13:00:34 #startmeeting senlin 13:00:35 Meeting started Tue Jan 5 13:00:34 2016 UTC and is due to finish in 60 minutes. The chair is Qiming. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:00:36 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 13:00:39 The meeting name has been set to 'senlin' 13:00:44 hi 13:00:52 hello 13:00:55 hi 13:01:14 hi 13:01:55 pls revise agenda if you have things to add 13:02:05 #link https://wiki.openstack.org/wiki/Meetings/SenlinAgenda 13:02:25 hello 13:02:55 let's check the etherpad 13:03:01 #link https://etherpad.openstack.org/p/senlin-mitaka-workitems 13:03:18 heat resource type 13:03:25 yes 13:03:34 mostly blocked by the header thing and action id 13:03:38 two patches are already merged 13:03:46 we have discussed that this afternoon 13:03:52 on block by openstacksdk 13:04:09 we still don't support cluster update properly I think 13:04:19 I will according to your suggestion to work on senlin side. 13:04:28 that would be great 13:04:35 senlinclient unit tests 13:04:52 finished? 13:05:04 v2/shell.py is finished 13:05:07 v1 13:05:09 mostly I believe 13:05:16 should we do something else? 13:05:34 coverage is 86^ 13:05:37 86% 13:05:40 pretty good 13:05:56 this one: senlinclient/openstack/common/cliutils.py 13:06:06 it only has 32% coverage 13:06:07 oh 13:06:14 it is from oslo 13:06:35 not easy to test? 13:06:47 I recalled that something is happening in oslo regarding this module 13:06:53 cannot remember the details 13:07:06 haiwei can you help do a search on mailinglist? 13:07:27 about oslo's change? 13:07:29 this is the only one we are relying on oslo incubation 13:07:46 maybe it has already graduated? 13:08:00 ok, I will check it 13:08:03 senlinclient/common/utils.py 13:08:08 this one has 42% 13:08:43 we are about to remove senlinclient/v1/models.py 13:08:54 since models are all supported in sdk now 13:09:06 great 13:09:30 service calls are all revised 13:10:01 health policy 13:10:26 we do have some progress on this, from lixinhui 13:10:44 we are adding do_check and do_recover for profiles 13:10:56 saw the patches 13:11:05 good start point I think 13:11:07 those will serve the foundation for failure detection and recovery 13:11:58 node update support in profiles 13:12:11 worked on metadata update support last week 13:12:19 patch in senlin side is ready 13:12:28 yep, saw that, brian has some comments 13:12:34 but it depends on the metadata refactoring in sdk side 13:12:35 good job, yanyanhu 13:12:38 maybe we should help push that work 13:12:39 brian is working on it I think 13:12:40 yep 13:12:52 will wait for brian's job 13:12:59 last time I checked it, it is still merge conflict 13:13:12 we'll have to push him, :) 13:13:12 yes, it is still in that status now 13:13:18 thanks :) 13:13:32 seems they are waiting for some work about image metadata 13:13:41 but I guess progress is not good 13:13:48 if he doesn't have cycles, we can take it over 13:13:58 yes 13:14:07 will keep looking at it 13:14:15 just need to understand server_metadata and metadata 13:14:27 the difference and the direction 13:14:33 yes, actually I think the latest design is better 13:14:40 okay 13:14:40 it's more clear for user 13:14:47 good 13:14:58 although the old one follows the sdk design a little bit better 13:15:35 will spend some time on disk update support, will do some investigation first 13:15:52 it is always difficult to unify things 13:15:54 cool 13:15:56 yep 13:16:01 receiver 13:16:13 haven't started this job 13:16:17 we are removing webhook functional tests 13:16:22 yep 13:16:28 will work on it soon 13:16:42 to recover our functional test :) 13:16:44 next time we work on a receiver test, please consider some actions simpler 13:16:58 yep 13:17:03 understand 13:17:05 because we are already testing scaling operations in other functional tests 13:17:26 yes, so just testing webhook workflow is enough 13:17:38 as we are adding more features, functional tests may eventually take a lot time 13:17:50 just be careful when moving forward 13:17:56 ok 13:18:01 release notes 13:18:16 proposed a patch containing most backlogs 13:18:27 saw your patch adding omitted release note 13:18:41 have +2 it 13:18:42 just realized how much we have accomplished .... really amazing 13:19:06 https://review.openstack.org/#/c/263083/ 13:19:39 merged now 13:19:42 :) 13:19:47 cool 13:19:52 so quick 13:19:54 please feel proud about ourselves, :) 13:20:08 lock breaker 13:20:12 yes 13:20:32 I'm thinking about adding a db table to record engine status 13:20:33 I'm blocking your reenabling proposal, elynn 13:20:40 yep 13:20:42 like other projects do 13:20:45 that would be great 13:20:58 maybe a dlm ? 