13:00:19 <yanyanhu> #startmeeting senlin 13:00:20 <openstack> Meeting started Tue Oct 18 13:00:19 2016 UTC and is due to finish in 60 minutes. The chair is yanyanhu. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:00:21 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 13:00:24 <openstack> The meeting name has been set to 'senlin' 13:00:31 <yanyanhu> hello 13:00:35 <zzxwill> Hello yanyanhu. 13:00:42 <lvdongbing> hello yanyan 13:00:42 <yanyanhu> hi, zzxwill 13:00:48 <yanyanhu> hi, lvdongbing :) 13:00:59 <Qiming> hello 13:01:11 <yanyanhu> hi, Qiming 13:01:46 <yanyanhu> lets wait for a while for other attender 13:02:29 <yanyanhu> ok, lets start our discussion and others can join us later 13:02:33 <yanyanhu> here is the agenda 13:02:35 <yanyanhu> https://wiki.openstack.org/wiki/Meetings/SenlinAgenda 13:02:49 <yanyanhu> please feel free to add items you want to discuss 13:03:05 <yanyanhu> #topic Newton/Ocata Work Items 13:03:17 <yanyanhu> ops, should be Ocata work items 13:03:26 <yanyanhu> first item, testing 13:03:36 <yanyanhu> performance test, no progress this week 13:03:44 <yanyanhu> API test 13:04:04 <yanyanhu> we may need some improvement here 13:04:36 <yanyanhu> by adding exception message verification for negative API test cases 13:04:56 <yanyanhu> to ensure the exception is expected 13:05:25 <yanyanhu> for example, there are many different cases badrequest exception will happen 13:05:39 <yanyanhu> and we'd better differentiate them 13:05:45 <Qiming> I'd like to propose postpone the message validation work 13:05:46 <elynn> so need to check the body? 13:05:58 <yanyanhu> elynn, yes, possible 13:06:07 <yanyanhu> Qiming, yes, it is no urgent 13:06:14 <Qiming> with the new REST/RPC parameter passing 13:06:39 <Qiming> most (if not all) parameter validation is done by jsonschema 13:06:56 <Qiming> that will give us a consistent message for verify 13:07:02 <yanyanhu> Qiming, great 13:07:26 <yanyanhu> currently, we can only make chaos matching between expected msg string and exception body 13:07:53 <Qiming> take this as an example: https://review.openstack.org/#/c/386942/1/senlin/api/openstack/v1/clusters.py 13:08:24 <Qiming> Line 171 and line 172 will be responsible for all schema validations 13:08:35 <yanyanhu> I see 13:09:00 <Qiming> it will check whether required field is missing, whether the data type is okay, whether there are some patterns followed etc 13:10:04 <yanyanhu> ok 13:10:27 <lvdongbing> How about unexpected field? 13:10:27 <Qiming> e.g. for cluster list request 13:10:28 <Qiming> http://git.openstack.org/cgit/openstack/senlin/tree/senlin/objects/requests/clusters.py#n24 13:10:30 <yanyanhu> I think we can postpone the api test improvement workitem and wait for api/rpc rework is done 13:11:04 <Qiming> it will check whether limit is provided and is an integer and non-negative 13:11:18 <Qiming> previously we were not checking it 13:11:46 <Qiming> unexpected fields can be ruled out using 'additionalProperties: False' at the object level 13:11:59 <Qiming> oslo.versionedobjects is still working on it 13:12:17 <lvdongbing> ok 13:12:18 <Qiming> once such an option is added, all requests will benefit from it 13:12:28 <yanyanhu> I see. 13:12:29 <lvdongbing> great 13:13:41 <yanyanhu> hi, Qiming, so this is part of Versioned Requests work item? 13:14:14 <Qiming> yes 13:14:21 <yanyanhu> ok 13:14:24 <Qiming> but we are not that urgent on this I think 13:14:33 <Qiming> we were not checking unexpected fields yet 13:14:40 <yanyanhu> yes 13:14:49 <Qiming> for cluster create request 13:14:50 <yanyanhu> that won't cause exception 13:15:05 <Qiming> once we modeled it using versioned object, we can get the jsonschema here: http://paste.openstack.org/show/586190/ 13:15:41 <Qiming> nova has been adding hand written schemas to their source tree for this 13:16:09 <yanyanhu> better way is generating it from definition of object storage? 