21:00:42 <oneswig> #startmeeting scientific_wg
21:00:43 <openstack> Meeting started Tue Jul 11 21:00:42 2017 UTC and is due to finish in 60 minutes.  The chair is oneswig. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:00:44 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:00:47 <openstack> The meeting name has been set to 'scientific_wg'
21:00:53 <oneswig> hey hey hey
21:00:59 <oneswig> #chair martial
21:01:01 <openstack> Current chairs: martial oneswig
21:01:01 <rbudden> hello
21:01:05 <oneswig> Hi bob
21:01:11 <hogepodge> o/
21:01:13 <martial> Hi Stig; I was not sure you were around :)
21:01:20 <oneswig> Hi hogepodge martial
21:01:22 * cdent passes the bourbon on to this meeting
21:01:25 * cdent waves
21:01:32 <oneswig> Cheers cdent :-)
21:01:36 <martial> (I was checking the channel name to be safe :) )
21:01:40 <armstron_> Hello everyone
21:01:59 <oneswig> greetings armstron_
21:02:09 <priteau> Hello
21:02:09 <oneswig> #link Agenda for today is https://wiki.openstack.org/wiki/Scientific_working_group#IRC_Meeting_July_11th_2017
21:02:17 <oneswig> Evening priteau
21:02:36 <oneswig> So we have only a short agenda for today but a couple of useful items.
21:02:42 <lizhong> Hello
21:02:50 <oneswig> #topic SC2017
21:02:56 <oneswig> Hi lizhong, welcome
21:03:15 <oneswig> SC2017 is the week after OpenStack Sydney
21:03:20 <oneswig> Call in on your way home :-)
21:03:25 <trandles> hi everyone
21:03:30 <oneswig> Hi trandles
21:03:45 <trandles> oneswig: I'm planning on going to Sydney then straight on to SC17 before home
21:03:53 <trandles> if I can help with either
21:03:54 <oneswig> Last year I think the books and stuff were pretty handy - thoughts anyone?
21:03:59 <martial> stig: I will add an item to the agenda as #4 :)
21:04:08 <oneswig> can't wait!
21:04:08 <rbudden> agreed, the books were great
21:04:34 <rbudden> i should be at SC as well and can help with booth/talks/books/etc.
21:04:38 <oneswig> hogepodge: is there a possibility for printing an update of the book, similar to last time?
21:04:59 <hogepodge> Definitely
21:05:22 <oneswig> cool - do you know how we can organize that?
21:05:23 <jmlowe> Hi everybody
21:05:27 <oneswig> Hi jmlowe
21:05:36 <hogepodge> We can't send staff this time, mainly because it's the week after the summit, but we want to support folks from the community who can go
21:05:57 <martial> trandles: lukcy you on Syndey -> Denver, as per our email discussion, I am in Limbo on that one for now
21:06:10 <oneswig> Likely it'll be my colleague John from our gang, and I'll take Sydney
21:06:26 <trandles> sounds lucky now, we'll see what state I'm in come mid-SC17 week ;)
21:06:38 <armstron_> is it late to submit a paper to SC2017?
21:06:56 <trandles> armstron_: paper submission is well past :(
21:07:03 <oneswig> #link deadlines for SC2017 are here https://submissions.supercomputing.org
21:07:28 <oneswig> I think it's gone but you can still do posters and BOFs
21:07:52 <martial> hogepodge: I am going to check with GWU, I know they usually go and a past contributor of ours might go, maybe they can help
21:08:03 <jmlowe> Do you think we would be successful with another BoF?
21:08:08 <oneswig> GWU?
21:08:20 <martial> stig: George Washington University
21:08:26 <oneswig> ah ok, thanks
21:08:27 <armstrong> Ok I will prepare a poster
21:08:29 <jmlowe> I was under the impression they wanted year over year variety
21:08:42 <martial> stig: a PhD student worked with us in the past
21:09:04 <oneswig> I think it's worth a shot.  Can't remember much about the BoF from last time - it was a blur.
21:09:26 <oneswig> I do recall it was busy, and I do recall most people attending were planning to try openstack rather than actual users
21:09:36 <hogepodge> The bof was well attended
21:09:38 <oneswig> So perhaps we should check in with them and see how they're getting on
21:10:06 <hogepodge> martial: great. Are there any organizing etherpads for the event?
