15:00:57 #startmeeting openstack-cyborg 15:00:58 Meeting started Wed Nov 1 15:00:57 2017 UTC and is due to finish in 60 minutes. The chair is zhipeng. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:59 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:01:01 The meeting name has been set to 'openstack_cyborg' 15:01:08 #topic Roll Call 15:01:15 #info Howard 15:03:27 hi zhipeng 15:03:50 hi shaohe 15:03:51 good evening 15:04:22 what does Howard mean? 15:05:42 my alias lol 15:06:19 got it 15:07:05 Any info? 15:07:19 are we waiting for others? 15:07:49 that is the cmd that will record your name in the minutes 15:08:03 we are waiting for justin and crushil 15:08:22 zhipeng, I am here 15:08:28 Just in another meeting 15:09:07 #info Rushil 15:09:16 no problem :) 15:09:19 zhipeng, Justin is on vacation till this friday 15:09:28 ah okey 15:09:34 then we can just start 15:09:45 he could catch up later with the minutes 15:11:27 any introduce for each? 15:11:27 #topic Sydney Summit slide review 15:11:45 crushil is our core reviewer from Lenovo :) 15:11:55 shaohe_feng_ is from Intel 15:12:13 crushil: hello, I'm shaohe feng 15:12:29 Sorry guys, it’s been a long time since we meet last time, unfortunately ,because I work very late today and I am driving home now, couldn’t pay attention to this meeting, so I will read the log history later and send some emails if needed 15:13:02 zhuli safe driving ! 15:13:03 crushil: from intel. focus on intel's accelerator 15:13:30 hope you are using audio to text ML converter zhuli :P 15:13:40 anyways back to the topic 15:13:49 we have one lightening talk at Sydney Summit 15:14:23 I have sent a simple slide I prepared to ttk[m] and crushil based upon our report to TC back in Denver 15:14:49 so I might add a few today or tmr, but I think it should be ok now content wise for the lightening talk 15:15:19 I made some changes 15:15:45 And I am talking about it to my coworker who will present for us at Sydney 15:16:32 ok gr8t 15:16:41 i think we should have few problem with the slide 15:17:03 and we could make the slide publicly available after the summit session 15:17:08 Ya 15:18:49 okey let's move onto the important topic today 15:19:02 #topic Placement-Cyborg Interaction Discussion 15:19:34 So I have been studying the Placement/Resource Provider/Trait stuff last week 15:19:45 and attended the nova-ironic trait meeting yesterday 15:19:57 so let me first send out some useful links here 15:20:18 #link https://github.com/jaypipes/articles/blob/master/openstack/placement-queens-update.md 15:20:46 #info Jay Pipe's writeup on Placement 15:20:52 this is as the best as you could get 15:21:12 and there are two specs that will be very useful to us, from Ironic 15:21:23 #link http://specs.openstack.org/openstack/nova-specs/specs/queens/approved/ironic-driver-traits.html 15:21:40 #link https://review.openstack.org/#/c/504531/ 15:22:17 #info first one is about Ironic report/accept trait with Nova, second one is about trait config in Ironic 15:22:45 ok there is another spec that will help our modeling, is the vGPU spec from jianghua 15:23:00 #link http://specs.openstack.org/openstack/nova-specs/specs/queens/approved/virt-add-support-for-vgpu.html 15:23:21 and even more interestingly 15:23:32 we could also learn something from the kubernetes discussion 15:23:35 #link https://github.com/kubernetes/community/pull/782 15:23:48 #info k8s discussion on resource class for accelerators 15:24:36 well based upon all the readings i did via the above mentioned links 15:25:06 I think there several points we need to consider when we modeling cyborg maneged resources 15:25:43 #info 1. We need to define CUSTOM_ traits for sure, but also some additions to os-trait lib on FPGA if there are any 15:26:41 #info 2. other than the reporting from cyborg-agent to the Placement API, we need (a) the driver also support trait api and (b) cyborg-api support trait 15:27:09 zhipeng, Ya, I have been following the traits discussion on the Ironic end too. So, I might have some inputs once there is a spec up on our end 15:27:44 #info 3. There is a caveat for Cyborg on resource provider modeling for accelerators which is we need to find a way to model different types of accelerators in a unified way 15:29:17 #info 3a. Traits and resource class are used to define qualitive and quantative side of resource, however we might need a third kind to descirbe accelerators, because only using traits and ResourceClass means we will have a lot them for different accelerators 15:30:35 #info 3b. We might need to consider using something like capability to describe the accelerator resource. For example FPGA and GPU both could be used for ML, a user might just want a VM to run a ML workload with the help of an attached accelerator 15:31:29 #info 3c. This means we need to enable user to be able to select one capable resource instead of identifing very specifically which resource they want 15:31:43 ok that is the end of my monologue lol 15:31:49 any thoughts ? 