13:00:54 #startmeeting senlin 13:00:55 Meeting started Tue Oct 13 13:00:54 2015 UTC and is due to finish in 60 minutes. The chair is yanyanhu. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:00:56 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 13:00:59 The meeting name has been set to 'senlin' 13:01:11 hello 13:01:17 hi 13:01:26 hi 13:01:37 Where is Qiming 13:01:41 since Qiming is in another call conference, I will hold this meeting :) 13:01:49 hope he can join us later 13:01:51 Okay :) 13:01:58 fine 13:02:08 please feel free to add items to the agenda 13:02:16 #link https://wiki.openstack.org/wiki/Meetings/SenlinAgenda 13:02:43 #topic deletion policy handling CLUSTER_DEL_NODES 13:03:27 so the first thing we may want to talk about is the patch about deletion policy proposed by haiwei 13:03:44 hi, haiwei, could you please give us some summary about it? 13:04:00 although I guess you're still working on it 13:04:02 o/ 13:04:04 yes 13:04:08 hi, elynn :) 13:04:30 deletion policy's TARGET contains CLUSTER_DEL_NODES action 13:04:56 but cluster_del_nodes action will delete specific nodes 13:05:20 so I thought deletion policy should not take effect when this action happens 13:05:44 maybe some of you will think the same 13:05:54 hmm, actually we don't need deletion policy to choose candidates for CLUSTER_DEL_NODE action 13:06:23 but Qiming noticed me there are other rules like grace_period and destroy_after_delete in deletion policy which will work for cluster_del_nodes action 13:06:51 yes, we just made a discussion about these two properties today 13:06:53 right, yanyanhu 13:07:33 these two rules are not working currently , I will try it fix it 13:07:43 actually we thought that maybe we should let end user to decide whether they want to destroy a node completely after it is deleted from a cluster 13:07:54 and also the grace period setting 13:08:02 https://review.openstack.org/#/c/232953/ this is the patch 13:08:43 yes, end user can define it in both deletion policy file and the command line 13:09:05 so the basic rule is if users specify the exact nodes they want to delete from cluster, they need to take responsibility to decide these two settings 13:09:07 and we will add a destroy_after_delete option for it 13:09:21 ok 13:09:27 yes 13:09:44 and about the comment I left before, I think I was wrong 13:09:57 welcome to review the patch after I pushed it 13:10:29 we can discuss it yanyanhu, if you have any questions 13:10:38 once user decided the deletion candidates, the deletion policy should not deny it 13:10:47 I also made mistakes at first 13:11:02 yes 13:11:28 ok, lets make further discussion about this patch later 13:11:31 the point is grace_period and destroy_after_delete rules 13:11:37 yep 13:11:37 ok 13:12:01 we can move on 13:12:10 #topic two placement policies or just one 13:12:53 hi, lixinhui, I saw you and liuwei are also working on this workitem recently 13:13:22 Qiming just proposed a patch of placement policy for cross-region yesterday 13:13:32 yes 13:13:37 I saw it 13:13:40 it is from the one for cross-az I think 13:14:21 I will find more time to review taht 13:14:34 so we haven't decided whether we should combine these two placement policies since their logic of building placement plan are the same 13:14:38 lixinhui, thanks :) 13:14:39 but not sure what does that mean by the topic 13:15:30 we hope to user the same placement policy to support cross-az and cross-region 13:15:40 if there is no much conflict 13:15:49 s/user/use 13:16:21 okay 13:16:25 but it seems a little difficult 13:16:37 yanyanhu, currently which two patches are almost the same, I saw three patches about placement policy, Qiming, lixinhui and liuwei 13:16:43 but I think more complex for across region 13:16:52 yes they both seem to be addressing placement 13:16:54 I think liuwei and qiming's 13:17:04 jruano, yes 13:17:12 makes sense to try to use one 13:17:16 lixinhui, right, we found some problems here 13:17:29 since region is usually carried in context and is supported by most openstack services 13:17:39 but az is different 13:18:00 only limited services support it (only nova and cinder now I guess)? 