General updates | MT JW Added you to #nrp-gpu-compute slack channel - If you’re trying new configurations and are unable to monitor them (ex overnight) it’s good to post the conditions under which other folks should shut them down and which commands to use. Cancel Friday one-on-ones? MT – Sounds good. What do we think about wednesday ones (this timeslot)? JW – Let’s keep these on for a few more weeks so we can figure out if they’re necessary/what we walk about here. MT – Agree. Hopefully lots can be handled in sit-downs when needed.
Could you come to an in-person workweek in Irvine in Feb? More date/location/travel details to come.
Leadership is planning an in-person work week in Irvine, CA the week of February 17th. Staff are encouraged to attend. Collaborators and contributors are welcome to join, but we will not be able to provide any travel support.
MT Re working on kubernetes - LW has pretty good looking workflow (quite nice, few hacks). Wondering about how to do validation/QA. Reached out to DS/EH and they sent some resources but none totally on target. Not sure about general approach (ex how do I do this in CI?). My kubernetes knowledge is quite old/limited. Would like to do better validation than “I shot it off to NRP and it worked”. JW – NRP tutorial is very good intro to kubernetes JW – Not sure about how to test - I think it’d be really difficult to set up a false k8s cluster to meaningfully evaluate how things are going. I’d almost prefer “continuous usage” as our tests, and expect science team to raise an alarm if something goes wrong/smells funny. MT – Uncomfortable about not doing any testing, would like to do a nonzero amount of research to see if there’s a good way that this could be tested. But LW won’t be back for a week so I’ll have at least that long to learn about the field.
|