03:02:53 #startmeeting openstack-cyborg 03:02:54 Meeting started Wed Apr 3 03:02:53 2019 UTC and is due to finish in 60 minutes. The chair is Sundar_. Information about MeetBot at http://wiki.debian.org/MeetBot. 03:02:55 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 03:02:57 The meeting name has been set to 'openstack_cyborg' 03:03:03 #topic Roll Call 03:03:08 hi Li_Liu Sundar_ 03:03:13 #info xinranwang 03:03:17 #info Sundar 03:03:21 Hi xinranwang 03:03:24 #info Li_Liu 03:03:37 Let's wait a moment for others to join 03:03:58 #info Yumeng 03:04:02 hi all 03:04:23 Hi all. 03:04:32 #info yikun 03:04:41 #info wangzhh 03:04:41 hello 03:04:51 Hi everybody, let's get started 03:04:54 #topic Status 03:04:56 https://review.openstack.org/#/q/status:open+project:openstack/cyborg+branch:master 03:05:11 It looks clean to me. Great job, everybody! 03:05:24 We should be cutting the RC2 within a couple of days 03:05:52 One question I have: What do we do with the patches with merge conflict, which will never really be merged? 03:05:52 any hope on https://review.openstack.org/#/c/645499/ ? 03:06:51 Li_Liu: We should do a bit more testing IMHO. We are working on a CI setup 03:07:07 Sundar_: my patches related to PoC will never get merged. 03:07:22 xinranwang: Do you agree on OPAE driver patch? 03:07:53 Li_Liu: the current opae still has some bugs, we'd better wait for next release of OPAE 03:08:45 xinranwang: I think the origial idea for the non-mergeable patches was to share with the team. They can download it and refer. 03:08:52 xinranwang, sure 03:09:05 Sundar_: yes 03:09:20 What do others think about these non-mergeable patches? 03:11:05 Li_Liu, wangzhh, Yumeng, yikun, anybody? 03:11:19 Yes. 03:11:58 We can keep them there for reference for now 03:12:11 +1 03:12:21 Might need to find a cleaner way to back log them 03:12:30 +1, It's better merge thoese in next release for me 03:13:28 #agreed Leave the non-mergeable patches as they are, for now 03:13:43 +1 03:13:45 Any other patch to discuss? 03:14:11 I think we are good 03:14:19 I will cut RC2 tomorrow 03:14:28 Yes, thanks Li_Liu. 03:14:33 np 03:14:35 #topic PTG preparation 03:14:51 #link https://etherpad.openstack.org/p/cyborg-ptg-train 03:14:58 We need to fill the schedule 03:15:33 Some folks have asked for a demo 03:15:57 Can Intel provide some infrastructure for the demo? 03:16:04 https://etherpad.openstack.org/p/ptg-train-xproj-nova-cyborg 03:16:23 I am preparing 2 demos -- was meant to be one, but will probably be 2 :) 03:16:38 Hi all 03:16:53 #info Coco_Gao 03:17:11 Hi Coco_gao 03:17:13 One is at the Intel booth, with a real-world app (genomics) + CYborg + Nova in upstreamable form. Probably no programming here. 03:17:44 is it on-site server or off-site server? 03:17:49 Other is for the Nova folks. There have been several asks, and I am trying to show programming. But we don;t have full discovery 03:17:56 so this one including the e2e device attach demo? :) 03:18:10 I mean the 1st one 03:18:34 Nova folks want to see discovery but we have to say the Intel OPAE driver is WIP. I have some setup scripts to populate the Cyborg db and Placemen directly 03:19:08 yikun: First one also does not have discovery 03:19:42 demo is always hacky :P 03:19:46 Li_Liu: on-site server : I'll either do a live demo or use a some video for backup 03:20:18 Yes, a few hacks but, IMHO, Nova's real interest should be in Cyborg APIs and how they get called 03:20:39 I see 03:20:39 Internals of device discovery shouldn;t matter -- except that Cyborg should not conflict with PCI whitelist 03:21:15 Li_Liu: my question is, is it too late to get a room for the demo? 03:21:23 We can just pre-populate the DB as if the discovery is happening 03:21:39 you mean during the PTG or summit? 03:21:54 PTG 03:21:58 Sundar_: agree, so the key is show cyborg api/client and the key interface of attachment. 03:22:41 Do we need an extra room for thedemo? 