13:18:08 yes 13:18:11 yanyanhu 13:18:16 Is there any benefit to use one instead of two? 13:18:43 good question, elynn 13:18:56 from my point of view 13:18:59 elynn, reduce the complication of policy type managing I guess :) 13:19:29 I think there should be more region specfic logic there instead of current code 13:19:34 I think one is better 13:19:41 I balancing 13:20:16 hmm, actually I also discussed this with Qiming this afternoon and we also haven't got conclusion... 13:20:20 by design, liuwei's policy can work together with mine 13:20:21 people will get confused by two placement policies if they are actually doing the same job 13:20:28 lixinhui, yes 13:20:30 I think so 13:21:07 haiwei, but putting them together could make the implementation of placement policy very complicated 13:21:12 lixinhui, that means the patches are doing different jobs? 13:21:23 yes, yanyanhu 13:21:28 but region is different 13:21:28 agree with yanyanhu 13:22:00 I guess maybe we can split common functiona out 13:22:01 despite now, the logic may looks similar but there should be more difference in future 13:22:20 I am not familiar with placement policy, what about making one multi-region placement policy, the other multi-az placement policy?? 13:22:48 :) 13:22:58 haiwei, yes, maybe we should follow this way 13:23:09 extract common function would be a good idea. 13:23:11 I am not sure if this is a right moment to work on trgion 13:23:20 But keep them two policy 13:23:41 what scenarios to drive region work? 13:24:07 lixinhui, you mean the use case of cross-region node placement? 13:24:13 yes 13:24:30 I think hybrid cloud is one of them 13:24:40 oh, okay 13:25:12 lixinhui: And site disaster recovery 13:25:25 okay 13:25:41 useful but complex 13:26:07 does current logic of across region well serve the two cases? 13:26:25 e.g. you have a small region which undertake your workload in most time. But one day, you find the capacity of this region is insufficient for your workload, you may want to scale into another larger region:) 13:26:28 yes, it might be complicate to handle multi region placement. 13:27:10 elynn, lixinhui, yes, we need to think it through 13:27:48 okay, I will read the patch then catch you all for more discussio 13:28:14 ok, I think we need more discussion about this topic. Maybe we can first look through these three patches and then make more talk about it :) 13:28:23 lixinhui, thanks:) 13:28:37 ok, lets move on 13:28:49 #topic big tent proposal review 13:29:10 I saw Qiming just propose a draft about application for bigtent 13:29:18 https://review.openstack.org/#/c/234162/ 13:29:26 yes 13:29:35 this is a hard work :) 13:30:10 hope you guys can help to review and any comments will be very helpful 13:30:38 What is big tent for? 13:31:08 Changing namespace from stackforge to openstack? 13:31:51 hi, elynn, I think it's about the principle for accepting new projects who want to join openstack family 13:32:09 elynn, nope I think. That should be the retirement of stackforge 13:32:13 elynn, the big tent means all the openstack project 13:32:16 let me try to find the patch 13:32:44 https://review.openstack.org/192016 13:33:11 hi, elynn, this is the patch and discussion about the retirement of stackforge 13:33:36 so I guess also stackforge projects will be migrate to openstack namespace after 17th Oct. 13:33:41 if they will 13:34:09 OK, got it. 13:34:13 big tent carries atc as well 13:34:27 but if a project wants to get the approval of TC, it needs to propose the application and accept review 13:34:34 jruano, yes 13:34:41 If a project in big tent , that means it's an official and mature project? 