2024-10-23 Mitchell/Wagner Check-in meeting notes

• JM –

  • Let’s kubernetes it up

    • (we kubernetesed it up)

    • Important links (be sure to be monitoring both of these when jobs are running):

    • https://grafana.nrp-nautilus.io/d/85a562078cdf77779eaa1add43ccec1e/kubernetes-compute-resources-namespace-pods?orgId=1&refresh=10s&var-datasource=default&var-cluster=&var-namespace=openforcefield&from=now-1h&to=now

    • https://element.nrp-nautilus.io/#/room/#general:matrix.nrp-nautilus.io

    • JM – I’ll put this on anything-goes

    • JW – Great. And for now don’t use mroe than 10 or 20 cores max. If that looks good on monday then I’ll approve you to use 100+

• JW –

  • Some hiccups with NRP access, should be un-throttled now. I’d set up some QCF workers without enough memory so they were dying over and over again

  • Thanks for digging into bespokefit issue

    • JM – I think I got to the bottom of it, took 12ish hours, not fun debugging bespokefit.

    • JW – Some chatter about “bespokefit 2” using smee and fast QM models. I said that we need to deprecate bespokefit 1 if we make bespokefit 2 because these are hard to maintain.

    • JM – If there’s a bespokefit 2 we need to organize it better. Current behavior would be fine if reorganized too.

    • (some discussion about possible things that could be moved/restructured in a refactor)

    • (Worked on response together, posted to issue, took roughly an hour)

  • Expect bespokefit release soon, to fix issue that BW identified

  • Let’s write up a check in for yesterday, and then you can do one for today when you’re done.

    • Weds: Kubernetes submission prep. Bespokefit issue support.

    • Thurs: JW meeting. Bespokefit issue support. Kubernetes submission prep

Trello

https://trello.com/b/dzvFZnv4/infrastructure