21:10:47 <martial> hogepodge: not yet, to my knowledge
21:11:50 <martial> #link https://etherpad.openstack.org/p/SWG-SC17
21:11:56 <martial> Created one :)
21:12:06 <martial> Blair?
21:12:10 <oneswig> Thanks martial, on fire today :-)
21:12:21 <b1airo> Hello!
21:12:23 <hogepodge> For the books, I think we'll be writing with Denise. We should start soliciting for new content now
21:12:23 <martial> #chair b1airo
21:12:24 <openstack> Current chairs: b1airo martial oneswig
21:12:42 <martial> blair: I thought you were on vacation? :)
21:12:47 <hogepodge> Kathy is on a leave of absence
21:13:01 <oneswig> G'day b1airo, how's the pancakes?
21:13:47 <oneswig> hogepodge: OK thanks, can you get some deadlines from Denise we can work to?
21:13:57 <b1airo> Getting started with a porridge entree oneswig
21:13:59 <hogepodge> Will do
21:14:54 <oneswig> Thanks hogepodge - that's BOF and books covered.  Anything else?
21:15:32 <jmlowe> I'm fairly confident I can get slots for booth talks again
21:15:44 <rbudden> jmlowe: +!
21:16:02 <oneswig> jmlowe: that went pretty well, I remember some awesome demos :-)
21:16:02 <rbudden> +1
21:16:09 <rbudden> are we planning more short talks again?
21:16:11 <jmlowe> I think rbudden also said he can get space
21:16:23 <martial> I think we can demo the short talks at the Summit :)
21:16:27 <rbudden> yes, we can use the PSC booth as well
21:16:29 <b1airo> Re. book, do we have any new sections possible or looking at a refresh of the existing?
21:16:54 <oneswig> b1airo: If somebody who knows was to pick it up, I'm sure a federation chapter is within reach
21:17:14 <jmlowe> I might be on the hook for doing a globus auth openid connect keystone federation
21:17:14 <oneswig> Otherwise, some refresh is needed I'd guess
21:17:31 <oneswig> jmlowe: I heard that, well volunteered :-)
21:17:52 <martial> jim: lots of keywords there :)
21:18:23 <rbudden> I’m in the middle of an Ironic refresh Liberty -> Ocata at the moment, possible I could have refresh on that use case
21:18:25 <oneswig> What we'd want to do is find someone else (eg, Nectar, CERN, etc) with a different case study, and combine them
21:18:33 <hogepodge> Federation chapter would be way cool
21:18:54 <oneswig> hogepodge: certainly would!
21:18:54 <martial> I can talk to my colleague at NIST, Robert Bohn, who is involved in a Federation effort
21:18:58 <rbudden> +1 on Federation
21:19:17 <rbudden> jmlowe: are we still going to attempt a federation setup at some point?
21:19:20 <martial> he was there in Boston and presented at the Open Research Cloud meeting
21:19:30 <jmlowe> yeah, I'd like to
21:19:33 <oneswig> Ok, lets try to make it happen - hogepodge tell us by when
21:19:34 <rbudden> probably not enough time to get it together for the book for SC, but just random thoughts ;)
21:20:37 <martial> jim: added an action item with your name on it on the Etherpad :)
21:20:51 <oneswig> any takers for filing the BoF submission?
21:21:28 <rbudden> sure, I can take that
21:21:30 <b1airo> rbudden: do you have architecture diagrams of your setup between SLURM and OpenStack? I've described it to other HPC people who have been interested...
21:21:40 <oneswig> It's probably worth submitting from a widely spread group, to add some gravitas - but someone's got to stump up the wording
21:22:04 <oneswig> b1airo: ooh, I'm doing some on that recently.
21:22:06 <b1airo> oneswig: suggest we start from last year's submission and update?
21:22:12 <martial> bob: can you use the etherpad to draft the wording and let us know if you want us to add/extend ?
21:22:12 <oneswig> Got heat stacks that'd make you scream
21:22:13 <rbudden> b1airo: we have some arch diagrams but i’m not sure they’re openstack related
21:22:18 <rbudden> more networking, etc.
