02:01:02 #startmeeting mogan 02:01:03 Meeting started Thu Jul 6 02:01:02 2017 UTC and is due to finish in 60 minutes. The chair is zhenguo. Information about MeetBot at http://wiki.debian.org/MeetBot. 02:01:04 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 02:01:06 The meeting name has been set to 'mogan' 02:02:01 liusheng, litao__, luyao, Xinran, zhangyang: o/ 02:02:07 o/ 02:02:29 o/ 02:02:33 o/ 02:03:13 o/ 02:03:43 as always, the agenda: 02:03:43 #link https://wiki.openstack.org/wiki/Meetings/Mogan#Agenda_for_next_meeting 02:03:48 let's jump in 02:03:55 #topic announcements and reminders 02:04:11 OpenStack Summit Sydney CFP, only 9 days left. 02:04:38 any ideas? 02:04:43 😬 02:05:32 just thinking 02:06:09 litao__: hah, ok, just need some abstract 02:06:31 zhangyang: would you like to prepare one about cloudboot or productions 02:06:51 zhenguo：yeah 02:07:26 zhenguo: i'm not sure... 02:07:28 litao__: is there a plan for Fiberhome to run mogan in production ? 02:07:42 zhangyang: hah, we can prepare a joint one 02:08:06 zhenguo: that will be very cool 02:08:36 zhangyang: need to show how our big Bank customer use it, lol 02:08:57 zhengguo：Currently not 02:09:12 litao__: need to wait it offical? 02:09:13 zhenguo: maybe need more time 02:09:37 zhenguo: not sure, maybe before 02:09:48 litao__: ok 02:10:14 The Pike release is coming, we don't have much time left. 02:10:34 zhenguo: we will relase our 1.0 ? 02:10:46 liusheng: yes 02:10:47 zhenguo: This is a big work for adding a new module, need to plan it in release version. 02:10:57 hi sorry I am late 02:10:58 liusheng: do we need a p-3 or just one for Pike 02:11:01 cool 02:11:03 Xinran: np 02:11:23 litao__: got it, same to us 02:11:25 zhenguo: maybe only one enough. hah 02:11:35 liusheng: hah, sure 02:12:12 liusheng: when we have a release, we should add API microversion and bump db schema migrations 02:12:46 zhenguo: ok, db migration is necessary 02:13:53 ok, and do you have something that must be done before release 02:14:04 liusheng: yes, I often encounter inconsistent issues for db 02:14:17 hah 02:15:06 zhangyang: do you still need server groups? 02:15:15 when do we stop merging codes for pike? 02:15:30 zhenguo: according to liudong, yes 02:15:43 zhangyang: at least one month 02:15:50 zhangyang: ok 02:15:51 zhenguo: i'll make it sure with him 02:16:07 zhangyang: ok, thanks 02:16:19 zhenguo: np:D 02:16:43 Contribution into mogan for the last 30 days 02:16:43 #link http://stackalytics.com/report/contribution/mogan/30 02:17:18 we got 5 person got more than 20 reviews last 30 days 02:17:25 thanks all for your efforts 02:18:00 hah 02:18:41 another thing I want to talk about is the 'review hour' 02:18:46 do you find it's useful? 02:20:00 seems not run very well and it's not effective 02:20:12 zhenguo: seems cannot ensure everyone have enough time 02:20:28 yes, that's one big reason 02:20:42 I think putting patches in etherpad for review is useful, hah 02:20:44 need to improve it or should we just cancel it 02:21:34 ok, let's talk about it later when shaohe_feng and wanghao back 02:21:54 not much of announcement, anyone else have a thing here? 02:22:15 ok, let's move on 02:22:21 #topic task tracking 02:22:31 #links https://etherpad.openstack.org/p/MoganWhiteBoard 02:22:39 looks like things are moving along nicely 02:22:46 * zhenguo gives folks a few minutes to review 02:25:14 do you think we should enrich quotas and notifications before Pike? 02:26:13 quota now just supports server number 02:26:39 litao__: yes, we should add more 02:27:04 I may prefer we'd better to enrich the stability of existing interfaces 02:27:33 liusheng: yes, before release we should make user the interfaces are stable 02:28:00 stable is important 02:28:23 sure 02:29:02 zhangyang: do you care about the notifications? 02:29:16 zhangyang: if you want to run mogan in production env 02:30:19 zhenguo: currently not using mogan notifications... 02:30:41 zhangyang: ok, got it 02:31:07 zhangyang: you can give us more feedback on your tests, so we can improve it 02:31:21 zhenguo: yes, sure! 02:31:33 zhangyang: thanks 02:31:47 zhenguo: :D 02:32:07 ok, let's go through the tasks list one by one 02:32:42 the new flavors one is done, maybe need a following up patch to deny deleting in use flavors 02:34:21 liusheng: next is placement, seems it's almost done, thanks for your hard work 02:34:35 zhenguo: np :D 02:34:50 zhenguo: after that, we may need to consider node aggregates 02:35:09 liusheng: sure, the next is node aggregates 02:36:19 liusheng: still not got how to use aggregates if it's only a notion of group without metadata/traits associated 02:37:11 zhenguo: not very sure, but maybe scheduler can use that to select grouped nodes ? 