21:00:34 <oneswig> #startmeeting scientific-wg
21:00:35 <openstack> Meeting started Tue Oct  3 21:00:34 2017 UTC and is due to finish in 60 minutes.  The chair is oneswig. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:00:36 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:00:40 <openstack> The meeting name has been set to 'scientific_wg'
21:00:43 <oneswig> hello
21:01:09 <martial> hello Stig
21:01:11 <oneswig> #link agenda for today https://wiki.openstack.org/wiki/Scientific_working_group#IRC_Meeting_October_3rd_2017
21:01:15 <oneswig> #chair martial
21:01:16 <openstack> Current chairs: martial oneswig
21:01:26 <oneswig> Good day Martial
21:01:38 * ildikov is lurking :)
21:01:54 <martial> Hello Ms Vansca :)
21:01:56 <oneswig> Hi ildikov!
21:02:00 <trandles> hello
21:02:05 <ildikov> oneswig: martial: hi :)
21:02:08 <oneswig> hey trandles
21:02:12 <rbudden> hello
21:02:22 <martial> thanks again for all the wonderful work and networking with the ORC people last week, truly appreciated
21:02:22 <oneswig> hey bob
21:02:43 <martial> Hi Tim, Robert :)
21:03:03 <oneswig> ildikov: thanks for attending ORC
21:03:16 <oneswig> OK, ready to go?
21:03:31 <oneswig> #topic Open Research Cloud: looking for volunteers to lead working groups
21:03:35 <ildikov> oneswig: martial: it was my pleasure, I always like to learn more about this community and that effort
21:03:41 <martial> I guess the ORC is the first topic of conversation, Khalil was going to try to join tonight but is unable, so I will paste some content I wrote and let Tim and Ms Vansca add to
21:03:55 <martial> First, what it is: The Open Research Cloud (ORC) is a collaboration of the international community supporting scientific research computing. Having recognized that a number of significant obstacles continue to interfere with the ability of globally dispersed researchers to effectively collaborate. It focuses on the federation of data, identity, security, shared compute and storage; and an interoperable cloud services management platform.
21:04:02 <martial> #link http://www.openresearchcloud.org/
21:04:16 <martial> The ORC just completed its second Congress hosted in Amsterdam last week, started to collect and curate the draft declaration
21:04:38 <martial> #link https://docs.google.com/a/datamachines.io/document/d/1nqlaFOWUX8jbmlGsDRCigVphaapaQl5M5C3kHS1VJOo/edit?usp=sharing
21:04:46 <martial> (the above link is a copy as of 2017/10/02 and read-only)
21:04:57 <martial> In particular, the ORC is looking to undertand what are the fundamental requirements needed  to create a succesful effort
21:05:17 <martial> ^ this in particular is important for people interested in helping
21:05:25 <martial> The ORC is looking for individuals to volunteer and to get involved in its Working Groups by committing time to some specific working groups where they feel their expertise will be well utilized.
21:05:38 <martial> For examples: conversations and communications on the topics of Identity, access to data (granting, provisioning, authentication), and the use of computing resources (allowing access to, data locality).
21:05:53 <martial> To do so, please add contact Khalil Yazdi <khalilyazdi@outlook.com> or Enol Fernández <enol.fernandez@egi.eu> and sign up for any and all group email lists here:
21:05:53 <martial> #link https://groups.google.com/a/openresearchcloud.org/forum/#!forumsearch/
21:06:03 <martial> (end of paste)
21:06:31 <martial> I was discussing the outreach with Khalil yesterday
21:06:54 <martial> and I plan to post this to our SWG ML as well as ask people who are interested to share it
21:07:10 <martial> Tim / Ms Vansca, anything to add re:ORC?
21:07:34 <ildikov> martial: thank you, I think this is a great summary
21:07:35 <oneswig> martial: I think that's a good idea to widen the discussion
21:07:49 <trandles> not really from me, I'm just pinging folks trying to get some more potential stakeholders engaged
21:08:04 <martial> thanks Tim, this is very useful
21:08:30 <ildikov> the effort is growing global and the intent is to involve more people and organizations from Asia and Australia as well onwards
21:08:33 <martial> I will work with Khalil on finalizing the language and provide a version of this text to share
21:08:41 <oneswig> how can ORC build upon, and get momentum from other projects in this area?  Was that discussed in Amsterdam?
