13:01:51 #startmeeting senlin 13:01:51 Meeting started Tue Nov 10 13:01:51 2015 UTC and is due to finish in 60 minutes. The chair is haiwei. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:01:52 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 13:01:54 The meeting name has been set to 'senlin' 13:02:22 please check the agenda and feel free to update it https://wiki.openstack.org/wiki/Meetings/SenlinAgenda 13:03:05 sorry I'm late 13:03:22 ok yanyan 13:03:26 hi 13:03:31 let's move on to the first topic 13:03:34 ok 13:03:38 #topic Go through TODO items to ensure everyone get them 13:04:12 here is the TODO list updated recently 13:04:14 https://github.com/openstack/senlin/blob/master/TODO.rst 13:04:25 yes, we just updated the TODO item list based on the discussion on Tokyo meetup 13:05:02 some backlog of liberty cycle were also added 13:05:25 In the first paragraph there are some notices we should pay attention to 13:05:38 yep 13:05:44 we need a BP to start these jobs 13:06:27 and once a workitem is picked up, please also propose a patch to remove it from the todo list and add it to etherpad 13:06:29 All of these jobs need a related BP? 13:06:40 here: https://etherpad.openstack.org/p/senlin-mitaka-workitems 13:06:52 elynn, most of them I think 13:07:11 it seems API response modification is not assigned to anyone, I will take this job 13:07:19 or if it is related to bug or it's really a trivial item, we can use a bug report to replace bp 13:07:45 elynn, you need to add a BP for engine lock broker 13:07:49 right, if it's something small a bug is sufficient 13:07:59 jruano, yes 13:08:16 test cases for example 13:08:23 yes 13:08:23 haiwei: I will add a BP if needed. 13:08:38 cool 13:08:59 ok, I think all of you have checked the tasks when this patch is under review 13:09:16 anything need to discuss more? 13:09:43 or if you have any confusion about some workitems, we can make a further discussion here 13:09:44 ok, lets move on 13:10:17 hi, lkarm 13:10:48 ok, next topic 13:10:51 hi 13:10:54 #topic Mitaka work items 13:11:22 yanyan, is there any document about this topic? 13:11:29 https://etherpad.openstack.org/p/senlin-mitaka-workitems 13:11:35 just the etherpad link 13:11:40 thanks 13:12:18 different from TODO.rst, this place is used to maintains all workitems in working progress 13:12:18 It's too slow to open this url for me. 13:12:46 got it, yanyan 13:12:54 I didn't update it 13:13:01 will update it later 13:13:04 so each time you guys pickup a workitem from todo list, plz also move it to etherpad 13:13:07 haiwei, no problem 13:13:23 this can help the team to track all in progress work 13:14:16 elynn, I just helped add the workitem about lock breaker which you have started work on :) 13:14:33 Thanks, still not open that link... 13:14:43 sigh... poor network 13:14:50 take it easy 13:14:53 I'm using proxy now 13:15:05 network is a little nervous 13:15:34 Poor connection tonight. 13:15:48 seems so 13:15:48 can we move on? 13:15:55 yes 13:16:08 #topic Quick update about Senlin resource support in Heat 13:16:10 I will check later 13:16:17 this is your job elynn 13:16:30 can you share your progress to us? 13:16:35 Yes, for now, only one patch submit to review. 13:16:52 https://review.openstack.org/#/c/240776/ Adding senlinclient in heat. 13:17:09 nice 13:17:21 this is the first step to support senlin resource type in heat 13:17:28 Still waiting for a new version of selinclient comes to global-requirement. 13:17:33 yep 13:17:53 I think senlinclient will have a new version soon 13:17:53 ethan, do you need some help with this? 13:17:54 So that I can shrink the codes for selin plugin. 13:18:06 since we rebuilt the senlinclient interface to make it consistent with other services client 13:18:52 jruano: thanks, I'm not sure if selinclient bump version or not. 13:19:15 elynn, if you need any help, no matter in heat or senlin side, you can just get it from us :) 13:19:24 is there any work left in senlinclient? 13:19:30 I think qiming will handle the client version bumping soon 13:19:31 After this patch, I will add 2 resources, OS::Senlin::Cluster and OS::Senlin::Profile 13:19:37 I just started the unit test for senlinclient 13:20:02 haiwei, nice, saw your patch this afternoon 13:20:04 yanyan: yes, count on him. 13:20:32 yanyan, please help review it, I am not so confident of the patch 13:20:43 haiwei, no problem 13:21:27 elynn, I recalled we will only propose profile and cluster resource types to heat at the first step? 13:21:41 yes, for now 13:22:14 ok, anything else for this topic? 13:22:24 I think it's enough to show the capability of senlin in heat for now 13:22:36 After all the codes merged 13:23:01 cool, and I think we should refer to resource group when consider what properties and attrs we should add to cluster resource 13:23:05 yes i think that is a great first step 13:23:08 I will upload some templates to show how to use these resource. 