03:17:20 <Yumeng> #startmeeting openstack-cyborg
03:17:21 <openstack> Meeting started Thu Dec 17 03:17:20 2020 UTC and is due to finish in 60 minutes.  The chair is Yumeng. Information about MeetBot at http://wiki.debian.org/MeetBot.
03:17:22 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
03:17:24 <openstack> The meeting name has been set to 'openstack_cyborg'
03:17:26 <Yumeng> #topic Roll call
03:17:43 <Yumeng> #info Yumeng
03:17:45 <xinranwang> #info xinranwang
03:18:03 <songwenping_> #info songwenping_
03:19:10 <Yumeng> # topic milestone announcement
03:20:05 <Yumeng> time flies we've passed wallaby-milestone1 on Dec 04
03:20:22 <brinzhang> #info brinzhang
03:20:46 <Yumeng> our next milestone will be Jan 18 - Jan 22: Wallaby-2 milestone, cyborg spec freeze
03:21:34 <Yumeng> It's good we've merged vGPU spec. other driver specs are also under review
03:23:00 <Yumeng> vGPU spec in nova side is still pending. :(
03:23:10 <brinzhang> Yumeng: vGPU specs are still pending
03:23:23 <brinzhang> I think you want to say smartNIC spec,right?
03:23:40 <Yumeng> brinzhang: I mean vGPU in cyborg side
03:23:59 <brinzhang> ack
03:24:15 <xinranwang> ah, why it is pending?
03:24:28 <Yumeng> need push nova cores to review..
03:25:09 <Yumeng> I was a bit busy with some urgent internal work in the last two weeeks.
03:25:17 <brinzhang> and there is the shelve/unshelve support patch, nova cores doesnot have time to review
03:25:24 <xinranwang> maybe you can push them in this week because after this week, it will be the holiday season..
03:25:31 <Yumeng> aha
03:25:40 <brinzhang> and there seems just only gibi to review it
03:25:48 <Yumeng> alright. I will ping them again this afternoon.
03:27:14 <Yumeng> maybe we should ask for their review one by one？rather than ask review for all？
03:27:43 <Yumeng> spec first?code later?
03:27:51 <xinranwang> spec is more urgent I think.  Jan will have spec freeze
03:28:00 <Yumeng> agree
03:28:49 <songwenping_> +1
03:29:47 <brinzhang> vGPU spec that have many people to review, but the shelve/unshelve patch just has gibi to review
03:30:08 <brinzhang> dansmith will leave in next month..
03:30:36 <Yumeng> oh. leave nova team?
03:30:51 <brinzhang> he will start his new job (congratulations from me ^^)
03:31:13 <songwenping_> good job
03:31:17 <xinranwang> brinzhang:  lol, he will not working on openstack?
03:33:21 <brinzhang> xinranwnag: maybe, but Iam not sure, I have got news from nova channel. In fact, I shouldn't have spread it.
03:33:53 <xinranwang> brinzhang:  ok, got it
03:34:17 <Yumeng> it might be ok. nova channel is open
03:35:40 <Yumeng> but agree we'd better not spread it .  :)
03:35:50 <brinzhang> yeah
03:36:31 <Yumeng> ok. ping nova to review vGPU spec and smartnic spec this afternoon.
03:37:03 <xinranwang> Yumeng:  smartnic spec is already merged  :)
03:37:08 <Yumeng> aha
03:37:27 <xinranwang> it is a good news for us
03:37:29 <Yumeng> that's great
03:37:48 <Yumeng> news latency in my side.
03:38:27 <xinranwang> Yumeng:  haha, I will continue working on cyborg side implementation
03:38:49 <Yumeng> cool, great! congrats!
03:39:02 <xinranwang> Yumeng:  lol, thanks
03:39:08 <Yumeng> ok. only needs to ping vGPU spec now.
03:39:36 <Yumeng> # topic review status
03:41:38 <Yumeng> I noticed we got new patches from other Inspur contributors
03:41:52 <songwenping_> @all, why we didnot consider the accel_info when resume guest state if a host is booted. pls see the file https://review.opendev.org/c/openstack/nova/+/767273/1/nova/virt/libvirt/driver.py#3490
03:42:13 <Yumeng> like openstack/cyborg-tempest-plugin
03:44:05 <Yumeng> but another question gibi asks is what if cyborg-agent service starts after nova-compute?
03:44:06 <brinzhang> Yumeng: yes, we have a team, and I asked them to contribute to the tempest plugin
03:44:53 <Yumeng> brinzhang: that's cool. Thanks!
03:45:02 <songwenping_> yes it indeed has the question.
03:45:56 <Yumeng> My opinion is we just let it go and follow what nova does today, because that's not a cyborg-vGPU-specific issue.
03:46:49 <songwenping_> is it still the question about vGPU, so how we solove it?
03:49:08 <Yumeng> In the production environment, operator can have their own way(watchdog/guard process) to do this kind of instance check/service status check.
03:50:13 <Yumeng> according to their own needs
03:51:41 <Yumeng> songwenping_: question just mentioned is still the gibi's one "but another question gibi asks is what if cyborg-agent service starts after nova-compute?"
03:52:22 <Yumeng> if cyborg-agent service starts after nova-compute, nova can not get acc_info as expected https://review.opendev.org/c/openstack/nova/+/767273/1/nova/virt/libvirt/driver.py#3490
03:55:09 <Yumeng> what do you guys think how can we solve this issue?
03:59:17 <Yumeng> brinzhang xinranwang songwenping_ are you still here?
03:59:41 <Yumeng> lunch time ha?
04:00:18 <xinranwang> ah, I am here.
04:00:26 <Yumeng> haha
04:00:37 <xinranwang> I am not very familiar with the spec
04:00:44 <Yumeng> no pb.
04:01:01 <Yumeng> let me ping wenping in wechat later
04:01:14 <xinranwang> can we use try..except to verify
04:01:29 <xinranwang> just a idea, not sure it works :)
04:01:46 <Yumeng> yes. that's a possible solution!
04:02:05 <xinranwang> seems they are not here. Maybe we can continue discuss offline
04:02:15 <Yumeng> we can add cyborg-agent service check in nova's init_host method
04:02:27 <Yumeng> sure
04:03:20 <Yumeng> xinranwang do you have anything else to bring up?
04:03:34 <xinranwang> nothing from my side
04:03:44 <Yumeng> ol. if we don't have other topics, let's wrap up today's meeting.
04:03:54 <Yumeng> #endmeeting