21:09:01 <ildikov> the plan is also to have a draft by the time of the Summit and the next Congress currently planned for Sydney
21:09:36 <martial> this was discussed more I think at the conversation we had yesterday on the phone, but yes, they want to expand, and they hope to have the expertise and volunteer
21:10:32 <martial> one of the big plan is to make more an International ORC Organization
21:10:52 <martial> (the first and last words are important)
21:11:31 <martial> we have weekly teleconferences at 11am EST on Mondays
21:11:40 <ildikov> there is another thing that I heard much last week in the room, which was to get big commercial providers into the room
21:12:30 <trandles> As far as I know only Microsoft has been active in ORC
21:12:46 <ildikov> trandles: that was my take from last week as well
21:13:00 <oneswig> ildikov: what's in it for them?  Abstracting the cloud supplier might be against their interests
21:13:15 <martial> Box and Cisco also presented their solutions
21:14:06 <ildikov> oneswig: there were comments like the researchers should spend time on research as opposed to managing infra in 90% of their time
21:14:11 <martial> I think it will have to come to similar term as the OpenStack foundation when a legal status is created: you plan to release the work to the foundation (or whatever status the ORC end up having)
21:14:13 <jmlowe> The first meeting was very vendor heavy in Boston
21:14:35 <ildikov> oneswig: I'm also less aware on where funding structures are evolving in the future
21:15:09 <oneswig> ildikov: show us the money, lebowski...
21:15:16 <ildikov> I didn't mean to throw a rock into the lake, but I felt to mention this as something I heard multiple times during the two days
21:15:39 <jmlowe> To a large degree for US researchers the funding agencies, which have not been participating, decide what resources researchers use
21:15:41 <ildikov> oneswig: lol, I've just got introduced to that nice piece of art this summer :)
21:16:29 <oneswig> ildikov: nice!
21:16:33 <martial> can not fault that and we are trying to figure this structure as we form the declaration and the groups
21:16:41 <oneswig> brb
21:16:47 <ildikov> jmlowe: so it can be on that side whether there's a policy for open source or they will go with any (specific) vendor?
21:17:38 <martial> ildikov: my belief is that it will have to grow into a sharing agreement between collaborating partners
21:17:54 <martial> ildikov: kind of like the OpenStack terms
21:18:04 <martial> (for contributing)
21:18:12 <ildikov> martial: that sounds reasonable as much as I can see this area
21:18:16 <jmlowe> Well, the funding agency could put that policy in place, but the program officers need to be engaged
21:18:36 <flanders_> +1
21:19:05 <ildikov> there are also activities where we can join and emphasize collaboration by having a broad forum as for the declaration itself
21:19:25 <ildikov> jmlowe: thanks for clarifying
21:19:29 <martial> jmlowe + flanders_: we are looking at forming the working groups as it is, firming the declaration, and creating a legal entity to represent the effort
21:20:00 <martial> ildikov: I look forward to seeing this happen
21:20:27 <ildikov> martial: me too
21:20:41 <ildikov> martial: let me know how I can help
21:20:51 <martial> so to sum it up, if you are interested in seeing this happen, I would ecnourage you to contact Khalil and Enol and/or join the mailing list or weekly teleconference
21:21:11 <martial> Khalil Yazdi <khalilyazdi@outlook.com> or Enol Fernández <enol.fernandez@egi.eu>
21:21:16 <flanders_> Is Wilfred still a chair?
21:21:19 <martial> https://groups.google.com/a/openresearchcloud.org/forum/#!forumsearch/
21:21:26 <martial> flanders_ : I believe so
21:22:55 <oneswig> More to add here?
21:23:25 <martial> There is a planning meeting for the Sydney Summit and ORC next week
21:23:35 <martial> flanders_?
21:24:02 <martial> I know it is 2am EST :)
21:24:56 <martial> #link https://etherpad.openstack.org/p/open-cloud-declaration-sydney
21:25:04 <ildikov> I have invites in my calendar, which indeed don't seem US friendly :)
21:25:12 <flanders_> We've yet to agree a room for this activity and need to decide if it happens at the summit or request susceptible from UTS.  Pending format of event. NB it shouldn't pull away from scientific-wg activities imho.
21:25:49 <ildikov> flanders_: +1
21:25:50 <flanders_> Susceptible=rooms
21:26:31 <martial> flanders_: I know you will keep us in the loop, and that some of us will attend :)
21:26:41 <ildikov> flanders_: assuming there are some activities during the Summit/Forum it would be great to have either less overlap or co-located space
21:26:59 <flanders_> Please do let me know if you would like me to add you to calendar item :)
21:27:11 <flanders_> Thanks martial
21:27:14 <martial> my personal concern is that they both overlap heavily
21:27:38 <flanders_> +1 though coming at it from top-down vs bottom-up?
