support G124 barium analysis

[pibion]

@MusaabFaozi is working on a G124 barium analysis and they are interested in trying this analysis on the XSEDE-hosted Jupyter system. I think this might be a nice use-case because he's processing raw data and the Dask server might be helpful!

He would need access to G124 data - we'd need someone on the CDMS side to work on copying this data to the OSN so it's accessible. I'll work on identifying someone who can help with this.

He'll also need a high-memory node. He could start with the 16GB node that's available, but it's possible he'll need more memory.

[zonca]

JupyterHub profiles

Here is the current configuration, that I haven't yet updated for Jetstream 2:

jupyterhub-deploy-kubernetes-jetstream/config_standard_storage.yaml

Lines 37 to 51 in 9b095af

	profileList:
	- display_name: "Default: 2-6 CPUs and 6 GB RAM"
	default: True
	- display_name: "Tiny: 1-3 CPUs and 3 GB RAM"
	kubespawner_override:
	cpu_limit: 3
	cpu_guarantee: 1
	mem_limit: 3G
	mem_guarantee: 3G
	- display_name: "Full node: 6 CPUs and 12 GB RAM"
	kubespawner_override:
	cpu_limit: 6
	cpu_guarantee: 4
	mem_limit: 12G
	mem_guarantee: 12G

I am currently using "m3.medium" instances for the worker nodes, they have 30 GB of RAM, so assuming we can use 25 GB, I think we can do:

• tiny, 6 GB RAM, 1 vCPU
• medium, 12 GB RAM, 3 vCPU
• full node, 25 GB RAM, 7 vCPU

I also want to remove all CPU limits, so numbers above are garantueed CPUs, but if there is availability, they can potentially use all 8 vCPU.

[pibion]

@zonca I think 25 GB RAM is a great place to start.

[zonca]

ok, it is deployed, as usual please open an issue with the error log if anything is broken.

Let's keep this issue open to track the barium analysis work.