02:37:44 liusheng: you mean we can specify aggregates when query RP? 02:37:56 zhenguo: I think so 02:39:13 we now use node aggregates to group node manually, and use the group while creating servers. 02:39:26 liusheng: at least, for now admins can set traits to every rp they want to, and flavor can match that 02:40:14 zhangyang: for placement, we need to set traits to every node instead of setting metadata to a group 02:40:24 so we are thinking whether we can let user select between the placement api and the aggregate / scheduler of our own. 02:40:26 zhenguo: yes, but if we specified node aggregates, we can also filter the nodes in specific aggregates 02:40:50 liusheng: node aggregate is seen by normal users? 02:41:06 liusheng: or admin only 02:41:18 litao__: no plan for now 02:41:48 I mean we don't use the host aggreates way like nova, but we can achieve the same thing 02:41:52 liusheng: placement now only provide apis, I am not sure about the policy 02:42:15 if you want to group node1 node2 node3, you can set a trait to all of them 02:42:32 zhenguo: morning 02:42:47 zhenguo: ping 02:42:47 zhenguo: we don't use the similar interface provided by placement ? 02:42:50 the difference is that, nova create a host aggregates A, then you add node1, node2, node3 to A, and set metadata to A 02:43:07 shaohe_feng_: o/ 02:43:16 zhenguo: you can also add a node to a aggregate 02:43:23 ah, i see. use placement api to set metadata to nodes you want to manage. 02:43:33 zhangyang: yes 02:43:48 liusheng: if we can't set traits to aggreatge, we don't need aggregate 02:44:17 zhangyang: before you set metadata to aggregates, now you set metadata to every single node 02:44:22 zhangyang: it's the same 02:45:24 zhenguo: I see :D 02:45:46 zhangyang: and we use flavor resource_traits to match the specified resource provider(node) 02:46:21 zhangyang: you can set AZ, RACK, ROW or whatever info to placement resource provider 02:46:30 zhangyang: that will meet your requirements 02:47:18 * litao__ But it is a little complexity to add metadata for every node? 02:47:32 litao__: yes 02:47:54 litao__: that's for admins/deployers, 02:48:25 zhenguo: yeah, I think so 02:48:47 zhenguo: oh, i see. liudong are concerned on whether we should all turn to placement so quick, so we wanted to keep both methods... 02:49:24 zhangyang: hah, placement's achitecture is simple, so no need to concern 02:49:50 zhangyang: just some query SQL to DB, it will be more fast, and we only use the basic functionalities 02:51:00 maybe we can hold a voice meeting to discuss more about placement things, hah 02:51:17 zhenguo: so we can deploy placement api along with mogan on a single host? 02:51:34 zhenguo: yes, we should 02:51:37 zhangyang: sure 02:51:46 zhangyang: it's very small 02:52:38 zhenguo: ok, i see. 02:52:57 wechat voice meeting. 02:53:06 shaohe_feng: hah 02:53:11 haa, that is cool 02:54:13 shaohe_feng: intel's conference bridge seems work fine for us, hah 02:55:17 oh, it's almost time 02:55:22 zhenguo: then I can setup a conference bridge 02:55:35 shaohe_feng: ok, we can confirm with others later 02:55:58 another task is the 'adopt servers' 02:56:14 seems wanghao is absent 02:56:27 litao__: you will do together with wanghao, right? 02:56:33 zhenguo: Ok. a fit suitable time for all 02:56:40 zhenguo: yes, 02:56:41 shaohe_feng: yes 02:56:55 zhenguo: I will commit my patch later 02:57:03 litao__: ok, t hanks 02:57:17 zhenguo: n 02:57:22 zhenguo: np 02:57:23 hi all, please help to review the specs more 02:58:17 zhenguo: sure 02:58:28 and I would like to land the placement patches this week, then we will get more time to prepare other things before release. 02:58:41 also please help to review it 02:58:50 thanks 02:59:04 ok, let's move to open discussion 02:59:08 #topic open discussion 02:59:27 who's got a thing here 03:00:34 ok, it's almost time, thanks everyone for joining the meeting! 03:00:43 bye 03:00:47 bye 03:00:56 #endmeeting