13:20:59 so we don't need to call rpc to check engine status 13:21:14 dlm means? 13:21:21 distributed lock manager 13:21:32 search back mailinglist 13:21:40 there have been some discussions on it 13:21:49 we may need a blueprint for this 13:21:53 I don't know what it is now, will search for it first. 13:21:59 yes, I remeber there is a long thread talking about this 13:22:04 it is more complicated than we thought 13:22:41 it is a solution for distributed processes to notify each other reliably 13:23:07 there is a project tooz 13:23:19 just heard of its name, no details 13:23:43 Seems we have a lot more project I don't know but helpful 13:23:55 deletion policy for cross-az/region 13:23:59 I will search for it. 13:24:17 elynn, yep, share the same feeling 13:24:23 just discussed it this afternoon 13:24:43 so ... whenever we start to work on some features, first thing is to check if there are existing wheels 13:24:49 it seems the use case we need will determine the design 13:24:57 yes, haiwei 13:25:03 really a good discussion 13:25:16 yes, it is 13:25:18 we should list out what we need 13:25:25 we won't be able to cover all use cases 13:26:00 one design philosophy we had when proposing these policies is that we want them to be orthogonal 13:26:05 will summarize it in a ether pad page for discussion 13:26:12 okay 13:26:18 okay 13:26:22 great :) 13:26:51 currently I have no idea about cross region 13:26:53 I do agree that all the problems we are solving can be combined into fewer number of policies 13:27:05 it maybe similar to cross-zones 13:27:10 yes 13:27:53 let's rethink about the design if we are not so sure where we are heading 13:28:04 yes 13:28:08 ok 13:28:33 etherpad discussion would be good 13:29:01 #topic mid-cycle meetup plan 13:29:14 #link https://etherpad.openstack.org/p/senlin-mitaka-midcycle 13:29:34 so we have a schedule for a 2-days discussion 13:30:01 first day is mostly about meeting users 13:30:11 can we make the schedule more specific? for example add time period 13:30:27 good suggestion 13:30:47 will try to contact presenter to confirm the time slot 13:30:52 also including some coffee time 13:31:13 sure :) 13:31:21 thanks 13:31:23 #action Qiming to refine meetup schedule (add **coffee** time) 13:31:58 haiwei, do you have something to share in the afternoon? 13:32:45 I have no idea 13:32:54 okay 13:33:03 lixinhui, how about you? 13:33:08 maybe about tempest? 13:33:15 the second afternoon? 13:33:31 haiwei, good topic, I'm intereted in it :) 13:33:45 ok, I will prepare for it 13:34:39 haiwei, will add a quality assurance session for you 13:35:05 I am working on health_manager 13:35:17 feel free to add other things you want to talk about 13:35:31 I do not know if proper to share some work about this 13:35:40 so we will talk in Chinese, right? 13:35:40 lixinhui, 'health management' session will be yours 13:35:50 oh 13:36:00 Thanks, Qiming 13:36:06 Qiming, about those topics related to Senlin, we should prepare presentation or they are feel talk? 13:36:13 haiwei, you can speak Japanese if you like 13:36:28 it would be mostly free talk 13:36:46 I like free talk 13:36:59 got it 13:37:05 :) 13:37:12 certainly you can create charts to show your design idea or whatever 13:37:14 but we can have a holder for them :) 13:37:43 It would be nice if Yanyan can share something about scalability analysis 13:37:59 currently who are 100% sure to be present? 