02:00:25 <zhenguo> #startmeeting mogan 02:00:26 <openstack> Meeting started Thu Mar 16 02:00:25 2017 UTC and is due to finish in 60 minutes. The chair is zhenguo. Information about MeetBot at http://wiki.debian.org/MeetBot. 02:00:27 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 02:00:30 <openstack> The meeting name has been set to 'mogan' 02:00:38 <zhenguo> o/ 02:02:45 <zhenguo> zhangjl, litao, luyao, wanghao_, liusheng, zhangyang: hi 02:02:59 <zhangjl> zhenguo:hi, i`am here 02:03:01 <wanghao_> okay let's roll 02:03:05 <zhangyang> hi 02:03:13 <litao> here 02:03:27 <zhenguo> as always, the agenda: 02:03:28 <zhenguo> #link https://wiki.openstack.org/wiki/Meetings/Mogan#Agenda_for_next_meeting 02:03:31 <liusheng> o/ 02:03:41 <zhenguo> let's jump in 02:03:46 <zhenguo> #topic announcements and reminders 02:04:01 <zhenguo> Our topic is not selected for Boston summit:( 02:04:25 <liusheng> :( 02:04:33 <zhenguo> seems some people don't like our way to mange baremetals, we should prove the benefits. 02:04:36 <zhangjl> :-( 02:04:39 <litao> what's a pity 02:04:42 <yuntongjin> hello mogan 02:05:04 <zhenguo> yuntongjin: o/ 02:05:30 <litao> and some people agree 02:05:49 <zhenguo> yes 02:06:08 <zhenguo> there's a discussion about adding support for RAID configuration for Ironic instance in Nova 02:06:18 <zhenguo> #link https://review.openstack.org/#/c/408151/ 02:07:01 <zhenguo> Please keep an eye on that, and we may also need to prepare a spec after the refacoring of flavor/scheduler 02:07:13 * zhenguo gives folks a few minutes to review 02:07:35 <yuntongjin> China Mobile is looking for host AGG in baremetel cloud 02:08:13 <liusheng> zhenguo: is Roman a ironic guy ? 02:08:15 <zhenguo> yuntongjin: seems they have many downstream hacks now 02:08:20 <yuntongjin> https://review.openstack.org/#/c/441924/ 02:08:38 <zhenguo> liusheng: no, a mirantis guy working on nova 02:08:39 <yuntongjin> zhenguo: they are trying to upstream this , 02:09:30 <zhenguo> yuntongjin: that's not easy possible 02:10:43 <yuntongjin> zhengguo: that's true, not easy job. 02:11:03 <zhenguo> ok, let's move on 02:11:11 <yuntongjin> Mogan can really be helpful here to address this requirement 02:11:31 <zhenguo> yuntongjin: sure, but seems they can't wait us 02:11:46 <zhenguo> yuntongjin: we can discuss this in the open discussion topic 02:11:49 <litao> they just want to use nova 02:11:56 <liusheng> may because people just know little about Mogan 02:12:13 <zhenguo> yes, and we are not that mature now 02:13:41 <zhenguo> Ok, really many refactors during last week ans so many conflicts on the in progress patches, sorry for the inconvenience. 02:14:53 <zhenguo> As the Placement service is not ready now, we will save node resources to db instead of cache, and split out scheduler later 02:15:06 <zhenguo> wdyt? 02:15:44 <zhenguo> in fact, we have already save them to db now :D 02:16:26 <litao> I found the placement service in the N verion 02:16:51 <zhenguo> yes, but there's still a long way to be mature 02:17:52 <litao> so maybe we use it in next version 02:18:18 <zhenguo> hope so, hah 02:18:34 <liusheng> it is better to make our scheduler easy to be migrated to the future placement 02:19:17 <zhenguo> yes, with placement we still have a scheduler, just need to save resources there and use it for scheduling 02:19:20 <liusheng> more closer to the placement service implementation 02:19:30 <zhenguo> sure 02:20:15 <zhenguo> okay, the last thing is about Valence 02:20:28 <zhenguo> Valence integration spec maybe need to update after the new flavor landed. 