11:00:40 #startmeeting scientific-sig 11:00:41 Meeting started Wed Nov 22 11:00:40 2017 UTC and is due to finish in 60 minutes. The chair is oneswig. Information about MeetBot at http://wiki.debian.org/MeetBot. 11:00:42 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 11:00:45 The meeting name has been set to 'scientific_sig' 11:00:57 Lets see what happens with the name change... 11:01:31 #link agenda for today https://wiki.openstack.org/wiki/Scientific_SIG#IRC_Meeting_November_22nd_2017 11:02:29 roll call - who's here? 11:02:52 o/ 11:03:01 * ildikov is lurking :) 11:03:08 hello again john, hi ildiko 11:03:24 awfully quiet this morning... 11:03:58 oneswig: johnthetubaguy: hi :) 11:04:15 * johnthetubaguy hides from ildikov 11:04:21 only kidding :) 11:04:42 johnthetubaguy: well, I have a new version of the new attach flow patch that you should review :) 11:04:53 evening 11:05:07 johnthetubaguy: I can ask oneswig to remind you after the meeting if that helps ;) 11:05:09 Hi b1airo 11:05:13 ildikov: totally should go have a look at that again, something for after lunch with a nice coffee :) 11:05:27 #chair b1airo 11:05:27 not sure i will manage to stay the course - quickly fading through combination of jetlag and heat 11:05:28 Current chairs: b1airo oneswig 11:05:40 johnthetubaguy: tests are supposed to pass now, so pretty please 11:05:51 b1airo: shall we do a roundup of SC first up then? 11:06:03 sure 11:06:04 #topic SC roundup 11:06:05 johnthetubaguy: mriedem has Turkey week going so you're my only hope 11:06:18 Turkey week - ha! :-) 11:06:54 #link SKA keynote at SC https://youtu.be/VceKNiRxDBc?t=1h3m2s 11:07:13 check out the A/V gear on stage there... 11:07:24 i haven't actually seen the keynote yet, was it good? 11:07:31 makes the OpenStack summit look like a school play! 11:08:10 b1airo: I think it was well received. John T was there and saw some enthusiastic response 11:08:53 sounds like they got Alexa to read the intro... 11:09:28 yes, it's a peculiar voice over. A bit like a very calm customer support kind of voice 11:11:03 b1airo: you mentioned the BoF was pretty well attended, but perhaps too many speakers up front for one session? 11:11:42 yep 11:11:58 i'd say slightly lower attendance than last year but still a pretty full room 11:12:08 Last year was a very big room 11:12:15 there was also a HPC Cloud BoF by DK Panda and co. on at the same time 11:12:27 so not great scheduling 11:12:35 John T tells me that went off on a docker vs singularity tangent 11:12:52 i didn't speak to anyone that went to that BoF yet though so can't say if it pulled much attention 11:13:00 and was not rich in data analytics content 11:13:02 oh ok 11:13:09 pity 11:13:14 yeah 11:13:56 the actual Shifter / Singularity / CharlieCloud session was very demilitarised 11:14:07 demilitarised? 11:14:17 as in, no religious war? 11:14:34 yeah, uncontroversial 11:14:58 that sounds better. 11:15:15 How were the OpenStack talks at the IU booth? 11:15:30 there were some guys (i forget names and orgs) apparently representing europe (big call) on that panel who made some interesting points about MPI integration 11:16:22 "speaking on behalf of Europe"... that's saying something, 11:17:30 good, i think Tim probably had the most new material to share. also had that Intel fellow talking about OpenHPC integration 11:17:54 Was that Jeff? I think he's moved internally within Intel, unfortunately 11:17:58 Did the books all go? 11:18:34 Jeff sounds good - I had John stuck in my head by knew it wasn't right :-) 11:19:10 it's 10:20pm and still almost 30C 11:19:26 Are you in Australia? It wasn't like that a couple of weeks ago... 11:19:53 books were well received in the BoF. i think the IU booth still had a few on Wednesday evening but probably distributed the rest on Thurs 11:20:13 yep - Melbourne weather! 11:20:24 it's not 30C everywhere, but it's definitely a pretty nice evening :) 11:20:41 OK, good to know about the books. 11:20:45 4 seasons in 1 day and all that 11:20:56 ildikov, where are you this evening? 11:21:03 b1airo: it's only dark season here today... 11:21:32 b1airo: Forster (great Lakes or something like 4 hours North from Sydney) 11:22:25 ildikov: nice! Is it safe to swim in the lakes or are they full of toxic Australian fauna? 