03:22:58 I thought each team should be getting a room and we can just use that 03:23:36 A room with a projector, which can potentially accommodate a large #people. ( I am thinking of the bw provider demo for Neutron at last PTG) 03:24:02 Li_Liu: Ok, does that room have a projector? 03:24:05 Another option is to show it during our project update session during summit 03:24:18 I will ask them 03:24:26 I think Nova folks will have tons of questions 03:24:44 Would be good to do this at the PTG, drive discussion and closure 03:25:04 Maybe you could send a mail to ML, and arrange it in the beginning cyborg meeting/(or nova-cyborg meeting) 03:25:11 Li_Liu: Thanks 03:25:23 Please please use your phone to record it 03:26:10 if the demo video is available, and we can also send it to ML, and let the tons of question comes before PTG. :) 03:26:15 yikun, all: Yes, we need to set a time in consultation with Nova. May be Thursday or Friday, since many folks will not be there on Sat -- including yours truly. 03:27:27 On a related note, I am thinking of setting a cross-project with Ironic, because (a) Ironic deployments are going up, (b) people want accelerators in bare metal, and (b) this may a path to host K8s + accelerators on Cyborg 03:27:39 That would also require setting a specific time 03:28:22 Request to all of you: please fill out https://etherpad.openstack.org/p/cyborg-ptg-train , so that we can estimate how much time each will take. 03:29:05 I'll stop here, and let you folks speak -- for a change :) 03:29:10 sure will do 03:29:15 ok 03:29:53 Sundar_: Note that 4.5 ~ 4.7 is china Chingming festival (So, Friday Sat Sun is vacation in China) 03:30:08 I see 03:30:55 Sundar_ big +1 on Ironic thought 03:31:14 zhipeng: :) 03:31:24 I've tried to get this started a while ago (Julia was not convinced back then ) 03:32:11 I spoke to Julia too a couple of times. I agree, it was a bit premature without getting Nova fully lined up 03:33:25 Keep up the great work ! 03:33:39 Another big thing coming up -- Intel PAC N3000 https://www.anandtech.com/show/13958/intel-fpga-pac-n3000 03:34:24 I'd like to present it in tomorrow's Zoom call. This has *huge* interest from telcos, and also represents quite some changes for CYborg. 03:34:48 great, looking forward to it 03:34:56 I tried to fold this into our device model, but the details kept changing 03:35:51 Great! 03:36:06 Sounds good. 03:36:16 What about the demo, do you need any help? 03:37:00 I think the discovery part may be functional, but not the nova interaction part. 03:37:41 Could you pls introduce you demo a little bit in tommorows meeting? 03:37:59 Coco_gao: Yes. I am about to make one more set of changes, including support for programming with v2 API. Please review the pilot branch patches after that! 03:38:56 Coco_gao: I need to merge the master with the feature branch, and then we can see what parts ork 03:39:44 Re. Intro the demo tomorrow, please give me some more time. Because only today I got permission to merge master onto the pilot. 03:40:17 Thank you, take your time. 03:40:57 The glorious git rebase 03:42:16 I tried a rebase, and #infra folsk said no, I need to do a merge. I tried the merge and that failed, because I need permissions. Then I have to figure out how to get those permissions. Now that is done -- so onto the glorious merge :) 03:42:20 Merged openstack/python-cyborgclient master: fix service name in client https://review.openstack.org/649493 03:43:00 aha, I see the gerrit information. 03:43:41 Any other topic on PTG preparation? 03:44:40 #topic AoB 03:45:59 Li_Liu, loking forward to the room info. All the best for the RC2 cut. 03:46:34 I have sent the email to confirm 03:46:45 Will let you know once they get back to me 03:46:51 Thanks, Li_Liu 03:47:24 np 03:47:54 Great job on Stein, and thanks everybody! Good day (or good night). Enjoy the Chingming festival! 03:48:04 #endmeeting