02:20:28 <zhenguo> #link https://review.openstack.org/#/c/441790/ 02:21:19 <zhenguo> aha, I have changed to use node_type instead of instance_type in ironic node's properties 02:21:56 <liusheng> is it still the instance type name ? 02:22:03 <zhenguo> yes, 02:22:07 <liusheng> ok 02:22:43 <zhenguo> but I still think of the flavor pattern, will draft a spec up later 02:23:19 <liusheng> you mean refactor the falvor model ? 02:23:25 <zhenguo> yes 02:23:41 <zhenguo> we need a new flavor 02:23:46 <liusheng> hah 02:24:39 <zhenguo> the flavor should not only give the information about size, also type 02:24:58 <liusheng> that is necessary for variety drivers support 02:25:24 <litao> will replace the current flavor? 02:25:46 <zhenguo> yes, do you like the current pattern? 02:26:22 <litao> I think a little simple 02:26:30 <zhenguo> hah 02:26:34 <zhangjl> agree 02:27:16 <zhenguo> we may need to prove cpu, ram ..., as we can also compose node from Valence 02:27:26 <zhenguo> s/prove/provide 02:27:49 <zhangjl> i see... 02:27:52 <wanghao_> did we have a spec about this refacoring? 02:28:01 <zhenguo> I will do that 02:28:06 <litao> great 02:28:21 <zhenguo> but I will split scheduler out first 02:28:41 <liusheng> sorry, a fool question, for Valence, we can provision instance by accurate cpu, ram .. right? 02:29:02 <zhenguo> seems yes 02:29:08 <liusheng> cool 02:29:13 <zhenguo> Valence is not a driver for us 02:29:16 <litao> maybe like nova? 02:29:30 <zhenguo> yes 02:29:32 <zhenguo> like vm 02:29:59 <zhenguo> we compose hardware first, then enroll it to our driver(ironic) then provision it 02:30:19 <zhenguo> so Valence is in the up layer of driver 02:30:54 <zhenguo> and we don't need to scheduling node to use if go to Valence to compose nodes 02:31:43 <liusheng> sounds like a magical Valence, lol 02:31:51 <zhenguo> hah, yes 02:31:59 <litao> Is Valence a pool contains storage, compute ? 02:32:13 <zhenguo> litao: yes, a specific hardware 02:32:37 <litao> sounds intresting 02:32:47 <zhenguo> hah, 02:33:02 <zhenguo> anything else here? 02:33:24 <zhenguo> ok, let's move on 02:33:31 <zhenguo> #topic task tracking 02:34:00 <zhenguo> hi guys, how's everything going on with your task 02:35:10 <wanghao_> Quota patch has some issues, I'm figuring them out. 02:35:39 <zhenguo> wanghao_: got it, thanks 02:35:43 <wanghao_> Not a big deal. 02:35:52 <litao> My multi-instance support is over 02:36:03 <liusheng> console patches are ready to review 02:36:09 <zhenguo> litao: hah, do you want to do more works :P 02:36:19 <zhenguo> liusheng: ok, thanks 02:36:20 <litao> zhenguo: 02:36:30 <litao> zhenguo: yes 02:36:48 <zhenguo> hah 02:37:19 <litao> I need to take on some other work 02:37:33 <zhenguo> litao: thanks very much, 02:38:34 <zhenguo> as there are many refactors comming, all tasks seems blocked, sorry 02:38:45 <zhenguo> I will try to finish taht as soon as possible 02:39:24 <zhenguo> but it's almost there, next week we can focus on flavor design 02:39:42 <liusheng> awesome works 02:39:59 <luyao> zhenguo, I do research on my thesis in past several days so I don't carry my work forward. I'll finish the patch about instance_fault next 02:40:02 <zhenguo> but I'm pretty sure, there will be many small issues 02:40:16 <zhenguo> luyao: oh, thanks 02:41:05 <liusheng> zhenguo: yes I have found several ones. lol 02:41:24 <zhenguo> liusheng: please feel free to fix them, thanks 02:41:30 <liusheng> zhenguo: sure 02:42:15 <zhenguo> zhangyang: how's everything going on with the cloudboot driver? 