13:38:09 ok, will try to share something about it :) 13:38:18 we do meet some issues about DB concurrency 13:38:21 Great, thanks, yanyanhu 13:38:22 all people who just talked 13:38:41 and some names appeared on etherpad 13:38:50 plus ... some people I don't know 13:38:56 lixinhui, no problem :) 13:39:12 yanyanhu, 'container management' session is yours 13:39:20 ok 13:39:36 cool 13:39:41 just haven't got too much idea how to talk about it :P 13:39:47 cool, I can learn a lot from your guys. 13:39:57 maybe some experience about mesos? 13:40:01 emm, if you like to lead a dicussion session, please sign your name there 13:40:25 yanyanhu, day 2 is mostly about senlin design 13:40:50 it would be good to share experiences learnt from mesos anyway 13:41:10 I'm interesting in mesos 13:41:10 understand, just try to think about something we can work on container management :) 13:41:20 right 13:41:35 interesting 13:41:40 #topic open discussion 13:42:02 will IBM offer coffee and cake? 13:42:10 for the 2 days 13:42:22 lunch, maybe 13:42:23 lixinhui, you can bring some from vmware if you can 13:42:32 ... 13:42:41 LoL 13:42:46 how about the lunch? 13:42:57 don't worry 13:42:57 that's important 13:43:07 enen, haiwei 13:43:17 we will try not to starve you 13:43:22 can we have si cai yi tang? 13:43:26 haha 13:43:39 we may order lunch boxes 13:43:43 you can, haiwei :) 13:43:56 too many topics during the day 13:43:58 hello, Daisy 13:44:04 si cai yi tang for each :) 13:44:07 in fact, I want to eat lunch boxes 13:44:10 elynn, ... 13:44:12 I miss it 13:44:25 Hi, yanyanhu 13:44:26 god 13:44:44 we can find some nice place for dinner together though 13:44:54 sure 13:44:59 that's important :P 13:45:10 critical? 13:45:28 absolutely 13:45:33 :) 13:45:51 no north gate resaurant of software park 13:45:59 btw, I have released senlinclient 0.2.1 today, seeing complaints about compatibility 13:46:26 we are in a good shape now, regarding the client side 13:46:55 why not after deleting v1/models? 13:47:11 doesn't harm though 13:48:00 didn't realize we still have garbage, :( 13:48:09 another heads up 13:48:10 https://review.openstack.org/#/c/258222/ 13:48:45 we have promised to do official release this cycle, together with all those famous projects 13:49:31 I understand your urge mode 13:49:50 hmm, do need to fix those bugs marked as critical ASAP 13:49:59 yep 13:50:27 http://docs.openstack.org/releases/schedules/mitaka.html 13:50:29 we still have? 13:50:49 Jan 16-22 would be m-2 milestone 13:51:01 haiwei, yep, e.g. lb policy doesn't work now 13:51:11 after meetup, we will do a m-2 freeze 13:51:32 I have tested lb policy a few days ago, got an 403 13:51:43 we can discuss more details about things we can finish in this cycle 13:52:33 haiwei, soft delete is removed now, so some logic about deleting lb member in lb policy doesn't work 13:52:59 ok, yanyanhu, I will read the source 13:53:04 403 is a rejection? 13:53:21 yes, 403 usually means policy check failed 13:53:37 it seems only admin user can get loadbalancer 13:53:50 what?! 13:54:30 hmm, didn't meet this issue before 13:54:32 we are requesting loadbalancer list , right? 13:54:33 will check it 13:54:44 pls double check 13:55:00 ok, I am not sure about that 13:55:07 will test it again 13:55:11 haiwei, I remeber we didn't do loadbalancer list in lb_policy 13:55:19 will also check it 13:56:10 haiwei, if it does exist, you can report a bug for it :) 13:56:16 4 mins 13:56:20 ok 13:56:36 we'll skip weekly IRC meeting next week 13:56:50 cool 13:57:07 haiwei, did you find proper hotel 13:57:10 more accurately, replace it with a face-to-face, beer-in-hand meeting 13:57:16 yes 13:57:20 :) 13:57:21 nice 13:57:22 on zhichun road 13:57:24 Will there be a conference call hookup for the meet? 13:57:27 ok 13:57:29 wow 13:57:35 not far from IBM I think 13:57:41 cschulz, it will last a whole day 13:57:42 very close to VMware, haiwei 13:57:54 Qiming, maybe we can enable a callin number? 13:58:00 we can use a etherpad though, cschulz 13:58:03 haha, have a visit to your company? 13:58:06 but not sure we can get a passcode 13:58:20 Welcome, haiwewi 13:58:23 :) 13:58:52 yanyanhu, we can 13:59:03 webex 13:59:23 cool, some people who can't be onsite can join by calling in 13:59:24 if need, I can share my account out 13:59:27 time's up 13:59:40 switch to #senlin now 13:59:43 see you 13:59:46 thanks 13:59:47 #endmeeting