15:01:50 #startmeeting openstack-cyborg 15:01:51 Meeting started Wed Nov 8 15:01:50 2017 UTC and is due to finish in 60 minutes. The chair is zhipeng. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:01:52 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:01:54 The meeting name has been set to 'openstack_cyborg' 15:02:04 #topic Roll Call 15:02:09 #info Howard 15:02:31 #info Michele 15:04:01 hey Michele 15:04:19 hi Howard! 15:06:07 let's wait to see if there are other folks coming 15:06:40 crushil ttk2[m] zhuli you guys there ? 15:06:49 yup 15:06:55 yup 15:07:00 daylight savings time tho 15:07:06 we're an hour early by my calender 15:07:07 hi folks, sorry for the late 15:07:08 Yup 15:07:09 or is my calender screwed up 15:07:16 It's DST jkilpatr 15:07:35 no biggie 15:07:46 we can just do a 2 hour meeting and then anyone who wants to show up at the other time can :P 15:08:02 #info Justin 15:08:53 :P 15:09:23 after all we don't want to make Howard have to stay up any later than he is now. 15:09:54 we want the bot to rest as well 15:10:02 lol bots rights 15:10:14 #info RUshil 15:10:19 we've got another 5 years before we have to worry about the meeting bots unionizing zhipeng 15:10:40 jkilpatr if they do we just close business 15:10:57 #topic Sydney Summit Initial Feedback Gathering 15:11:20 So I think I'm very thankful for Richard to step in for us to do the presentation 15:11:31 he even got the joke right (about our meeting time) 15:11:56 for those of you haven't seen the recording 15:12:08 link? 15:12:18 how was attendance? 15:12:35 #link https://www.youtube.com/watch?v=DXA0c62R4jI 15:12:39 awesome guy, is there a video about the presentation, would love to see it 15:12:53 and for presentation slide itself 15:13:23 #link https://docs.google.com/presentation/d/1tERW4CVhyxNdX50AOPZRa44iPEhico8O_vQ0Ou75L80/edit?usp=sharing 15:13:39 crushil I was gonna to ask you, any feedback from Rich > 15:13:42 ? 15:13:51 It went really well 15:13:52 about the attendance, etc ? 15:14:06 thanks for the link, will watch later 15:14:12 It was a lighting talk 15:14:24 So, there were like 15-20 folks there 15:14:30 anyone approached him after the talk ? 15:14:41 Obviously there was no time for questions 15:14:52 That we didn't get a chance to talk about 15:15:04 It was around midnight when he finished 15:15:16 ah ok 15:15:18 I'll catch up with him my evening (his morning) 15:15:27 sounds great :) 15:15:34 Sydney is 15 hours ahead of EST 15:15:48 It's like 3 hours ahead of you 15:15:49 ye we have 3 hour difference 15:16:34 let's try to get together in Dublin next year, for rocky ptg 15:16:50 late Feb 15:17:07 I figure I'll get sent to the PTG, to be honest we get more work done there anyways 15:17:16 might freeze my balls off tho, considering the weather ... 15:17:22 jkilpatr for sure 15:17:53 any other comments re this topic ? 15:18:29 speaking of getting work done, can we try and wrap up the interaction patch? Sorry I haven't gotten my nova hack ready been on PTO 15:19:01 so on that note, moving on to the next topic 15:19:12 #topic Placement-Cyborg Interaction Discussion 15:19:23 so last week crushil made a really good proposal 15:19:40 that before Xmas we should have like weekly video sprints 15:19:48 i asked Sean about the release sched 15:20:00 Feb 5th 2018 is the hard cutoff date for coding 15:20:09 cool 15:20:12 that will be when RC is released 15:20:13 I can refocus on Cyborg last week of November and into December 15:20:41 Wrapping up a lot of other projects in the next few weeks. 15:20:42 so I think let's just concentrate on Nov and Dec and try to hammer everything we need out in this period 15:20:58 jkilpatr no problem 15:21:18 then when would be the dates works for everyone about the sprint ? 15:22:46 crushil jkilpatr zhuli 15:23:55 Nov 27 - Dec 4th short break for Ansible training then Dec 8 - EOY and probably beyond. 15:24:42 I can attend meetings and maybe get some stuff done earlier in November but I' 15:25:01 I'm getting a few racks of gear for release qualification so I need to focus on that when I have it. 15:25:03 how about you crushil ? 15:26:24 I should be ok to start sprint next week. Still wrapping up one project that has been pending for months 15:28:02 then I think zhuli and I could start with crushil first, starting next week 15:28:15 on driver side and traits 15:28:15 since the summit is done, next week is ok for me too 15:28:21 yep 15:28:33 and jkilpatr could jam in later on the agent and conductor parts 15:28:54 then maybe EOY we could have some E2E stuff, hopefully 15:28:54 been looking at Nova don't see any reason why we can't add devices to the whitelist live 15:29:04 there's actually not too much going on with the whitelist at all, just gets loaded and checked 15:29:27 jkilpatr I could also ping Matt if we encounter any problems at Nova side 15:29:48 easier to write code and ask for a review than ask them to do it :P 15:29:54 especially if it's short and easy to review 15:30:22 yep :P 15:30:32 then crushil which day of the week shall we pick ? 