User support: JupyterHub hardware requirements

[isabelizimm]

Questionnaire

1. User details:
   izimmerm

2. Desired resource change:
   I currently have a 2GiB PVC, but would like to request 8GiB. This is in order to avoid an error in spawning JH at https://jupyterhub-opf-jupyterhub.apps.zero.massopen.cloud/hub/spawn/[email protected]

CC @tumido

[tumido]

On it! 👍

[tumido]

Original issue tracked upstream: opendatahub-io-contrib/jupyterhub-singleuser-profiles#107

[durandom]

@isabelizimm did you check your PVC usage and quota?
Maybe in the Grafana Dashboard?

[tumido]

@durandom related discussion: https://operatefirst.slack.com/archives/C01RMPVUUK1/p1617017072043200

This is actually pretty nasty from the user perspective, hence reported upstream. Full PVC is generating 503 error in the Spawner UI with a suggestion to restart the server.. which is not helpful at all.

[isabelizimm]

Is there an easy way to get to Grafana for this PVC? It does not look like it is enabled when I look at the ODH dashboard when logged in.
Or maybe a better question is, is there a self-service way to tell if I hit this limit again?

[4n4nd]

@isabelizimm We do have a grafana dashboard available where you can check your PVC usage (link). You should be able to select your user id and it should show you your PVC, CPU and Memory usage.

[4n4nd]

A little more self service way would be checking the usage in your notebook terminal with a cli command.
For example, you can run df -h /opt/app-root/src and it would show you the disk usage.

[isabelizimm]

Hello!! I seem to have hit my limit again... am I able to request upgrade to 16Gi?

cc: @tumido

[HumairAK]

@isabelizimm the pvc was increased to 16Gi

[tumido]

@isabelizimm I think you want to cleanup your pipenv cache. It's consuming 8Gi of your storage now. Just remove it 😉

[1000950000@jupyterhub-nb-izimmerm-40redhat-2ecom ~]$ du -hc --max-depth=1 .
2.3M    ./.npm
512     ./.config
221M    ./.pyenv
36M     ./.local
45M     ./mailing-list-analysis-toolkit
254M    ./.flair
92M     ./.keras
8.0M    ./.virtualenvs
8.3G    ./.cache
25K     ./.jupyter
1.4M    ./.ipython
512     ./.kube
0       ./.ssh
215M    ./ocp-ci-analysis
0       ./.ipynb_checkpoints
5.5K    ./.virtual_documents
9.1G    .
9.1G    total

[1000950000@jupyterhub-nb-izimmerm-40redhat-2ecom ~]$ du -hc --max-depth=1 .cache
6.0K    .cache/pip-tools
22K     .cache/fontconfig
27K     .cache/matplotlib
0       .cache/yarn
38K     .cache/black
24M     .cache/jedi
63K     .cache/python-entrypoints
528M    .cache/pre-commit
7.7G    .cache/pipenv
684K    .cache/huggingface
8.3G    .cache
8.3G    total

If you want, you can delete the whole .cache directory, nothing bad is gonna happen.