None yet

Development

Successfully merging this pull request may close these issues.

5 participants