21:22:58 <rbudden> but i’d be happy to talk to ppl that are interested in what it looks like
21:22:58 <jmlowe> I can go in with rbudden on the BoF
21:23:03 <oneswig> b1airo: good idea on starting from last year.  er... anyone got a copy?
21:23:09 <b1airo> I'm looking for stuff I can adapt into slideware for OpenStack+HPC talks
21:23:30 <rbudden> starting from last year’s submission would be great ;)
21:23:31 <oneswig> b1airo: gotta mention the SKA if this is at Pawsey
21:23:57 <martial> stig: I can also reach out to Mike to see if he can formalize his presentation at the last lighting talk into a book chapter
21:24:03 <b1airo> I always tryvto talk about different ways of doing things but it's generally very hand wavy and I'd like to make it clearer for the HPC Advisory Council tal
21:24:08 <rbudden> b1airo: if you want to ping me offline we can chat about the setup
21:24:29 <b1airo> thanks rbudden , will do
21:24:48 <oneswig> b1airo: graphs are good for that kind of talk.  preferably with the axes labelled
21:25:15 <b1airo> I'll find the BoF submission, let me right with Google Drive for a moment - if I don't come out can someone please call Anonymous to find me...
21:25:32 <b1airo> *fight with
21:26:41 <martial> while blair is looking this up, should we see if there is anything else to discuss re: SC17?
21:27:11 <rbudden> social?
21:27:26 <rbudden> the few times we’ve been able it seems to be a hit
21:27:36 <oneswig> true
21:27:37 <rbudden> even if it’s just informal
21:27:45 <rbudden> i don’t want to throw trandles under the bus ;)
21:27:45 <martial> you mean for the summit or sc17?
21:27:48 <oneswig> I also added an item on booth space for lightning talks
21:27:52 <rbudden> SC17
21:28:05 <rbudden> unfortunately I don’t think i’m going to be able to acquire funding to get to Sydney :(
21:28:34 <martial> bob: welcome to my club
21:29:23 <b1airo> They're just worried you won't return!
21:29:25 <martial> bob: I would tell you that Mr Hoge was kind enough to remind me that the foundation has a travel aid
21:29:36 <martial> (assistance)
21:29:53 <martial> not sure if it would work, but it is worth trying if you are presenting
21:30:05 <rbudden> martial: true, kinda forgot about that. i’d consider it, but would want to be presenting or something to put it to good use
21:30:11 <b1airo> Here's the panel draft, I think we just adapted / used the same text...
21:30:16 <b1airo> https://docs.google.com/document/d/19kjs_HQ2bQw4ydjqY6ECMSBlAq6w2N0BELrxHtSfzn0/edit?usp=drivesdk
21:30:21 <rbudden> yeah, not sure if i can make the presentation deadline, we’ll see
21:30:21 <oneswig> My instinct is an evening social at SC17 could be a big table (or two) for a meal and beers.
21:30:25 <oneswig> Good job b1airo
21:30:32 <martial> #link https://docs.google.com/document/d/19kjs_HQ2bQw4ydjqY6ECMSBlAq6w2N0BELrxHtSfzn0/edit?usp=drivesdk
21:30:40 <oneswig> Reminder: deadline for Sydney presentations is THIS FRIDAY
21:31:00 <martial> stig: shoot
21:31:04 <rbudden> oneswig: ugh
21:31:10 <rbudden> thanks though, good to know
21:31:24 <oneswig> so plenty of time, right?  I've got three on the go!
21:32:40 <b1airo> I haven't proposed anything myself but I think I'm featuring in three vendor submissions :-P
21:33:10 <martial> :)
21:33:14 * rbudden counts hours left to work on ironic/ocata and contemplates if sleep is futile ;)
21:33:16 <oneswig> ... nvidia, dell, mellanox?
21:33:27 <martial> okay, 30 minutes to go and 3 topics left :)
21:33:43 <oneswig> martial: good point , which brings us to...
21:33:44 <armstron_> how is the presentation evaluated? like conference papers?
21:33:51 <oneswig> #topic Application catalog(ue) updates
21:34:11 <oneswig> armstron_: exactly like that - a review committee
21:34:22 <jmlowe> I'm sitting in the cloud and container image bof at PEARC right now
21:34:42 <oneswig> jmlowe: what's PEARC?
