02:09:05 #startmeeting openstack-cyborg 02:09:06 Meeting started Thu Mar 4 02:09:05 2021 UTC and is due to finish in 60 minutes. The chair is xinranwang. Information about MeetBot at http://wiki.debian.org/MeetBot. 02:09:07 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 02:09:09 The meeting name has been set to 'openstack_cyborg' 02:09:40 first of all, congrats inspur guys, we have inspur ssd driver merged lol 02:09:57 lol^ 02:10:00 congrats thanks 02:10:33 #topic vgpu support 02:10:51 next topic is vgpu implemt 02:11:25 xinranwang: when do we FF in this cychle? 02:11:34 cycle 02:12:20 https://wiki.openstack.org/w/index.php?title=Cyborg/Wallaby_Release_Schedule&oldid=176980 02:12:48 xinranwang: i have checked your comment, the patch need to solve conflict 02:13:07 brinzhang_: it will be next week 02:13:12 songwenping__: which patch? 02:13:26 https://review.opendev.org/c/openstack/cyborg/+/769482 02:14:21 about vgpu management 02:14:34 yeah, I've seen your updates. 02:15:02 I think throwing out an exception when gpu is not configured in file is ok for me 02:15:24 what do you mean by "conflict" 02:15:51 conflict with master branch 02:16:16 yes, may you need to rebase on master 02:16:34 sure 02:16:45 oh, so you may need more time to rebase it, right ? 02:17:11 yeah 02:17:17 songwenping__: sure 02:17:22 thanks wenping 02:17:40 n[ 02:17:42 np 02:18:13 I think vgpu cyborg side is almost good, is it possible about we can merge this in this release. 02:18:27 agree 02:18:40 Is there any dependency with nova side which not yet be nailed down 02:19:59 seems no dependency 02:20:37 ok, great. 02:20:49 I have some concern, do we need to test nova-cyborg interaction about the vGPU feature? 02:21:05 but nova side cannot merge in this release. 02:21:18 brinzhang_: I think we should do the test 02:21:34 at least in local env 02:21:44 or, we merged cborg support, then later to test, if there is something need to improve, that we can fix bug 02:22:01 and in next release we can use launchpad to trace bugs 02:22:07 give out the test report? 02:22:37 I think we need make sure the current vgpu support in cyborg works with nova side patch. 02:23:29 songwenping__: yes, a test report is needed 02:23:47 hi all. sorry for later. 02:23:56 chenke: hi, it's ok 02:24:05 ok. thanks. 02:24:06 we are talking about vgpu patches 02:24:07 hi chenke 02:24:18 hi wenping. ok. please all continue. 02:24:56 chneke^ morning 02:25:31 morning brin. 02:25:41 i havenot test nova-cyborg interaction yet because there is not enough time. 02:26:36 can we merge the bp in next release with nove side? 02:27:19 (TO AGREE) we need to test the vgpu cyborg patches with nova patch, and generate a test report to show how it works. If cyborg patch has no dependency with nova patch, we can let cyborg patch in. the test report is also useful for nova guys to understand how it works. what do you think? 02:28:54 not sure whether there is dependency with nova or not now. 02:29:33 because we donnot know nova cores' ideas 02:30:30 songwenping__: ok, if it's too urgent to merge. we can do it in next release 02:30:46 i mean we donnot need hurry to let cyborg patch in 02:31:05 ok 02:31:28 songwenping__: sure, better to have an agreement with nova and test it before merge 02:31:39 agree. 02:31:55 btw, do you need to cross-project with nova to discuss the vgpu details? 02:33:28 if there are questions while testing, we need to discuss with nova. 02:34:31 you means on the PTG? xinranwang? 02:34:51 from now, we are not have the implementation, there is not questions, right? songwenping__ 02:34:51 yeah 02:35:44 and there is another feature, cyborg resume/paused support patch, also need to continue work in next release 02:36:21 donot need to talk on PTG now. 02:36:42 songwenping__: ok, please let me know if you need any help 02:36:57 thanks 02:37:20 brinzhang_: got it, do you need discuss it during ptg with nova? 02:37:54 not necessary 02:38:35 lol, ok. 02:38:42 and how to implement there is clearier, the same as shelve/unshelve 02:38:59 nice 02:39:20 just need to update continuely ^, so we dont joined the cross project talking 02:39:50 xinranwang: how about the edge with cyborg interaction? do you have some plan? 02:39:50 btw, do we have any vgpu nova-cyborg spec to show the workflow? 02:41:34 brinzhang_: I do connect with edge team but for now there is no clear plan for this part 02:41:45 ack 02:41:59 xinranwang: seems no 02:43:17 songwenping__: do you think it's better to have a spec to show all the workflow so that we can have an agreement with nova on the spec first, like smartnic support? 02:43:51 sure, agree. 02:44:35 great, it will be more easier to discuss 02:45:40 #topic ptg 02:45:51 next topic is ptg 02:46:01 will you all attend the ptg? 02:46:01 I remeber xinran's workflow graph is very nice. 02:46:47 lol, there is a tool to draw that. 02:47:29 will you all attend the ptg? chenke brinzhang_ songwenping__ 02:47:31 oh, i though you type many spaces for the workflow graph 02:48:01 where the link of the ptg scheduler? 02:48:07 I want to . but need to negotiate with the team 02:48:08 xinranwang:sure^ 02:48:09 songwenping__: no way, I would got out of battery if do this/ 02:48:24 chenke: it is virtual! free! 02:49:14 lol. ok.hope to listen and talk our cyborg on ptg. 02:49:16 there is no specific schedule for each team. but the date is Apirl 19 -23 02:49:39 i will be ok. 02:50:02 which time is ok for you? 02:50:42 all times, lol. 02:51:10 lol 02:51:31 I will align with you guys again when booking the meeting room 02:52:00 songwenping__: you can use asciiflow or textik to draw the workflow 02:52:24 ok i'll try, thanks. 02:52:54 nice tool 02:53:11 ok, that's all for today, the end of next week is FF, please help review the pending patches :) 02:53:37 and fill the etherpad with topic you wanna discuss on ptg 02:53:44 https://etherpad.opendev.org/p/cyborg-xena-ptg 02:54:15 xinranwang: https://review.opendev.org/c/openstack/cyborg/+/778183/2/cyborg/accelerator/drivers/fpga/intel/sysinfo.py#333 i think we cannot start index from 0 at L333. 02:55:33 as it may be changed in the for loop above. 02:55:38 bye 02:56:42 songwenping__: good catch, I will improve it. thanks 02:57:05 np 02:57:12 nothing from myside 02:57:31 ok 02:57:39 bye 02:57:49 if nothing else, let's wrap up the meeting 02:57:55 #endmeeting