21:00:03 <oneswig> #startmeeting scientific-sig 21:00:04 <openstack> Meeting started Tue Apr 13 21:00:03 2021 UTC and is due to finish in 60 minutes. The chair is oneswig. Information about MeetBot at http://wiki.debian.org/MeetBot. 21:00:05 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 21:00:07 <openstack> The meeting name has been set to 'scientific_sig' 21:00:08 <oneswig> ... away we go! :-) 21:00:17 <oneswig> so what's new? 21:00:34 <oneswig> trandles: LinkedIn tells me congratulations are in order for your new role? 21:00:46 <julianp> (looks at LinkedIn) 21:01:04 <trandles> ah, thanks oneswig 21:01:14 <trandles> it's a title.../shrug 21:01:23 <oneswig> Couldn't happen to a better guy :-) 21:01:37 <trandles> I'm fighting more responsibility until I get my boss to agree to push some of my work on someone else :P 21:01:46 <oneswig> So no significant change? 21:02:52 <trandles> significant change, but trying to ease into it 21:03:07 <trandles> Probably _more_ cloud-related stuff in this role 21:03:27 <trandles> I have a platform now to push for better HPC/Cloud integration and adoption 21:03:34 <oneswig> The Next Platform will be queuing up for your pearls of wisdom before you know it 21:03:41 <trandles> Before I'd talk and they could just ignore me :P 21:04:05 <oneswig> Well that's just excellent. 21:04:52 <julianp> Congrats trandles. 21:05:41 <oneswig> So what else is new? I'm attending GTC (virtually) 21:06:08 <julianp> What is GTC? 21:06:26 <oneswig> Nvidia's GPU technology conference 21:06:42 <trandles> thanks julianp 21:06:52 <oneswig> All things Nvidia. More green branding than you ever thought possible / bearable 21:07:09 <trandles> The fact LANL and CSCS are taking delivery of Grace systems is big news for me at least 21:07:32 <oneswig> LANL too? I think they were an also-ran in that announcement! 21:08:23 <oneswig> I didn't get a feel for what the Grace CPU was capable of by itself, it was mostly described as a sidekick to an A100 21:11:15 <trandles> Don't look at me, I'll just point at the press releases 21:11:23 <oneswig> martial: you there? 21:11:28 <martial> am here 21:11:33 <oneswig> hey martial 21:11:37 <oneswig> #chair martial 21:11:38 <openstack> Current chairs: martial oneswig 21:12:23 <oneswig> trandles: did they say when? I hope Nvidia hasn't picked up Mellanox's habit of pre-announcing tech 21:12:35 <trandles> for us it's "early 2023" 21:12:55 <trandles> I think that's the same for CSCS 21:14:04 <oneswig> The system diagrams were reminiscent of the Transputer. I was wondering about NUMA overheads. 21:14:24 <martial> have not had a chance to view the keynote yet 21:16:33 <oneswig> It's a bish-bash-bosh of one announcement after another. But this time he doesn't get any hardware out of the oven. I was disappointed. 21:17:06 <martial> true :) 21:17:15 <julianp> I want a Transputer. 21:17:17 <oneswig> In the idle moments, I was wondering how he gets the fryup grease out of that plaster frieze above his cooker. 21:18:23 <oneswig> Now if you see the keynote, you'll be thinking this too 21:19:38 <b1airo> hi gents - sorry i'm on a call with SoftIron crew 21:19:55 <oneswig> Hi b1airo, livestream it :-) 21:20:02 <oneswig> #chair b1airo 21:20:03 <openstack> Current chairs: b1airo martial oneswig 21:20:36 <oneswig> OK we should probably cover some business wrt the PTG next week. 21:20:38 <martial> so on the todo list for next week is for me to share a google meet link for everyone to join 21:20:53 <oneswig> quick work martial! 21:21:40 <trandles> PTG next week....oof... link to the times? I better get it on my calendar if possible. 21:21:47 <oneswig> That would be good. Can we share it via the mailing list? 21:22:01 <martial> I was going to ask how to share it 21:22:36 <oneswig> trandles: its Wednesday, two 1-hour sessions at 14:00 UTC and 21:00 UTC 21:22:45 <martial> I was thinking Slack originally 21:23:33 <oneswig> Slack would cover most people. There's always a bit of missionary evangelism needed to find others though. 