maxrjones
commented
2 weeks ago I'm going to use this comment to keep track of possible additions to either a FAQ or other jupyter sections based on my experiences:
- [ ] What determines which kernels launch on startup? (I've been having some very old notebooks start up every launch even though I shut them down each time)
- [ ] Where should I put ephemeral data products? (nasa-veda-scratch or /tmp; possibly using
SCRATCH_BUCKETin the future) - [ ] Document how to find out which docker image is used while in an instance
- [ ] Document science support's strategy for upgrading the default images
One recurring issue is JupyterLab reporting notebooks as corrupted JSON. We should add guidance that when this happens the general recommendation is to re-check the notebook out from git if you haven't changed anything important and use the notebook checkpoints or a json validator if you need to recover more recent changes. This would likely be best organized into an FAQ which could be modularized for reference by other Hubs (e.g., MAAP - #57, OpenScapes, CryoCloud)
cc @wildintellect