should I build my own environment to access new libraries?

[jzweifel]

Hi all,

I'm moving onto some experimental analysis on jupyterhub through my connection with Casper and have a simple question:

If there are python libraries I want to use that are not recognized, should I just create my own environment that has them and operate within said environment?

Asking this because yesterday I tried to import xarray and was suprised it was not a known module.

If the answer is yes, I do have to create and work within my own environment, is there anything I need to look out for?

 

[mlevy]

There are a few available conda environments, and I recommend starting with npl -- it should have xarray, numpy, matplotlib, cartopy, etc. If it turns out to be insufficient, then the next step would be to create your own environment. I don't think there is anything specific to look out for, though I'd recommend loading the conda module instead of installing conda yourself.

 

[katelynfitzgerald]

There's some great info and recommendations (for using existing envs and creating your own) in the ARC Documentation on the Using Conda and Python page as well.

 

[jzweifel]

Thanks, I was guessing there was an environment that had these common packages!

I appreciate the help!

Also, any idea on when Jupyterhub will be available again? I believe its down today

There's some great info and recommendations (for using existing envs and creating your own) in the ARC Documentation on the Using Conda and Python page as well.

Very helpful, thank you!

 

[mlevy]

The notifier message that went out said downtime would begin at 7:00a and is expected to last 12 hours:

NCAR's JupyterHub service will be down for maintenance Tuesday, January 9th. CISL staff will perform software updates and reconfigurations required as part of the Cheyenne decommissioning process.

During this outage, we will migrate the JupyterHub to newer hardware, and remove the Cheyenne server types from the interface, and we will add a Derecho batch option to the production JupyterHub to enable access to Derecho queues, though we expect most JupyterHub work will continue to happen on Casper. If you have custom language kernels or personal lab extensions, these will still be available after the outage. However, to facilitate the migration to the new hardware and improve the reliability of the service, we will be clearing the Hub database of all existing users and servers. When you sign back in after the outage, you will start with no running servers and any "named" servers will need to be recreated as needed; make sure you save all active Jupyter work before the morning of the 9th.

START: Tue Jan 9 2024 7:00 AM MST
DURATION: 12 hours