working on PPI

[jerabaul29]

A few tricks and tips @tomaloki :

• use a tmux session on PPI (learn about the basis of tmux)
• to work on PPI, request a queue session and work on it; typically:

[LOGIN NODE] qlogin -q gpu-r8.q -l h_rt=7:00:00,h_data=40G
[QUEUE SSSION] cd MY_FOLDER
[QUEUE SSSION] source /modules/rhel8/conda/install/etc/profile.d/conda.sh && conda activate development-11-2023 && jupyter-lab --no-browser --ip=$(hostname -f)

and then you can work in jupyter by opening it in your browser. It should also be possible to use VSC instead.

See online help about qlogin, qstat, qsub, etc.

See documentation on dokit: https://dokit.met.no/brukerdok/post-processing_infrastructure?s[]=ppi

[tomaloki]

This is working smoothly now. For some reason I can't figure out how to activate the conda environment directly in vscode when I'm on ppi, do you have any ideas? But as of now I am very comfortable just working directly in jupyter lab, so it's not a problem.

[jerabaul29]

Good :) .

Not sure - I think you should somehow use an extension that makes it possible for VSC to interact with the remote machine. I used this in the past but got frustrated about VSC on some aspects, so not looked into this for a long time.

[tomaloki]

I see. Hm, it sounds like a bit of a hassle. Think I'll just stick to working directly in jupyter for now - thank you! :)

[jerabaul29]

I can ask Paulina how she does it a bit later today - I think she uses VSC on PPI :) .