13:34:47 yes 13:34:51 elynn, I guess so 13:34:57 Cool 13:34:58 blessed by the tc 13:35:02 we usually call them 'TC-approved' projects 13:35:06 yep 13:35:07 maybe not mature, but official project 13:35:09 :) 13:35:23 so we really should try :) 13:35:32 definitely 13:35:45 hope more people can join us to make the project better :) 13:35:51 so help review the patch 13:35:54 stronger, more mature 13:36:32 yes, appreciated for any suggestion :) 13:37:04 Will review the patch after meeting ;) 13:37:09 thanks :) 13:37:36 ok, let move to next topic 13:37:40 #topic PTL election 13:37:57 umm... 13:38:12 Qiming put this item in the agenda 13:38:26 it's strange that senlin doesn't have a PTL yet :) 13:38:35 we need to make the election now? 13:38:42 I thought he was the PTL 13:38:44 I thought Qiming was the PTL 13:38:58 at least we have a candidates now:) 13:39:20 I think since we will elect it, we should do it formally 13:39:30 like other projects 13:39:39 ok, so let me start a vote 13:39:40 use the vote tool 13:39:55 cool, yanyanhu 13:39:56 He is the only candidate, still need to vote? 13:40:09 actually we dont I think :) 13:40:30 anyone else want to join the competition? 13:40:35 hmm 13:40:39 :) 13:40:40 Haha 13:40:42 welcome 13:40:52 hahaha 13:41:14 it is not bad to vote 13:41:18 probably needs to be formal for move to big tent 13:41:23 actually after getting familiar with the project, everyone can have a try I think 13:41:33 jruano, yes, I guess so 13:42:07 ok, so since only one candidate now, I think we can make the decision 13:42:08 i vote qi ming :) 13:42:27 +1 13:42:29 Qiming +1 13:42:53 +1 from me 13:43:23 ok, we have decided it 13:43:29 Qiming +1 13:43:50 hopefully he accepts 13:43:53 what about senlin-dashboard's PTL? 13:43:58 jruano, ;p 13:44:01 the same? 13:44:03 lol 13:44:07 haiwei, nope I think 13:44:18 they need their own PTL I guess 13:44:22 need a PTL also? 13:44:32 not decided here 13:44:34 ok 13:44:38 not very sure about it, but I guess so 13:44:56 I will talk with Qiming and zhengguo about it 13:45:19 senlin-dashboard could have the same PTL -> Qiming 13:45:41 maybe zhengguo will run for it 13:46:05 yes, so I guess a discussion is needed :) 13:46:29 since both them are not here, maybe we can talk about it later 13:46:56 #topic open discussions 13:47:08 so now is the open discussion 13:47:56 anyone of you guys have decided to join the summit? 13:48:05 about the big tent application patch , when is the deadline of review locally? 13:48:19 I should be there 13:48:19 i will go to the summit 13:48:22 Don't get the ticket from IBM T_T 13:48:27 haiwei, I'm not sure about it. But I guess there is no deadline for this kind of application 13:48:35 elynn, sigh... 13:48:51 hi, jruano, will you be there 13:48:57 and also lixinhui ? 13:48:58 yes i am going 13:49:01 cool 13:49:07 I mean we will push another patch for TC to review after we review it locally 13:49:26 will you be there yanyan 13:49:38 I think you can discuss this patch at summit 13:49:43 oh, haiwei, I think after we get feedback from all guys, we can propose the applicatoin formally 13:49:47 jruano, yes :) 13:49:58 I will go to tokyo with Qiming together 13:50:03 great. we should do a meetup 13:50:07 elynn, sure 13:50:13 definitely 13:50:21 yes, so I think we need a deadline for the formal proposal, yanyanhu 13:50:59 haiwei, yes, I guess we will make the proposal before summit 13:51:10 we should try to be accepted before the summit 13:51:11 yes 13:51:26 just try our best:) 13:51:44 is there any information about the meeting room for senlin team? 13:51:55 haiwei, no yet ;( 13:52:02 it seems very limited 13:52:10 I guess the meeting room is very limited 13:52:12 yes 13:52:26 hope we can get one 13:52:43 in worst case, we can have a meal or coffee :) 13:52:50 there is no big enough place to hold so many people in Tokyo 13:53:01 and make our discussion 13:53:36 :) 13:54:15 ok, is there any other topics you guys want to discuss? 13:54:22 we still have 5 mins left 13:54:58 no from me 13:55:03 im good 13:55:26 not for me 13:55:27 hi, elynn, lawrancejing, lixinhui, anything else? 13:55:49 ok, so lets end the meeting 13:56:02 thank you so much for joining :) 13:56:11 by the way, good job for charging the meeting 13:56:21 thanks :) 13:56:21 yanyanhu 13:56:34 lets move back to senlin channel and release the meeting channel 13:56:41 #endmeeting