11:22:53 oneswig: haven't asked/tried yet 11:22:58 re. SC17, if I had to sum things up I'd say my impression is that OpenStack seems quite entrenched across quite a few large scale lab and university hpc centres now 11:23:10 oneswig: but it certainly sounds discouraging to do so :) 11:23:45 i think next year we would do well to spend more time on going around the room and trying to pull out bugbears and common problems 11:23:45 b1airo: good to know. Part of the wider movement out of the lab and into production in research computing 11:24:09 b1airo: that sounds pretty good! 11:24:09 the challenge is to pull the conversation away from specific version and upgrade issues towards unmet requirements 11:24:55 b1airo: I think you are right. It's too easy to talk without encouraging the quiet ones at the back to add their thoughts 11:25:07 b1airo: is there any chance to have them involved in the SIG for instance or it's trickier than that? 11:25:23 I was encouraged at how many people in Sydney had not taken part in the SIG previously 11:26:27 In the past an etherpad to type your thoughts on certainly helps get thoughts from those that don't want to speak up over the loud people 11:26:39 johnthetubaguy: +1 11:26:47 agreed, i did use an etherpad :-D 11:26:50 johnthetubaguy: sure, we had one 11:27:02 but I'm not sure how much it was used. 11:27:16 there were maybe 8 people on it at most though 11:27:26 There's a culture of debate-via-etherpad that doesn't come naturally outside of upstream contributors 11:27:31 yeah, making people use the thing can be harder 11:27:45 (however i was only able to put the URL up during the first 5 minutes as presenters had their own slides and there was no other screen or board 11:27:50 probably not all their with their laptop charged ready to type in an etherpad 11:28:03 ah... slides, that kills etherpad use 11:28:35 I think the lightning talks went quite well though - we had some good diversity of content there. 11:29:00 oneswig, yes i find the debate via etherpad thing very hard to follow - basically killed my following of updates to the recent LTS discussion 11:29:09 (clearly there is no perfect solution here, very curious how it goes outside of the OpenStack world!) 11:29:54 oneswig, agreed, was good to have some new (to me at least) stories there 11:30:04 johnthetubaguy: me too. Is K8s developed in this manner I wonder? 11:30:37 oneswig: yeah, don't know 11:30:54 good to see more folks stepping up to speak, sounds healthy :) 11:31:29 ildikov: do you know how well subscribed the openstack-sigs mailing list is now? Just wondering if we should keep discussions on operators and sigs lists 11:32:13 oneswig: I don't have the number, but will ask 11:32:55 at some point discussions should cut over from cross-posting but don't want to lose the relevant audience by doing that prematurely... 11:33:05 oneswig: we keep referring to it in some Dev ML discussions etc and encouraging people to join, so I would think it's going up 11:33:17 OK, that's good. 11:33:41 I will get back to you on this 11:35:16 I think we are too thin on the ground today to do much with the other items on the agenda. I suggest I follow up on the mailing list about them. 11:35:56 I had one item - for next week, a Slurm-on-OpenStack theme. b1airo do you think Chris Hines would be interested? 11:36:36 oneswig, yes i'm sure he would 11:36:47 i suspect many HPC operators would be 11:37:17 Piyanai I know was interested in presenting their work again 11:39:12 oneswig, did you attend the ORC meeting on Thurs after the Summit? 11:39:19 OK, let's aim for that. I'll see if we can get the Intel OpenHPC team involved too. 11:39:27 b1airo: yes I did - with ildikov 11:39:43 i heard you might have been a bit worse for wear... 11:40:03 didn't see you at Karaoke though?? 11:40:09 I went on for high-level discussions with Tim afterwards... 11:40:10 yep, I can confirm oneswig was there too :) 11:40:30 ildikov: sorry to leave abruptly 11:41:01 Tim said he thought it was an improved meeting 11:41:12 oneswig: all good, everyone was leaving quickly after the group photo 11:41:39 I must admit, I'm glad I didn't make it to karaoke, could have been far worse... 11:41:54 b1airo: good news I guess, I wanted to ask for feedback :) 11:42:04 by the way on this topic - I was chatting to Resa Reynolds from Cornell, they are one of the partners in this project: https://federatedcloud.org/ 11:43:11 i know Resa through the DellXL HPC community. they were previously using Eucalyptus but are currently making the switch to OpenStack 11:43:13 oh, with the NSF stamp on it, nice 11:43:54 yeah, i wondered whether the ORC chairs have been in touch with them - should be looking to collaborate 11:44:06 ildikov: The impression I got was the initiative wants to arrive at a (distant) goal but should focus more on first steps. Perhaps I'm not active enough in the discussions but I was expecting to see something more like a roadmap. Early days still? 