21:24:20 <oneswig> Perhaps instead of disclosing the link in a public way on the list, suggest people contact you for it? 21:25:12 <martial> I can do that 21:25:38 <martial> actually we can put it in the Etherpad 21:25:42 <b1airo> yeah i'd just ask people to reply if interested 21:25:55 <b1airo> and we can put it in etherpad at the start of the meeting 21:26:00 <oneswig> For the lightning talks, I was just thinking we should ask Kristi from MOC to present their work on Adjutant for user on-boarding. We are planning to try that out ourselves. 21:26:12 <oneswig> good idea b1airo 21:26:15 <julianp> Oh my yes. I'd love to see that. 21:26:35 <martial> cool, we have a basic etherpad ready 21:26:45 <martial> not sure if it was going to be an official one 21:26:52 <oneswig> I'll mail now before I forget 21:27:07 <martial> #link https://etherpad.opendev.org/p/202104PTG_ScientificSIG_LightingTalks 21:27:31 <oneswig> martial: previously I think they've created a bunch of etherpads with a predictable naming sequence. I haven't checked if they've done that again this time 21:27:37 <martial> we have currently 6 slots open 21:28:09 <oneswig> Only 6 remaining? :-) 21:28:41 <martial> oneswig: agree, but I was unclear if that was still true, so I created this one 21:30:54 <martial> from what you were saying we only have 5 slots available ... they go out fast 21:32:21 <oneswig> I had a look on the OpenStack wiki - this is the Xena cycle, right? No sign of an etherpads page yet 21:34:00 <martial> well we have a placeholder then :) 21:34:39 <oneswig> Yes, that's great. 21:35:20 <oneswig> I'm going to do a talk but I'm not sure what on yet. 21:35:40 <julianp> What are the candidate topics? 21:36:02 <martial> I can do a talk about edge use case on jetson nano using containers but that seems less openstack-y and HPC to me 21:36:14 <oneswig> julianp: Monitoring for Slurm, Blazar and preemptible instances and the Scalable Metal Service cloud. 21:36:41 <martial> (although it might interest b1airo and his edge idea) 21:36:43 <b1airo> sounds interesting to me martial - and no doubt other "Scientific" types :-) 21:36:56 <oneswig> martial: one of my team has a Jetson nano in a peanut butter tub. It's awesome apparently 21:37:04 <julianp> Yup. I'm into that. 21:37:04 <b1airo> don't have to limit ourselves to HPC 21:37:45 <b1airo> peanut butter IS AWESOME, oneswig...? 21:37:54 <martial> it is a nice little "cheap" GPU processor that can easily be placed at the edge 21:38:21 <b1airo> anyone else got 30 GTC tabs open at the moment? 21:38:25 <oneswig> b1airo: yeah, the peanut butter got eaten first, should have said that 21:38:53 <martial> b1airo: nah, will watch it "after", too many fires 21:38:55 <oneswig> b1airo: alas no, too distracting! It's putting a dent in my day 21:39:51 <martial> if people are interested in the nano, I am happy to recommend ... 21:39:54 <martial> #link https://github.com/datamachines/cuda_tensorflow_opencv/tree/master/JetsonNano 21:40:00 <oneswig> martial: I saw you'd organised the PTG sessions with a proper meeting in the first slot. If we get EMEA people with a lightning talk, would you object to them presenting it then? 21:40:16 <b1airo> have heard of food grade lubricant being used as an immersive liquid cooling solution, not sure peanut butter would be good though... wonder what its specific heat capacity is 21:40:32 <martial> we can have them present their lighting talk then of course, we can only record that section 21:40:40 <trandles> We happily never cared about GPUs at LANL. Then Nvidia bought Mellanox. Then Nvidia got in the ARM game. Now I should be watching GTC but I can't be bothered. 