21:27:47 <martial> (the ORC / Summit)
21:28:25 <martial> stig, we ought to move to the next item, still
21:28:51 <oneswig> martial: lets do that
21:28:55 <ildikov> martial: flanders_: I will send out a reminder to the ORC list when we have the Forum schedule out to further encourage collaboration
21:29:19 <ildikov> oneswig: yep, we can move on :)
21:29:20 <oneswig> #topic Discussion - a convention for image names between federated clouds
21:29:26 <martial> ildikov thank you Ms Vansca
21:29:51 <flanders_> Thanks ildikov
21:29:52 <trandles> well, this next topic sounds like something that's perfect for the ORC declaration  :P
21:30:04 <oneswig> ok, I had a side-discussion with Kalle Happonen as a part of the ORC meeting
21:30:36 <ildikov> trandles: +1 :)
21:30:41 <oneswig> He said, their group at CSC in Finland found a pain point was inconsistency on simple stuff like image naming between clouds
21:31:13 <oneswig> So they thought of putting a stake in the ground and saying, "heres your standard CentOS image" and "heres your standard Ubuntu image', etc
21:31:42 <oneswig> #link The CSC proposal for a convention on image naming https://github.com/CSCfi/CloudImageRecommendations
21:32:15 <oneswig> it's a punctuation mark or so from most people's naming, so why not harmonise?
21:32:36 <oneswig> Or even harmonize, for the Americans :-)
21:33:39 <clarkb> keep in mind that that naming scheme doesn't work with how many clouds want to do it at least in the public space (they like timestampingthem and it causes allthe problems as a user)
21:33:45 <trandles> We've found that simple base image naming is an easy one to solve, much like that document describes.  It gets hairy when folks build images for/with specific complex applications installed.
21:33:46 <oneswig> The question for me is, is this a human-level construct?
21:34:17 <oneswig> clarkb: that's true, we timestamp all the n-1 editions of our images.
21:34:45 <trandles> and it's a discussion (bordering on religious war) when folks start to discuss how to handle image versioning and curating
21:35:02 <trandles> whether it's timestamping or some other meta-metric
21:35:06 <oneswig> but not the nth, latest.  I was questioning our rationale but seems to make sense
21:36:20 <oneswig> trandles: agreed - "lets all do something together" works so long as everyone believes it's their thing we'll all be doing...
21:36:42 <clarkb> we;ve also discovered that an ubuntu 16.04 here and one over there are often significntly different
21:36:50 <clarkb> hardware enablement kernels here not there, etc
21:36:57 <clarkb> (so we just build and ship our own images now)
21:36:59 <martial> oneswig: this is when I want to be a "funding agency" and say ... "fix it"
21:37:44 <trandles> clarkb: +1  - just because we agree on "centos-7" doesn't and shouldn't imply equivalency
21:37:46 <oneswig> clarkb: obvious now you've said it, but I hadn't considered that.  Good point.
21:38:32 <trandles> not to air too much dirty laundry, but when the emacs user found out that "centos-7" only included vi, feelings were hurt
21:38:56 <martial> trandles: this is why centos is dead to me ;)
21:39:28 <oneswig> trandles: ouch.  Suspect it went super-critical?
21:39:46 <martial> but joking aside (emacs power), tim is right about this kind of things being religious wars ... fast
21:40:11 <trandles> oneswig: not that bad, but it could have spiraled if one side didn't just give in
21:40:11 <oneswig> We add properties to our images, I wonder if that's enough to get a first-order approximation of consistency?
21:40:37 <oneswig> trandles: you mean the emacs user said "hey, vi's actually pretty good", right?
21:40:53 <trandles> oneswig: haha, sure ;)
21:41:46 <oneswig> It's the kind of discussion at an OpenStack level where I wonder if ORC could widen to other cloud infrastructure.
21:42:15 <martial> oneswig: the ORC was planned as cloud agnostic, so hopefully yes
21:42:24 <trandles> the combinatorics on the distro packages is too big, properties might not be rich enough to describe the level of detail people might want, maybe <gasp> centers should be prepared to document their images in some standard way
21:42:48 <oneswig> OK, I thought I'd throw that one out there
21:43:06 <ildikov> oneswig: I think that would be part of the idea there, yes
21:43:13 <oneswig> I think there's a discussion to be had.  I think EGI do this across multiple infrastructures, for example.
21:43:35 <trandles> perhaps ORC should define a metadata field for images that is a hyperlink to documentation for that image
21:44:06 <oneswig> trandles: interesting idea.  A layer of indirection
21:44:30 <martial> trandles: we are always interested in people participating in the Working Groups. so this is a great idea to share on the ML
21:45:08 <oneswig> OK, we should move on
21:45:20 <oneswig> #topic Cloud for Science and Engineering - new book and resource portal
21:45:58 <oneswig> Not much to cover here: I put a chapter on OpenStack into a book on cloud for scientific compute.