13:16:20 <Qiming> we have just taken a further step in that direction 13:16:21 <yanyanhu> sorry, resource object 13:16:37 <Qiming> you cannot do that 13:16:47 <Qiming> cluster create is different cluster update, for example 13:16:58 <yanyanhu> yes, it is 13:17:24 <Qiming> reading code there: http://git.openstack.org/cgit/openstack/senlin/tree/senlin/objects/requests/clusters.py 13:17:30 <yanyanhu> maybe giving properties different tags? 13:17:37 <Qiming> it is much more easier to maintain 13:18:04 <yanyanhu> yes, saw that implementation 13:18:27 <Qiming> then .. how would you model cluster list? 13:18:48 <Qiming> we are not there yet 13:18:51 <yanyanhu> like the way we are using in API reference :) 13:18:57 <yanyanhu> I see 13:19:16 <Qiming> even the whole versioned object infra is still evolving, we are riding the wave 13:19:27 <yanyanhu> great 13:19:43 <Qiming> e.g. http://git.openstack.org/cgit/openstack/senlin/tree/senlin/objects/fields.py#n54 13:19:58 <Qiming> and this: http://git.openstack.org/cgit/openstack/senlin/tree/senlin/objects/fields.py#n35 13:20:57 <Qiming> so ... the framework has been proved to be more effective, and we get VERSIONing ... 13:21:16 <Qiming> so I have revised the blueprint for this: https://blueprints.launchpad.net/senlin/+spec/objectify-service-requests 13:21:46 <yanyanhu> cool. I think this will be an important basement for our "versioned everything" discussion in design session 13:21:48 <Qiming> if anyone is interested in this, please join me in the effort 13:22:06 <yanyanhu> as an reference implementation 13:22:14 <Qiming> with versioned object, I'm even considering deprecate the senlin.common.schema model 13:22:25 <yanyanhu> Qiming, yes, that is possible 13:22:27 <Qiming> for each API request, do this: 13:22:41 <Qiming> 1. prepare a request class, e.g. https://review.openstack.org/387893 13:22:42 <yanyanhu> we can use this way to validate spec of profile/policy as well 13:23:02 <Qiming> 2. revise the engine service to handle objectified request, e.g. https://review.openstack.org/387894 13:23:27 <Qiming> 3. revise the API layer to generate such a request object, e.g. https://review.openstack.org/387912 13:23:32 <Qiming> that's all 13:24:04 <Qiming> rebase profile/policy schema on ovo would be a next step 13:24:06 <yanyanhu> so the object will be a param of rpc call 13:24:16 <Qiming> yes 13:24:22 <Qiming> just one single call 13:24:28 <yanyanhu> and it will be serialized and deserialized during transmission via rpc 13:24:48 <Qiming> only this one: http://git.openstack.org/cgit/openstack/senlin/tree/senlin/rpc/client.py#n54 13:24:59 <yanyanhu> haha, call2 :) 13:25:13 <Qiming> the 'call2' method will be left there, all other methods will be deleted 13:25:22 <yanyanhu> I see 13:25:47 <Qiming> yes, don't want to break any existing users who are following the master branch 13:25:48 <yanyanhu> maybe we can replace the current 'call' with it after everything is ready 13:25:57 <Qiming> yes 13:25:57 <yanyanhu> I see 13:26:02 <Qiming> we will do that 13:26:16 <yanyanhu> ok 