| |
|---|
Updates from MolSSI | BP: We are looking to move the server after getting new HDDs. Hoping there won’t be any downtime, will reroute traffic through my desktop but I will post on the slack channel. BP – May be as early as thurs or fri, depends on when equipment comes in DD – PB, will this disrupt work with BSwope and LDamore? PB – I’ve already made local copies of the data they need, so they should have what they need, though if they try to rerun the script it may fail during the transition. The whole thing takes 2-3 hours to run. The pickle file containing 73000 equilibrium conformer final records was about 4GB. They should be able to quickly get dipoles from that. So they have the script and I’ll do a one-on-one with them to help them understand how to change it to do slightly different things.
BP – Cleaning up various packages for MolSSI, including the webapps to do the ML viewer. There’s a lot that’s not totally needed and not totally done. The ML app is actually working pretty well. The reaction dataset viewer is a bit of a mess. PB – FWIW, I don’t think we use reaction dataset viewer at all. BP – Lots of unnecessary stuff around there, like social media features and some copy-paste cruft from other projects. So I’ll probably keep the ML dataset apps alive, and while I think the reaction dataset viewer isn’t very high value, I may keep it around as a template for how to do other things.
BP – Still targeting first week of Dec for initial deployment of QFractal next. Haven’t had time to take a crack at QCSubmit DD – I probably won’t have time to touch qcsubmit in the next week or two. MT – I don’t have the expertise to finish this. I did some general housekeeping around QCSubmit the last few days but was mostly porting some PRs forward. Â
JW: I think it’s fine to move slow on qcsubmit migration since it is not critical. It’s OK if it’s broken for a little while DD – Bespokefit uses QCSumit internally, and this is our interface to QCArchive. So we may be unable to do FF fits. PB – We should be OK if the old data is still available. BP – That’s right, the old datasets will still be available, we’ll just not be accepting new jobs on the old server DD – So we should keep in mind that we won’t be able to submit new sets
MT – Toolkit/QCSubmit version compatibility? JW – I think this will mean that we can’t have a single env that fetches datasets (will require new OFF toolkit) and uses SMIRNOFF plugins (requires old OFF toolkit). So as long as those can be run in separate processes it should be ok. DD – We’ll need to make a build of bespokefit (and maybe some other packages) that pins to coherent versions and targets the old server.
PB – So what will be the working combinations moving forward? DD – The old stack will still be able to access old QCFractal, but we won’t have write access. But we won’t be able to submit new datasets/access the new QCArchive until QCSubmit is updated. PB – Expecting a new dataset from trevor soon, I’ll let him know that submitting that soon would be best.
|
Infrastructure advances | QCEngine v0.25.0 released last week
Openff-qcsubmit 0.4.0 released
DD – I may need to update compute environments. Right now our openmm workers are pulling in toolkit 0.11 and qcengine 0.24, which are incompatible. I think we just haven’t noticed yet. I’ll go ahead and fix this. JW - DD, let’s talk about the logistics of hosting our own qcf instance when you have time next week (may be during team leads meeting) DD - Sure, that works. Ben can you send over the configuration you would use for hosting an instance as big as openff needs. BP - I think for 100K you can get pretty good hardware for a community hosted qcf instance. I can send over the info reg. one of the VaTech proposals. Â
|
Throughput status | May have to wait for CC to dig into the status of these RNA Trinucleotide Single Point Dataset v1.0 - Complete! After DD moved remaining 7 calcs to Vulcan, thank you! TG may submit a new dataset of small molecules (< 10 atoms).
|
User questions/issues |
PB: Since we have compute available should we give another try ? PB – I’ve put these back into error cycling DD – Great, I’ve started some workers on Lilac. PB – I’ll also add DES 370K monomers, since that’s in a similar place.
|
Science needs | Â |