02:42:57 <zhangyang> zhenguo: need to push change to cloudboot to fit glance, still working :D 02:42:59 <zhenguo> zhangyang: I have made the driver interfaces more general, hope it will help you. 02:43:42 <zhangyang> zhenguo: that's great 02:44:31 <zhenguo> zhangyang: ok, please ping me if something still need to be changed 02:44:58 <zhangyang> zhenguo: ok, thanks 02:45:04 <zhenguo> np 02:45:17 <zhenguo> ok, thanks everyone for the hard work with your task this week 02:45:22 <zhenguo> let's move on 02:45:32 <zhenguo> #topic open discussion 02:45:44 <zhenguo> who's got a thing here :) 02:46:24 <liusheng> a little idea to listen your opinions 02:46:49 <liusheng> how about more our meeting to specific openstack meeting channel ? 02:46:56 <liusheng> s/more/move 02:47:24 <zhenguo> is official project a requirement? 02:47:37 <liusheng> not sure 02:48:15 <liusheng> maybe we can ask someone of non-official project previously 02:48:23 <zhenguo> I would like to, but not sure there's still suitble time slot for us. 02:48:56 <zhenguo> you mean to invite more people to attend the meeting, right? 02:49:34 <liusheng> at least to make our voice can be listen to public 02:49:54 * litao maybe foreigners 02:50:10 <liusheng> I think our project is too close currently, that many people don't know about us 02:50:29 <zhenguo> liusheng: yes, that's the problem 02:50:49 <zhenguo> liusheng: we don't even send out emails to public 02:50:58 <zhenguo> that's maybe a more effective way 02:51:04 <liusheng> zhenguo: yes :-( 02:51:27 <zhenguo> lt's try to change 02:51:40 <liusheng> zhenguo: thanks 02:51:54 <zhenguo> liusheng: thanks for point out that 02:52:33 <zhenguo> liusheng: and we may need some blogs to introduce Mogan 02:53:00 <litao> I agree 02:53:17 <zhenguo> but of course not only chinese internal blogs 02:53:56 <liusheng> zhenguo: yes, I think that is even more important than coding works for current period. 02:54:21 <litao> It]s best to introduce mogan to more foreigners 02:54:24 <zhenguo> liusheng: hah 02:54:50 <zhenguo> yes, but now we still don't have enough advanced thing to say :( 02:55:47 <liusheng> zhenguo: hah, yes, we can just draw a blueprint first, but just personal thougts 02:56:15 <zhenguo> liusheng: yes, 02:56:55 <zhenguo> maybe RAID is a good point for us, 02:57:12 <liusheng> zhenguo: good idea 02:58:25 <zhenguo> and another thing is docs 02:58:54 <zhenguo> I think it's hard for users to deploy Mogan, or use Mogan with the current little docs 02:59:08 <liusheng> hah, yes 02:59:10 <wanghao_> agree 02:59:29 <zhenguo> there's even anything about the CLI :( 02:59:43 <zhenguo> not anything 03:01:02 <zhenguo> now wiki is the only place we can find what's Mogan #link https://wiki.openstack.org/wiki/Mogan 03:01:12 <liusheng> hah 03:01:15 <zhenguo> oh, sorry, it's over time 03:01:33 <zhenguo> thanks everyone, we can continue the discussion here later 03:01:41 <liusheng> thanks, bye 03:01:45 <zhenguo> bye 03:01:48 <zhenguo> #endmeeting