13:26:18 <Qiming> after all transitions are done, we remove the '2' part, because 13:26:24 <Qiming> 2 is not a good word in Chinese 13:26:31 <yanyanhu> LoL 13:26:34 <lvdongbing> haha 13:26:44 <Ruijie_> :) 13:27:13 <XueFengLiu> hehe 13:27:25 <Qiming> sorry for occupying too much time 13:27:28 <yanyanhu> great work. hope everyone can spend some time to see it. 13:27:47 <Qiming> but I believe this is a huge change so everyone else want to understand the 'why' and 'how' 13:28:03 <yanyanhu> no problem at all, it will be the key for senlin to providing backward compatibility and consistency in future 13:28:22 <Qiming> we will be able to provide smooth upgrading ... 13:28:32 <yanyanhu> yep 13:28:43 <lvdongbing> Great work 13:28:47 <Qiming> we rely on versioned object to automatically translate requests between different api versions 13:29:27 <Qiming> if you have 3 engines, they can even run at different versions, hopefully 13:29:55 <yanyanhu> yes. that will be very helpful for service upgrading 13:30:32 <XueFengLiu> Cool 13:30:46 <Qiming> shall we move on? 13:30:49 <Qiming> 30 mins already 13:30:59 <yanyanhu> ok, we can have more discussion later, lets move on to next item 13:31:05 <yanyanhu> HA 13:31:11 <yanyanhu> Qiming, lixinhui_, your turn 13:31:31 * Qiming obviously not working on this thread, :( 13:31:36 <lixinhui_> haha 13:31:43 <yanyanhu> I guess xinhui as well :P 13:31:56 <yanyanhu> for she is preparing the summit topic 13:32:00 <lixinhui_> a demo run through 13:32:06 <yanyanhu> nice 13:32:29 <yanyanhu> so that demo will be part of our HA presentation in summit? 13:32:30 <haiwei_> lixinhui_, what kind of demo are you preparing? 13:32:30 <lixinhui_> for sending notice from Octavia 13:32:32 <elynn> There's a demo for that topic? 13:32:56 <haiwei_> can you explain the demo briefly? 13:33:00 <lixinhui_> there is misunderstanding here :) 13:33:11 <lixinhui_> Ethan 13:33:25 <lixinhui_> and I are preparing chart for summit 13:33:33 <elynn> Okay :) 13:33:40 <lixinhui_> where we will show out ideas, now, and future:) 13:34:01 <lixinhui_> the demo I mentioned is about getting notice from octavia 13:34:36 <lixinhui_> maybe this time we can catch stephan 13:34:40 <lixinhui_> the new ptl to discuss if we can submit some BP for this 13:34:50 <yanyanhu> so notification can now be sent out while lb members' status change? 13:35:02 <yanyanhu> great 13:35:09 <lixinhui_> since this should be useful for senlin and other purpose since it suppose to replace lbaas 13:35:30 <lixinhui_> Yes, yanyanhu 13:35:46 <yanyanhu> receiving notification from lb service is much better than polling the status of lb member for senlin 13:35:55 <lixinhui_> https://specs.openstack.org/openstack/neutron-specs/specs/newton/kill-neutron-lbaas.html 13:36:04 <Qiming> I like event listeners over pollers 13:36:20 <lixinhui_> totally agree, yanyanhu and Qiming 13:36:34 <lixinhui_> the link I just sent is the decision to kill lbaas in next step 13:36:48 <Qiming> good to know someone is dead 13:36:58 <yanyanhu> :) 13:36:59 <Qiming> and ... someone is born 13:37:14 <lixinhui_> :) 13:37:37 <yanyanhu> ok, waiting for good news from you guys 13:37:43 <lixinhui_> sure 13:37:50 <yanyanhu> ok, next one 13:37:52 <yanyanhu> document 13:37:57 <yanyanhu> no progress I guess? 13:38:09 <Qiming> #link https://wiki.openstack.org/wiki/Senlin 13:38:15 <Qiming> Senlin frontpage 13:38:27 <Qiming> at least someone treat it as frontpage 13:38:49 <Qiming> I have updated the page with some pointers to the talks we delievered 13:38:55 <yanyanhu> all those presentations we have made have been added :) 13:39:12 <Qiming> it took me 2 hours this afternoon 13:39:16 <yanyanhu> and also the link to development and user doc 13:39:38 <Qiming> convert to pdf, upload to slideshare, annotate them, then writing a short summary on the wiki 13:40:24 <yanyanhu> you deserve a meal as compensation :) 13:40:30 <haiwei_> quite quick IMO 13:40:39 <Qiming> I will spend some time unfold that Documentation section, because it is still not obvious what we can do with senlin 13:40:57 <Qiming> oh, forgot to mention 13:41:07 <Qiming> I have been talking to the product-wg guys 13:41:27 <Qiming> it is on the product-wg@lists.openstack.org 13:41:42 <Qiming> I was asking why senlin wasn't shown on the openstack software page 13:42:00 <Qiming> the answer is that they require a >10% deployment, or a PTL request 13:42:00 <yanyanhu> saw that mail :) 13:42:25 <Qiming> and the team was raising question about senlin, what is it ... 13:42:34 <Qiming> the wiki wasn't telling them that 13:42:34 <yanyanhu> sigh... 13:42:42 <Qiming> again, we need to improve the wiki 13:43:00 <yanyanhu> yes. maybe some use cases later 13:43:03 <Qiming> maybe adding a use case section will help 13:43:13 <yanyanhu> e.g. the one we cooperate with cmcc 13:43:32 <Qiming> moving terminology to a separate page ... etc 13:43:43 <yanyanhu> great, in this summit, the presentation and video will be availble 13:43:52 <Qiming> will work on that 13:44:00 <yanyanhu> great, thanks for working on it 13:44:17 <yanyanhu> ok, lets move on? 13:44:22 <Qiming> line 20 on the etherpad is done, removing that 13:44:32 <yanyanhu> ok 13:44:45 <yanyanhu> versioned request. I think we have already covered it? 13:44:51 <Qiming> yes 13:45:00 <yanyanhu> ok, next one, container profile 13:45:11 <yanyanhu> hi, haiwei_, any updating about it? 13:45:29 <haiwei_> I have been watching sun these days, they have added image support 13:45:33 <haiwei_> https://review.openstack.org/#/c/383678/ 13:45:52 <Qiming> beijing has been cloudy, no sun here ... 13:46:01 <Qiming> LOL 13:46:06 <yanyanhu> ok, madhuri's work 13:46:10 <yanyanhu> :) 13:46:13 <haiwei_> I am thinking whether we can refer to it 13:46:36 <haiwei_> sorry, my pc converted it automatically 13:46:45 <yanyanhu> yes, maybe we can have some discussion with zun team in summit 13:46:45 <haiwei_> z un is not a word 13:47:04 <Qiming> I hoped so 13:47:05 <yanyanhu> just there is conflict on design session time slot... 13:47:14 <Qiming> but the timeslots all overlapped 13:47:22 <yanyanhu> yes... all on friday morning 13:47:31 <Qiming> there is no possibility to do cross-team talk now 13:47:33 <haiwei_> managing image seems out of senlin's scope 13:47:48 <yanyanhu> haiwei_, yes, it is 13:47:53 <Qiming> yes, so you can keep watching sun, haiwei_, :) 13:48:19 <haiwei_> yes, it's shining 13:48:25 <yanyanhu> :) 13:48:32 <Qiming> move on? 13:48:39 <yanyanhu> ok, lets move on 13:48:43 <Qiming> no update on receiver or event notification I think 13:48:44 <yanyanhu> message receiver 13:48:48 <yanyanhu> yep 13:48:51 <yanyanhu> just added doc 13:48:56 <Qiming> oh, some docs merged, yes 13:49:03 <yanyanhu> hope that will be helpful for user and other developer to understant it 13:49:14 <yanyanhu> any issues or questions, plz just ping me 13:49:24 <yanyanhu> or leave comments in patch :) 13:49:33 <Qiming> Then ruijie has started some work on batching policy, please folks have your eyes on them 13:49:36 <yanyanhu> batch policy 13:49:39 <yanyanhu> yes 13:49:48 <yanyanhu> thanks a lot for working on this, Ruijie_ :) 13:49:59 <Ruijie_> my pleasure, yanyan :) 13:50:02 <yanyanhu> if you need any help, plz find us in irc channel 13:50:21 <Ruijie_> I was working on the cluster_replace_action last week. 13:50:22 <Qiming> didn't we forget to welcome lvdongbing and XueFengLiu on board? 