Skip to content

Conversation

@ocaisa
Copy link
Member

@ocaisa ocaisa commented Jul 30, 2025

Currently, the script that installs CUDA and cu* libraries to unbreak symlinks has different behaviour depending on whether or not a GPU is available on the node where it is run (due to #27 where accel subdirectories are enforced for site installations).

This change standardises behaviour, ignoring the subdirectory, as this is what the hook that creates the symlinks expects (and is ok since the packages being installed are binary installations)

@ocaisa ocaisa changed the title Don't use a site installation mode when installing CUDA and libraries Don't use an accel subdirectory when installing CUDA and libraries to unbreak symlinks Jul 31, 2025
Copy link
Contributor

@casparvl casparvl left a comment

Choose a reason for hiding this comment

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

Fixes the issue I had on a bot instance I tested here SURF-hpcv/software-layer#1 (comment)

Tested, and found to work here: SURF-hpcv#2 (comment)

@casparvl casparvl merged commit ce4f213 into EESSI:main Jul 31, 2025
58 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants