13:00:28 #startmeeting senlin 13:00:29 Meeting started Tue May 17 13:00:28 2016 UTC and is due to finish in 60 minutes. The chair is Qiming. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:00:30 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 13:00:33 The meeting name has been set to 'senlin' 13:00:46 evening 13:00:58 Hi~ 13:00:59 hi 13:01:03 Hi 13:01:09 o/ 13:01:38 oops 13:02:00 if you have item to add to the agenda 13:02:17 please use this link: https://wiki.openstack.org/wiki/Meetings/SenlinAgenda 13:02:35 as usual, let's check the etherpad first 13:02:44 #topic netwon work items 13:02:51 #link https://etherpad.openstack.org/p/senlin-newton-workitems 13:03:23 blockers for scalability test 13:03:52 with experience sharing from CMCC on CCTC, I believe we can do something more 13:04:21 the general suggestion is to have separate rabbitmq instances 13:04:31 I'm removing that item for now 13:04:50 what okay 13:04:59 Okay, Qiming 13:05:01 tempest testing 13:05:21 ethan was working on it last week 13:05:22 elynn has commited patch bug the gate is actively rejecting it 13:05:26 I'm maintain an etherpad for tests we need 13:05:39 https://etherpad.openstack.org/p/senlin-tempest-plugin 13:06:14 and also submit a patch for enable gate for it. 13:06:27 https://review.openstack.org/#/c/316544 13:06:46 Just saw your comment, will change it later :) 13:07:04 And will continure to add rest tests, then negative tests. 13:07:11 That's it. 13:07:25 jenkins_jobs.errors.JenkinsJobsException: Failed to find suitable template named '{pipeline}-senlin-dsvm-tempest-{test_type}-{backend}{job-suffix}' 13:07:46 nice 13:07:46 seems the job name is no good 13:07:46 I don't know what qualifies a good job name 13:07:46 but having cloud_backend name injected into that string is weird 13:07:47 we should have named the dummy backend just 'dummy' instead of 'openstack_test' 13:07:51 looks like the job template definition is incorrect 13:07:53 it is bad name 13:08:24 {backend] 13:08:35 oops , should change ] to } 13:08:38 other than that, I'd also suggest we remove {backend} from the job name template 13:08:50 right 13:09:04 Why? 13:09:11 if remove {backend} from name 13:09:21 I don't think we need to do each test type using two or more backend drivers 13:09:29 How do we pass the backend parameter? 13:09:46 for api_test, it will only use the dummy driver, that is fine 13:10:00 hi, ethan, maybe we can tell it based on the test-type :) 13:10:01 for scenario test, we may need to do things with real drivers 13:10:15 e.g. for api, default backend is dummy/openstack_test 13:10:23 yup 13:10:36 okay , I get it. 13:10:57 Then I just judge the test_type and decide what kind of backend to use. 13:10:59 each test type has a goal, we don't need too many combinations 13:11:11 sounds a life saver 13:11:40 so from etherpad, I can remove line 6-8, right? 13:11:52 just leave line 9 there, and remove it when gate is up 13:12:12 yes, I think so 13:12:16 yes 13:13:09 xinhui, do we have news on stress tests? 13:14:13 These days 13:14:22 I am packing everything together 13:14:49 I prepare use the same test bed for a demo for integration between VIO and Senlin 13:15:02 after that, we can continue the stress things 13:15:11 you mean you are building an end-to-end solution demo? 13:15:12 based on the first item just discussed 13:15:25 just first step 13:15:30 okay 13:15:32 use ansible to deoloy Senlin 13:15:39 with VIO together 13:15:40 would love to see how that works 13:15:47 sure 13:15:50 still some problems 13:16:02 for the network settings in VIO env 13:16:02 ask questions on IRC then 13:16:06 okay 13:16:11 thanks in advance 13:16:14 okay, ask your colleagues, not on IRC 13:16:20 :) 13:16:23 both I think :) 13:16:28 np 13:16:39 Rally support ? 13:16:52 patch #301522 is there for some time now 13:17:06 yes, didn't get time to work on it last week 13:17:11 need to pick it up 13:17:22 yes, it has been more than 40 days 13:17:28 hope in next week, push a new patchset 13:17:41 yes 13:17:55 you mean after May 24? 13:18:13 nope, the weeking from tomorrow :) 13:18:16 week 13:18:29 just tried to clarify 13:18:33 cool 13:18:42 Health 13:19:13 neutron bug is still there, lixinhui ? 13:19:26 yes, Qiming 13:19:30 today, I checked the code 13:19:39 will discuss with Kobi 13:19:39 would be nice you paste the bug link, we need to revisit it now and then 13:19:47 who is the neutron core 13:19:54 okay 13:19:56 okay 13:20:20 btw, I just spent this afternoon on the HA etherpad 13:20:53 Please read, comment, fix, add ... 13:20:53 https://etherpad.openstack.org/p/senlin-ha-recover 13:20:59 okay 13:21:01 will read 13:21:08 https://bugs.launchpad.net/neutron/+bug/1548774 13:21:09 Launchpad bug 1548774 in neutron "LBaas V2: operating_status of 'dead' member is always online with Healthmonitor" [Undecided,Confirmed] - Assigned to Yang Yu (yuyangbj) 13:21:10 this one 13:21:33 healthmonitor does not update the status of member in v 13:21:35 2 13:22:09 that is the critical feature we need 13:22:33 documentation 13:22:54 we have completed the migration of senlin API doc from WADL to the new format 13:23:09 thx yanyan for the help 13:23:19 my pleasure :) 13:23:42 it still has some room for improvement for sure, but the content, look-and-feel are all in good shape 13:24:32 em .. seems I need to write some other articles ... 13:25:27 haiwei_, are you working on improving the container profile? 13:26:06 or, do you believe we can leave it as is 13:26:19 errhh... seems he is asleep? 13:26:43 engine 13:26:54 I am thinking about shall we wait for Higgins or do by ourselves? 13:27:21 I'm signing up on the NODE_CREATE, NODE_DELETE rework 13:27:35 haiwei_, if you are asking my suggesion 13:27:39 I guess it will take some time for higgins to provide basic support for container management 13:27:56 I think we should do it using docker directly 13:28:27 when higgins does provide an API to use, we add that as another driver 13:28:52 then we need to have a discuss on how to implement it, the impact to the existing code 13:29:05 my current feeling is that it will take a looooong time to get higgins up 13:29:09 I will rework the spec file 13:29:25 okay, looking forward to a revision 13:29:57 line 32-38 was added by chuck 13:30:08 agree. But hope higgins API can become stable ASAP, especially those interfaces for basic container management, e.g. CUDR 13:30:17 by using docker api directly, most things are in my mind, still some issues are not sure 13:30:48 how about we work and discuss 13:31:20 yes, we can discuss in the spec patch 13:31:56 you know, there are many verbs for patchs: commit, abandon, merge and revert 13:33:49 but if the new change will affect the existing API or database, shouldn't we be very careful? 13:34:49 that is for sure 13:35:04 we are not changing the existing API 13:35:35 maybe move to a new version 1.1 13:35:51 we already have micro-version implemented 13:36:06 let's see if we do need to change the API 13:36:33 if we do, we will need to ensure we NEVER break user land 13:37:38 sure 13:37:45 there are always a million items on the todo list, what we can do is to prioritize, work on them one by one 13:39:54 okay, that is all for things on etherpad 13:40:08 #topic open discussions 13:40:55 anything you want to discuss? 13:41:12 nope from me 13:41:36 nothing from me. 13:41:52 No for me 13:41:53 from the mail list, tacker seems doing auto-scaling things 13:42:03 yes, I saw that 13:42:11 but they don't want to use seniln currently 13:42:26 em, I have left some comments there 13:42:40 but we cannot force people to do this or that 13:43:00 tacker has relationship to nfv, I will spend some time on it 13:43:28 yes, want to learn their need first 13:43:50 that is all from me 13:44:21 tbh, I really don't like the fact that people tend to ignore suggestions from "outsiders" 13:45:07 we have learned our lessons in a hard way, and now they want to experience that again ... 13:45:21 and ... if you are watching mailing list 13:45:38 magnum wants to reinvent a cluster management thing 13:46:08 yes, saw that mail 13:46:28 have the link now? 13:46:32 actually, senlin can help them achieve their goal 13:46:44 in a much easier way 13:46:48 in easier way 13:46:50 yea 13:46:50 with less effort 13:47:12 Which mail? 13:47:15 anyway, we did what we had to do 13:47:29 Agree 13:47:50 http://lists.openstack.org/pipermail/openstack-dev/2016-May/095044.html 13:48:32 okay, if nothing else from you 13:48:38 let's call it a day 13:48:40 thanks 13:48:42 thanks for joining 13:48:46 #endmeeting