15:30:50 I think time wise the meeting time would be ideal 15:31:58 I agree 15:32:19 we could do like the same time but on Tuesday, for example 15:32:45 that's fine too but then what do we talk about during the Wednesday meetings? 15:33:33 ya, that works 15:34:03 jkilpatr to get you update :P 15:34:43 or shall we just use the weekly meeting time ? 15:35:07 I think we should just use the weekly meeting time and be sure to note things in IRC don't want to turn away newcomers in the channel 15:35:32 crushil that ok for you ? 15:36:49 yup 15:37:11 okey great, then I will send out ZOOM links for the video conf 15:37:15 starting next week 15:37:42 jkilpatr regarding your current spec, I will have another set of comment, but I think we could wrap it up this week 15:38:20 sounds good let me know 15:39:23 zhuli, these times good for you? 15:39:45 no problem 15:40:48 jkilpatr last week meeting, don't know if you have comb through the log yet, I have thought about adding another dimension of resource provider 15:41:03 nope haven't read it, care to summarize? 15:41:12 we could implement it later, but I think I should submit a spec about it 15:41:19 so in a gist 15:41:33 link me when you have it up 15:41:36 currently Placement uses Resource Provider to model resources 15:42:19 which consists of two dimensions: traits on the quantative side of things, and resource class for the qualitive side of things 15:42:45 e.g. you could use trait to define what you have and resource class for how much/many you have 15:43:00 however, i think for cyborg to deal with accelerator 15:43:15 purely enumeration would cause problems 15:43:26 since for example for FPGAs 15:43:44 you can't trait and resource class with every VF/PF combination/layout 15:43:50 where you have possible traits that consume the same resource class? 15:44:13 zhipeng, I would have a resource class fpga that was consumed by many traits which would be the VF/PF 15:44:35 jkilpatr my point is that many traits would be really hard to manage 15:45:02 we need a more fuzzy way to describe accelerator resource, besides traits and resource class 15:45:26 what do you propose 15:45:34 so I think propably something like capability, that it could generally map to a set of resource that user would need 15:45:50 instead of specifically id the resource 15:46:01 for example for deep learning 15:46:25 there will be cases that operators just want to run their framworks on OpenStack Infra 15:46:44 doesn't really care about which GPU, FPGA, SoC solution it should use 15:47:15 and then if say, the infra got some GPU good for temporalachitecture 15:47:36 and some FPGA cards that good for spatial architecture 15:48:00 but we need to get more specific at some point? 15:48:03 we should just label these like CAPABILITY: DL/TEM 15:48:15 we will still have traits ands resource class 15:48:17 the way I see it we can just map a buncha traits to the same resource and it will be kinda ugly but work. 15:48:30 ye that is what I think 15:48:56 so my thought was that if we could add another dimension, to use with trait and resource class together 15:49:06 so that operators could do a specific request 15:49:10 or a fuzzy request 15:49:34 because unlike the general computing infras, which could almost be enumerated 15:49:44 accelerators just got so many types and combinations 15:50:09 so we let operators aggregate things when their workload could run on either and schedule agains that? 15:50:32 against* 15:51:06 not necessarily aggreagating, but just select a (super/sub) set of resource that satisfy the fuzzy request 15:51:24 if it deems to be a super set, then it is aggregation 15:51:43 if it just needs a subset, say just some of the FPGA card funcationalities 15:51:53 we could also do that mitigation 15:52:07 still not sure I understand the usecase 15:52:26 I think I will just writeup a spec and we could review it :P 15:52:40 sure, a better outline of the potential usecase would help a ton 15:52:40 I need to elaborate more 15:52:44 yep 15:52:50 okey 15:53:29 #agreed have weekly sprint video conf using team meeting timeslot starting Nov 15th 15:53:47 #topic AoB 15:54:02 I think we have covered the most important ones 15:54:07 any other business ? 15:58:54 okey let's conclude the meeting then, thanks everyone :) 15:58:59 #endmeeting