15:37:32 need time to reading the link, too many info. 15:38:11 ya , any thoughts on my comments above ? besides the links :P 15:43:05 I will chime in once we have specs in our backlog 15:43:36 I have some thoughts about the way we should be implementing traits on our end and handing them over to Nova 15:44:22 crushil that brings mine proposals actually 15:44:23 And that will alter the implementation of the generic driver as well which I am working on currently 15:44:36 i think besides justin's general spec 15:44:46 we will need one on the driver side and one on the api side 15:45:06 would you like to lead on the driver trait spec ? 15:45:11 Sure 15:45:37 When is the stable Queens release date? 15:45:40 zhuli and I will work on the api side 15:45:54 crushil I need to double check on that 15:46:21 I think we need to be reasonable with the amount of work we promise 15:46:30 I would hate to overpromise and underdeliver 15:46:37 #action crushil to work on a new spec on cyborg driver trait support, and zhipeng will work on a new spec on api trait extention 15:46:43 crushil that is for sure 15:48:15 and on that note, let's also start simple 15:48:32 we could cover very basic trait descriptions for Queens 15:48:57 meaning for example not necessarily model FPGA resources in a Tree structure in full 15:49:11 let's deliver something basic in Queens 15:49:42 zhipeng, Exactly with at least a shim that vendors can build upon 15:49:56 crushil +1 15:50:13 okey any more comments on this topic ? 15:50:38 folks you are more than welcomed to provide further feedback on mailing list later on 15:50:53 zhipeng, We need to have something like a midcycle to go over things that need work on 15:51:10 that is what I think 15:51:20 maybe a video conf meeting for like 2-3 hours ? 15:51:24 Ya 15:51:27 maybe the week of Sydney Summit ? 15:51:34 since all of us will not be there lol 15:51:38 lol 15:51:48 let's do something useful instead 15:51:48 Or we can do that monthly 15:51:58 of cource that'd be great 15:52:10 Justin and I can be in the same room 15:52:59 lol don't eat ice cream in front of me 15:53:41 We'll join from Coldstone 15:53:50 lol 15:54:00 okey moving onto next topic 15:54:14 #topic Generic Driver progress update 15:54:32 crushil any thoughts on what or how much work needed there ? 15:54:45 i think shaohe could also lend a hand here 15:55:29 So, I am working on creating interfaces between Cyborg APIs and generic driver. After that is done, I will work on creating a shim for the vendors 15:55:44 That can be helpful if shaohe can help 15:55:54 Does shaohe work with you? 15:56:42 no but we are in the same country, so it'll be easy for us to communicate 15:56:53 shaohe_feng_ interested ? 15:56:53 I'm planning to support inte 15:57:01 yes. 15:57:07 gr8t ! 15:58:02 we are planning to support intel's accelerator in cyborg 15:58:09 #action shaohe_feng_ and crushil collaborate on the driver development for backend support 15:58:23 such as intel's GPU and FPGA 15:58:32 shaohe_feng_, That's awesome 15:59:04 I have reached out to other vendors. Intel was one of the vendors that I hadn't talked ot 15:59:05 *to 15:59:15 nice :) 15:59:40 okey moving on to next topic 15:59:43 shaohe_feng_, Are you specifically supporting Intel products or Altera products? 16:00:43 sorry zhipeng We can talk about this offline. I don't mean to start another tangent 16:00:56 crushil: I can help on some general work 16:01:01 no worries 16:01:13 no important items left :P 16:02:45 we know better Intel's and Altera specifically products 16:02:58 ^ crushil 16:03:01 zhipeng, Please send a recurring calendar invite for every month where we analyze where we are 16:03:15 shaohe_feng_, Cool 16:03:39 #action zhipeng to send meeting invites for monthly deep dive 16:03:46 #topic patch discussion 16:04:05 #link https://review.openstack.org/#/q/project:openstack/cyborg+status:open 16:04:12 any outstanding issue there ? 16:04:48 We need more patches 16:04:51 :) 16:05:08 haha yep 16:05:34 okey then 16:05:38 #topic AoB 16:05:48 any other business 16:06:13 Nothing from me 16:07:04 anyone else ? 16:07:42 I got an email from NXCA technologies 16:07:50 They want to know more about Cyborg 16:07:53 me and justin as well 16:08:01 not sure if Amol is at the meeting now 16:08:06 I will schedule a call with them next week 16:08:10 I don't think so 16:08:10 probably not 16:08:15 okey great 16:09:33 if nothing else, then I will conclude the meeting 16:09:44 we have an awesome one today, thx everyone 16:09:54 #endmeeting