21:40:55 <b1airo> lol trandles 21:41:10 <martial> my plan was to record each talk independently so that everyone can get their own slot 21:41:19 <martial> and their own video 21:41:36 <b1airo> trandles: i suspect you might be interested in GRACE 21:41:36 <oneswig> martial: you work the tech, we have faith in you :-) 21:41:43 <trandles> isn't the viscosity of food grade lubricant kinda high? 21:42:10 <martial> but if there is interest in the nano stack, I can talk about the k3s stack for the nanos :) 21:42:31 <oneswig> trandles: have you got another chat going? :-) 21:43:02 <b1airo> trandles: https://dug.com/dug-cool/ 21:43:11 <trandles> b1airo: suddenly, yes, I'm interested in GRACE and BlueField 21:43:16 <martial> SC virtual this year still? no idea of the oil to use to dip my server in anymore, that is what SC's floor is about, right? :) 21:44:14 <trandles> b1airo: that's an interesting cooling solution 21:44:36 <trandles> martial: I hear from someone on the organizing committee that SC is still going to try and be in-person 21:44:44 <martial> wow, shiny 21:45:16 <martial> nice, looking forward to it! 21:45:31 <oneswig> The focus of this discussion is wrong, we should move from the cooling of computers to the heating of swimming pools - by computers - https://www.leaf.cloud 21:45:34 <trandles> I fear that SC can be in-person for certain people, and a complete miss for others, based on factors totally outside of the organizer's control. 21:46:29 <b1airo> yeah i can't see in-person being a good idea this year 21:46:32 <trandles> FWIW I'm "fully vaccinated" from the US CDC's perspective, more than 2 weeks past my second pfizer dose 21:46:59 <oneswig> trandles: nice work. I have my first jab scheduled for Monday. 21:47:04 <b1airo> even for those people who have been vaccinated there will still be considerable institutional inertia around international travel 21:47:17 <trandles> no tail, no lizard scales...and I don't live too far from a 5G tower 21:47:44 <trandles> so disappointed 21:47:48 <martial> same as Tim 21:47:50 <oneswig> I'm getting the GH-325 vaccine - no, wait that was something else... 21:47:54 <trandles> b1airo: I think that's highly likely to be the case. 21:48:15 <martial> b1airo: good point 21:48:26 <trandles> For some reason all of my dreams are now Bill Gates lectures. 21:48:38 <b1airo> i always thought there was something super about you oneswig ;-) 21:49:16 <oneswig> ha :-) good point though b1airo, we are still without an office. That might change but plane flights are like going to another galaxy from where we are now 21:50:02 <martial> oneswig: you have to watch the Falcon/Winter Soldier, it is a show for you :) 21:50:39 <oneswig> martial: My daughter is crazy about that, I've watched a few with her. Pretty good! 21:52:34 <oneswig> What can we do to get a few more folks signed up to speak? 21:53:37 <trandles> I'll reach out to someone who's doing a new RHOSP deployment and mentioned something about "Ceph going on hyperconverged Compute" 21:53:50 <b1airo> we should ask the Nectar cloud core team if they've got something 21:53:58 <julianp> Can we convince jomlow and/or jfischer to talk about Jetstream 2.0 progress? 21:54:08 <trandles> I don't want to throw this person under the bus by dropping a name. 21:54:19 <julianp> XD 21:54:30 <martial> and I can ask Rion to talk about Agave on k8s too 21:54:32 <oneswig> All good ideas 21:54:42 <trandles> julianp: I don't know if Jetstream 2 hardware arrived on their floor yet. 21:55:11 <trandles> I'm always interested in what rbudden is doing with Discover at NASA 21:56:31 <trandles> I can't talk about our production deployment yet. Fingers crossed for Berlin though... 21:56:45 <trandles> If they re-run last year's plan 21:56:57 <martial> *crossing my fingers for Tim too* 21:57:00 <oneswig> trandles: true, rbudden and jonmills take a fundamental approach to managing things 21:57:21 <martial> getting close to the hour 21:57:36 <oneswig> Nearly at time - let's all go away and aim to get some lightning talk content. 21:57:47 <julianp> Good plan. 21:58:26 <oneswig> the etherpad again: https://etherpad.opendev.org/p/202104PTG_ScientificSIG_LightingTalks 21:59:16 <trandles> bye folks...gotta get on another call 21:59:30 <oneswig> Thanks all, hopefully see you next week 21:59:50 <martial> bye 21:59:58 <martial> #endmeeting