21:34:59 <jmlowe> it was the xsede conference but was renamed
21:35:14 <jmlowe> pearc.org
21:35:40 <trandles> sorry, I stepped away for 10 mins...RE: social...ugh ;)
21:36:00 <oneswig> trandles: you've done your time on that!
21:36:01 <jmlowe> yeah, trandles you didn't say anything so you have to do the social again :)
21:36:16 <trandles> +1 oneswig   -1 jmlowe
21:36:25 <armstrong> oneswig: Thanks, I will be submitting a paper this Friday.
21:36:29 <oneswig> ah, for a united front
21:36:39 <trandles> that said, we go to Denver several times a year, so I'm happy to make suggestions but organizing might be a step too far
21:36:59 <oneswig> OK - app catalogues - there has been some recent discussion here
21:37:17 <martial> tim: I think Jim was volunteering in truth :)
21:37:48 <oneswig> Some interest because the OpenStack.org app store is going EOL and there may be a brighter future for scientific apps
21:38:12 <oneswig> b1airo mentioned that Nectar have a well-stocked instance of Murano
21:38:25 <oneswig> I think last week there was some interest from CERN
21:39:03 <oneswig> Earlier tonight we deployed Murano on our Kolla dev system
21:39:12 <oneswig> Is there a mechanism for sharing these things?
21:39:33 <oneswig> rbudden: jmlowe: I recall you guys exchange apps for Murano, right?
21:39:45 <b1airo> I think like Tim suggested, simply GitHub?
21:40:02 <oneswig> jmlowe: just noticed you're in New Orleans.  Is this bof in a piano bar?
21:40:45 <oneswig> b1airo: right, that makes sense
21:40:50 <rbudden> oneswig: not yet, that was the idea for an end goal of a proposal we have in
21:41:18 <rbudden> jmlowe: that reminds me we should chat offline about the next steps with the PIF
21:41:26 <jmlowe> oh, nice, we are still waiting to get paid to make a well stocked catalog
21:41:47 <oneswig> rbudden: jmlowe: in your experience, how portable are Murano apps?
21:42:36 <jmlowe> oneswig: no, but they did chase us out of breakfast with a jazz band
21:43:26 <rbudden> oneswig: i’m not sure how much we (i) can attest to the portability. what was largely done was a POC setup using devstack to flesh out the ideas of using Murano for OpenStack integration
21:43:46 <b1airo> I guess they are similar to Heat templates, portability wise, with possible extra Murano version requirements
21:44:00 <oneswig> I remember it sounded much fancier than that in the presentation abstract :-)
21:44:06 <rbudden> so we didn’t exactly test across OpenStack versions or even Murano version for that matter
21:44:29 <jmlowe> I definitely had trouble pulling the aps.openstack packages due to backing images
21:44:51 <rbudden> think of the presentation as selling the idea ;) backed by some usage to show the idea was feasible :)
21:45:26 <rbudden> in all honesty it was slightly more difficult than I originally imagined, but that have been due to my expectations going in
21:46:07 <oneswig> rbudden: I get it, might be something where the feasibility of sharing is learned through experience
21:46:20 <jmlowe> This bof has some of it's roots in that talk, if we could convince our learned openstack colleagues, then we could sell this to xsede as a viable project
21:47:13 <oneswig> ... anyone else in the room with jmlowe who can help out with that? :-)
21:47:14 <jmlowe> I think we both struggled with the ui elements, confirms my belief that I need to be kept away from end users
21:47:54 <oneswig> I might be in the same boat on #scientific-wg come tomorrow
21:48:17 <oneswig> OK, I'll gather these thoughts and follow up
21:48:36 <oneswig> shall we move on to the next topic - martial what have you got for us?
21:48:42 <martial> stig: maybe follow up on this next week then
21:48:49 <oneswig> indeed
21:49:04 <martial> stig: well there is topic #3 AOC unless I missed that?
21:49:17 <martial> make that AOB
21:49:23 <rbudden> oneswig: i’m not at PEARC but I’m in the same boat and can help jmlowe
21:49:43 <oneswig> martial: that's any other business - put it at the end
21:50:05 <martial> ahhh well then #4 is a #3 then
21:50:15 <martial> #topic Introducing Conducere and DMoni
21:50:39 <oneswig> aha!