13:23:35 yanyan: Yes, property is define in spec. 13:23:58 we still need to decide what kind of attrs to add. 13:24:25 yea, we need make further discussion about this 13:24:35 For cluster, as we discuss in irc, we have 2 attrs: 'current_size' and 'node_details' for now 13:25:04 elynn, just a question , why node resource is not to be supported in the first step? 13:25:09 and also cluster id and name I think 13:25:40 if node resource is not supported, can a cluster have nodes? 13:26:00 haiwei, sure 13:26:22 cluster resource type can work individually without node resource type 13:26:26 haiwei: To support senlin functions, without node resource we can still do autoscale, cause cluster can create nodes itself. 13:26:30 the only dependency is profile 13:26:37 yep 13:27:15 ok, that means we can start nodes without node resource in heat? 13:27:49 we can define desired_capacity when creating cluster resource. then cluster will create desired number of nodes. 13:27:57 haiwei, yes. Just those nodes will belong to a cluster. 13:28:00 haiwei: Yes. 13:28:08 got it 13:28:29 ready to move on? 13:28:43 yes, I'm done 13:28:45 I'm ok 13:28:48 #topic discussions on -dev mailinglist 13:29:11 currently senlin mailing list is not well used 13:29:45 yes, most of our communication happened in irc channel :) 13:29:47 sometimes we think IRC is fast, so we discussed there 13:29:57 i saw that comment from doug 13:30:04 That's true 13:30:12 jruano, yea 13:30:13 but for USA guys like jruano, there are not there maybe 13:30:37 you will miss the discussion sometime, jruano 13:30:42 true 13:31:18 you can only get some information from irc log :) 13:31:19 so for something big, we should use senlin mailing list to make sure everyone can join the discussion 13:31:44 yep 13:31:50 agree 13:31:56 agreed... 13:32:03 I just sent a mail last week, but ... 13:33:01 really need to use maillist better :) 13:33:07 yes 13:33:23 won't miss your message next time I believe 13:33:36 the mail is still there, anyone can reply it 13:33:41 sure 13:34:07 if you have read it, it's ok to discuss it now 13:34:19 #topic open discussion 13:35:30 about auto-scaling for container, I discussed it with qiming, he said senlin had better not do it 13:35:46 because k8s and docker team will do it 13:36:04 haiwei, actually I have the same feeling 13:36:09 senlin should only concern VM layer auto-scaling 13:36:25 yes, that makes sense to me , yanyan 13:36:28 since this may beyond the scope of senlin 13:36:57 yep, we should focus on the clustering service itself in the first step I think 13:36:57 at least that is not the thing we should consider now 13:37:07 to make the servie itself more stable and scalable 13:37:13 haiwei, right 13:37:25 yep... container space is still very fragmented 13:37:36 no need to move into that mess 13:37:56 right, jruano 13:38:26 So we won't support container in near future? 13:38:31 by the way, when will qiming be back to Beijing? 13:38:48 I think so , elynn 13:39:10 elynn, I think senlin itself won't provide related support directly I think 13:39:46 haiwei, in next week I think 13:40:09 next thursday? 13:40:15 and also xinhui 13:40:30 I saw some new guys had joined to senlin, maybe they don't know the weekly meeting's time, like dongbing 13:40:39 I notice that there's discussion about doing autoscaling in magnum. 13:41:06 Do you know if that related to senlin? 13:41:10 we should have invited them to the meeting 13:41:11 yeah... too many people, too fragmented looking to do autoscaling of containers 13:41:11 is it now senile meeting? 13:41:23 yes 13:41:33 haiwei, yes, I think we can give an update in both senlin irc channel and maillist to ensure everyone can get the latest information 13:41:33 welcome yingjun 13:41:40 you are diving now :) 13:41:42 hi, yingjun, welcome :) 13:41:42 :) 13:41:43 tc is already concerned about scope of senlin 13:42:03 so at least for now, no need to worry about container support 13:42:16 hi yingjun 13:42:25 welcome yingjun 13:42:55 jruano, I believe they will help us to make the correct decision :) 13:43:10 hi 13:43:11 yes, I think so 13:43:34 yes, or at least block us until we make the right one they agree with :) 13:43:52 yanyan, you mean we should send a mail to announce the meeting time? 13:44:14 haiwei, yes, I think we can do that 13:44:16 jay from mirantis brought up autoscaling that murano is attempting to do now 13:44:22 which was odd 13:44:22 ok 13:44:50 #action send announcement mail for weekly meeting 13:44:58 jruano, cross-project collaboration is always important 13:45:30 yep 13:46:24 ok, anything else? 13:46:43 nope from me 13:46:47 not for me 13:47:20 ok, that's all for the meeting 13:47:28 #endmeeting senlin