11:44:18 b1airo: not sure, but I can bring this up on one of the upcoming meetings 11:44:20 (in fact the NSF people showed up at their booth when I was there!) 11:45:00 b1airo: oh, nice! 11:45:37 please do ildikov - like oneswig I feel a bit too removed from the conversation to put my oar in and don't have the bandwidth either 11:45:41 oneswig: something like, I think we would like to set up a goal to agree on and encourage more organizations/institutions to get involved and set up that roadmap 11:46:02 ...i wonder whether we could convince them to piggy-back on our meetings every so often 11:46:24 oneswig: or as they operate now that's what seems "doable" to me at this point 11:46:52 ildikov: institutions will want to see evidence of an advantage to the effort of taking part, I guess 11:46:52 however trying to encourage people to set up some achievable items as I think that would be more appealing 11:47:23 b1airo: will do, I will also encourage them to take some notes from the meetings 11:47:58 b1airo: I think I'll fly during the next one and it's almost impossible to figure out what happened during these calls if you don't attend :( 11:47:59 Isn't Martial a co-chair now? 11:48:09 yep, he is 11:48:37 so there should be a close connection besides me 11:48:44 the teleconference thing is not great 11:48:56 I think he usually attends this meeting when it's in his TZ 11:49:22 b1airo: there are communities that handle it, but they take at least some notes on IRC or some other format 11:49:22 last time he was up at 2am, as I recall. There are very few things I'll get up at 2am for. 11:49:30 ORC has Slack now 11:49:44 but no real activity there at this point 11:49:50 i also heard some gossip that set off alarm bells for me - was Khalil suggesting the concept of a broker appliance ? 11:49:54 I'll try and see whether I can get some life on it :) 11:50:28 he has a few figures now with stuff like that 11:50:50 I need to look up the slides though, I don't remember the structure that well, maybe oneswig does 11:51:25 There was a hardware appliance in the concept diagram which was not well received. 11:52:05 I think it was more of a gateway - don't remember any reference to a broker. 11:52:25 I remember the gateway impression 11:52:40 I got confused on some of the parts 11:53:08 and I also mentioned that if any SW gets developed we should have an open source reference implementation 11:53:54 we almost got into a don't close out commercial players debate on that one... 11:54:21 hmmm 11:55:02 I don't think we're on a wrong track, they just need heads up on aspects every now and then 11:55:06 IMHO 11:56:14 Okay, I confess I hadn't updated my calendar entry for the time change. Bah. 11:56:37 hi verdurin :-) 11:56:59 verdurin: never mind, we can run for a second hour :-) 11:57:00 verdurin: been there, done that, I missed almost all my meetings for days for the last one... 11:57:16 Definite interest from the RCUK Cloud WG for input to our workshop from the SIG 11:57:22 one of the interesting things about the federatedcloud.org project that Resa mentioned was that they were figuring out some sort of capacity exchange mechanism - from the website: "This project will serve as an important model for campus cyberinfrastructure that other institutions can follow providing flexibility in support of multi-discipline scientific workflows, elasticity by sharing resources, and an allocation model that provides a fair 11:57:22 exchange mechanism for resource access between and across multiple institutions." 11:59:17 OK, alas we are out of time 11:59:32 b1airo: that indeed sounds interesting 11:59:38 b1airo: there's a good deal in common there. Can we get them along to a meeting? 11:59:47 b1airo: would be good to connect with them 12:00:12 b1airo: Whenever I see bartering of resources I think of https://en.wikipedia.org/wiki/Stone_Soup 12:00:54 : I seem to be 1h late when I thought to be on time 12:00:57 OK, time's up, thanks ildikov johnthetubaguy b1airo verdurin 12:01:04 bye 12:01:06 ha, yes a classic oneswig 12:01:11 oneswig: hmmm, didn't know the Hungarian version before :) 12:01:12 ah hello martial_ - alas it seems so 12:01:43 I'm sure you'll be out in force next week :-) 12:01:46 #endmeeting