21:50:50 <trandles> gotta run to another meeting folks, ttfn  o/
21:50:51 <martial> So after many things to clear, I am happy to give you a link to the software we promised for a bit
21:51:02 <martial> #link https://github.com/usnistgov/Conducere Conducere
21:51:10 <martial> #link https://github.com/usnistgov/DMoni DMoni
21:51:38 <martial> Lizhong, Maxime and Pooneet are here I believe, they helped make this release
21:52:12 <martial> it is still a little rough but it is public, and I know a few people have been interested to get their hands on it
21:52:47 <martial> so for those interested, please test it, let us know what you thing and how we can make it better and thank you for letting me announce it here :)
21:53:03 <rbudden> apologies, i have to go AFK as well… little man just got back from daycare
21:53:20 <martial> bob: no worries, thank you for spending time with us :)
21:53:50 <martial> as you can guess, I am very interested in knowing what people think :)
21:53:53 <oneswig> martial: looks great, I could start using conducere tomorrow!
21:53:58 <b1airo> Bye rbudden
21:54:40 <martial> stig: we welcome users and contributions obviously
21:55:17 <oneswig> how does packer compare with disk-image-builder?
21:55:28 <martial> stig: I was really worried that I would not be able to release it before this meeting, but I got the final okay late this morning, so we have scrambled to make it "finished"
21:56:08 <b1airo> Packer came first?
21:56:21 <clarkb> oneswig: that came up in the infra channel earlier today in a different context. A big difference is dib lets you build one image and push it everywhere. packer is more snapshot based and has a dsl rather than just letting you execute code
21:56:26 <oneswig> In the heat templates, would I be able to provision groups of nodes of different types?  I was working on a problem earlier (in Dask) where a few nodes might need to be high memory, or example
21:57:01 <oneswig> clarkb: thanks, that's interesting
21:57:13 <martial> I am not sure about Packer vs disk-image-builder
21:57:35 <jmlowe> ok, BoF over gtg
21:57:38 <martial> stig: I do not see why we could not mix flavors in heat
21:57:46 <oneswig> we've been hacking on DIB, I'm interested to know the alternatives
21:58:29 <martial> thanks clarkb
21:58:33 <oneswig> Here's what we've been working on
21:58:34 <oneswig> https://galaxy.ansible.com/stackhpc/os-images/
21:58:34 <b1airo> We use Packer for Nectar "official" images, have had to resort to DIB for some Trove stuff
21:58:59 <oneswig> and science-y image elments as extensions https://github.com/stackhpc/stackhpc-image-elements
21:59:00 <hogepodge> Before the meeting end, word is book update may be too much of a lift for us with Kathy out. We're taking about it internally. Sorry to talk out of topic
21:59:26 <oneswig> hogepodge: OK thanks for the update.  A reprint might be manageable though?
21:59:27 <martial> stig: sweet, I wonder if we can mix your solution with ours
21:59:39 <hogepodge> Yeah
21:59:48 <oneswig> martial: my thoughts exactly, I like the apps you've got going!
22:00:09 <lizhong> Stig: Packer can support multiple clould providers (OpenStack, AWS, etc.). Basically, it connenct to the cloud env to create an instance, do configurations and save it as a image to be reused.
22:00:12 <martial> hogepodge: thanks for letting us know. We can still publish a PDF version if we have time
22:00:26 <oneswig> we are out of time, alas
22:00:44 <clarkb> lizhong: right, whereas dib you'd make a single image, convert to N output formats then use it wherever you like
22:00:53 <clarkb> lizhong: for us that is important because we can use the same image everywhere easily
22:01:00 <Maxime_> @oneswig About the flavors, you should be able to customize it in the heat template
22:01:28 <oneswig> Maxime_: today's case was 3 types of node in the topology.  Can I do that?
22:02:00 <oneswig> I think there's some exciting potential here - thanks for sharing
22:02:18 <oneswig> we need to wind up the meeting.  Thanks everyone!
22:02:23 <martial> stig: we can and will gladly follow up next time if welcome
22:02:27 <oneswig> #endmeeting