Skip to content
Snippets Groups Projects
Unverified Commit ff03d562 authored by Andreas Wassmer's avatar Andreas Wassmer Committed by GitHub
Browse files

Update installation guide (#573)


Update installation guide in order to address CUDA driver issue.

Co-authored-by: default avatarLukas Gehrig <32157249+Lukas113@users.noreply.github.com>
parent d80f759e
Branches
Tags v0.25.0
No related merge requests found
......@@ -29,6 +29,8 @@ conda env config vars set LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/lib/wsl/lib
Karabo versions older than `v0.15.0` are deprecated and therefore installation will most likely fail. In addition, we do not support Karabo older than latest-minor version in case dependency resolving or online resources are outdated. Therefore, we strongly recommend using the latest version of Karabo. If an older version of Karabo is required, we recommend using a [container](container.md), as the environment is fixed in a container, or extract the environment dependencies using `conda env export --no-builds`. However, outdated online resources may still occur.
Please refer to the troubleshooting section at the bottom of this page if you feel something does not work.
## Update to latest Karabo version
A Karabo installation can be updated the following way:
......@@ -41,3 +43,4 @@ conda update -c nvidia/label/cuda-11.7.0 -c i4ds -c conda-forge karabo-pipeline
## Additional Notes and Troubleshooting
- Dont' install anything into the base environment except libraries which are supposed to live in there. If you accientally install packages there which are not supposed to be there, you might break some functionalities of your conda-installation.
- If you're using a system conda, it might be that you don't have access to a libmamba-solver, because the solver lives in the base environment, which belongs to root. In this case, you can ask your admin to install the solver, try an installation without the libmamba solver OR we recommend to just install conda into your home (which is the recommended solution).
- In some rare cases the correct CUDA drivers are not installed. This may be the case when you get the error `RuntimeError: OSKAR library not found.` More information about this issue can be found [here (issue #568)](https://github.com/i4Ds/Karabo-Pipeline/issues/568). Refer to this issue for a workaround.
\ No newline at end of file
0% Loading or .
You are about to add 0 people to the discussion. Proceed with caution.
Please register or to comment