21:46:27 <oneswig> The nice thing is, the authors have created a website and portal so the book becomes like a 'living document'
21:46:43 <oneswig> #link cloud for science and engineering https://cloud4scieng.org/
21:46:52 <ildikov> oneswig: that sounds pretty cool :)
21:47:08 <oneswig> Hopefully we can all enjoy people's additions here :-)
21:47:26 <martial> congratulations
21:47:37 <oneswig> thanks martial :-)
21:48:28 <oneswig> I think the book has only just come out - you can download it digitallly for free from that site.
21:48:36 <oneswig> Has a lovely retro TeX feel
21:48:51 <oneswig> OK, lets move on
21:49:01 <oneswig> #link OpenStack-OpenHPC project code published
21:49:07 <oneswig> #item OpenStack-OpenHPC project code published
21:49:10 <oneswig> I mean...
21:49:15 <oneswig> #topic OpenStack-OpenHPC project code published
21:49:26 <oneswig> late in the day...
21:49:41 <ildikov> oneswig: I'm with you on that!
21:49:50 <oneswig> Even later for you I guess?
21:50:12 <oneswig> #link OpenHPC / OpenStack on Github https://github.com/hpc-cloud-toolkit/ostack-hpc
21:50:38 <oneswig> The team at Intel who have presented previously on this have released their code
21:50:57 <ildikov> oneswig: yep, one hour ahead
21:51:06 <oneswig> First seen (by me) last November at the IU booth at SC, thanks to jmlowe
21:51:41 <jmlowe> Yes, they are giving an update in our booth at sc this year
21:51:47 <oneswig> There's an opportunity here to get all the OpenHPC + OpenStack efforts out there under one roof
21:51:52 <jmlowe> along with lustre mounting in openstack vm's
21:51:57 <oneswig> So get your PRs in :-)
21:52:21 <oneswig> jmlowe: is that from the LUG workshop at IU earlier in the summer?
21:52:23 <rbudden> jmlowe: +1, i have some ongoing work for potentially Lustre/Manilla
21:53:04 <hogepodge> hi everybody! (52 minutes late... conflicting meeting)
21:53:19 <jmlowe> I think we have the 11:00am Tuesday slot in our booth, need to double check
21:53:40 <oneswig> Hi hogepodge - damn those 52-minute meetings
21:53:54 <oneswig> rbudden: jmlowe: have you seen this presentation https://docs.google.com/presentation/d/1kGRzcdVQX95abei1bDVoRzxyC02i89_m5_sOfp8Aq6o/edit#slide=id.p3
21:54:00 <trandles> hogepodge: -1  I'm in here, on a concall, and listening to a webex  (and have very little idea what's going on in any of them)
21:54:34 <rbudden> oneswig: i have not, will check it out
21:54:36 <hogepodge> trandles: I was on a concall too, but it was a really exciting meeting. The michael bay transformers of meetings. :-)
21:54:48 <trandles> LENS FLARE
21:55:13 <oneswig> ??
21:55:16 <jmlowe> oneswig: I have not, thanks, btw I'm totally stealing your ligo hook from your CSCS talk when rbudden and I go visit trandles in two weeks
21:55:39 <oneswig> jmlowe: get in there, nobel prize winners and all now :-)
21:56:14 <oneswig> How come you're heading to Arizona?
21:56:23 <oneswig> Or New Mexico I guess?
21:56:25 <jmlowe> oneswig: I may need your cheesiest head shot to give you credit on a slide
21:56:34 <hogepodge> sorry to drop this in at the last minute, but I need mailing address for stickers and eventual mailing address (hotel I'm guessing) for books for SuperComputing (Denise and Kathy approve this message)
21:56:58 <oneswig> hogepodge: Can they be addressed to a booth?
21:57:04 <rbudden> hogepodge: you can send them to PSC or IU or both
21:57:05 <jmlowe> oneswig: trandles has some directors wanting an overview of cloud computing
21:57:05 <martial> hogepodge: I will email you my info
21:57:06 <oneswig> Seems sketcy
21:57:25 <hogepodge> Stickers are lighter and can be carried by hand, so we want to send them out earlier.
21:57:47 <jmlowe> hogepodge: what rbudden said
21:57:57 <hogepodge> Books will come from the publisher or our warehouse, and are better timed for the event. But I think we can try to do both in one shipment. I'll check.
21:58:52 <trandles> oneswig: as far as jmlowe and rbudden know, the invite to LANL is totally on the up-and-up
22:00:29 <jmlowe> this does sound a bit like the wind up for a horror movie
22:00:42 <oneswig> OK better close up for the week - on that happy note :-)
22:01:05 <oneswig> jmlowe: next time, you'll have added bit errors
22:01:11 <martial> good meeting :)
22:01:15 <oneswig> #endmeeting