13:50:29 <Ruijie_> Will keep working on batch policy later 13:50:30 <yanyanhu> oh, right! 13:50:36 <yanyanhu> Ruijie_, great 13:50:37 * Qiming applauds hard ... 13:50:39 <yanyanhu> will help to review 13:50:55 <yanyanhu> welcome our new core reviewer, lvdongbing and XueFengLiu 13:50:57 * Qiming keeps applauding ... 13:51:07 <haiwei_> welcome 13:51:10 <lvdongbing> thanks guys :) 13:51:11 <yanyanhu> looking forward to more coodination with you guys :) 13:51:16 <lvdongbing> my pleasure 13:51:19 <elynn> welcome! 13:51:21 <XueFengLiu> hi all 13:51:44 <yanyanhu> and hope more people will join the core reviewer team soon :) 13:51:46 * Qiming keeps applauding ... 13:51:52 <yanyanhu> Qiming, ... 13:51:58 <yanyanhu> don't just applaud 13:52:16 <yanyanhu> maybe some simple self introduction, lvdongbing XueFengLiu ? 13:52:22 <yanyanhu> lvdongbing, maybe you first? 13:52:31 <lvdongbing> OK 13:52:54 <lvdongbing> I come from KylinCloud 13:53:41 <lvdongbing> Have been working with Openstack for 3 years 13:53:43 <yanyanhu> lvdongbing is a veteran of openstack I think :) 13:54:02 <yanyanhu> great 13:54:16 <yanyanhu> hi, XueFengLiu 13:54:30 <yanyanhu> XueFengLiu is from ZTE 13:54:33 <XueFengLiu> I work in ZTE 13:54:40 <XueFengLiu> yes 13:55:31 <yanyanhu> ok, welcome again, you guys 13:55:34 <lvdongbing> Nice to meet you guys in openstack world! 13:55:40 <yanyanhu> me too :) 13:55:44 <yanyanhu> ok, last 5 mins 13:55:48 <XueFengLiu> mee too 13:55:55 <yanyanhu> #topic summit presentation 13:56:17 <yanyanhu> lixinhui_, and Ethan are preparing the HA one 13:56:33 <lixinhui_> Yes, yanyanhu 13:56:36 <yanyanhu> and Qiming and me and cmcc colleague are preparing the large cluster topic 13:56:46 <elynn> yes, might need your review and skilled experience ! 13:56:50 <lixinhui_> Will schedule meeting with you and Qiming 13:57:00 <lixinhui_> Yes, Ethan 13:57:02 <lixinhui_> :) 13:57:06 <yanyanhu> no problem, please send out the slides and we can help to review 13:57:13 <lixinhui_> cool 13:57:30 <yanyanhu> ok, next one 13:57:35 <yanyanhu> irc meeting reschedule 13:57:48 <yanyanhu> I think we have no enough time to discuss this 13:57:52 <yanyanhu> maybe next time 13:58:04 <yanyanhu> BTW, I will send mail to cancel next irc meeting 13:58:14 <yanyanhu> for most of you guys will be in Barcelona next week :) 13:58:20 <Qiming> just one question, maybe we can discuss it on senlin channel 13:58:26 <yanyanhu> Qiming, sure 13:58:38 <yanyanhu> if you guys have time, we can make a quick discussion in senlin channel after meeting 13:58:47 <Qiming> do you prefer having this meeing at night (this same time), or moving it to day time in China timezone 13:59:20 <Qiming> let's continue the discussion on senlin for a few minutes 13:59:25 <yanyanhu> ok, no further items in agenda 13:59:27 <Qiming> and release this channel? 13:59:32 <yanyanhu> Qiming, sure 13:59:39 <yanyanhu> thanks you guys for joining :) 13:59:50 <lvdongbing> :) 13:59:51 <yanyanhu> lets move back to senlin channel 13:59:56 <haiwei_> thanks 13:59:56 <yanyanhu> #endmeeting