13:01:10 <yanyan> #startmeeting senlin
13:01:10 <openstack> Meeting started Tue Oct 11 13:01:10 2016 UTC and is due to finish in 60 minutes.  The chair is yanyan. Information about MeetBot at http://wiki.debian.org/MeetBot.
13:01:11 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
13:01:13 <openstack> The meeting name has been set to 'senlin'
13:01:19 <Qiming> o/
13:01:22 <yanyan> hello
13:01:38 <Ruijie_> ^^
13:01:40 <zzxwill> Hi Qiming, yanyan.
13:01:41 <yanyan> :)
13:01:58 <elynn> hi
13:02:01 <yanyan> hi
13:02:07 <yanyan> welcome you guys
13:02:20 <lixinhui_> hi
13:02:24 <yanyan> this is our first meeting in Ocata cycle I think :)
13:02:27 <yanyan> hi, xinhui
13:02:44 <yanyan> here is the agenda, plz feel free to add items you want to discuss
13:02:45 <yanyan> https://wiki.openstack.org/wiki/Meetings/SenlinAgenda
13:03:09 <yanyan> #topic ocata work items
13:03:15 <yanyan> https://etherpad.openstack.org/p/senlin-ocata-workitems
13:03:30 <yanyan> I have created a new etherpad to track workitems in ocata cycle
13:03:44 <yanyan> currently, I just moved those backlogs from newton workitems etherpad
13:03:57 <yanyan> maybe lets go through it first
13:04:08 <yanyan> perform test
13:04:17 <yanyan> I will keep working on it
13:04:32 <yanyan> adding more senlin related support into rally
13:04:49 <yanyan> and also building some sample test cases
13:05:03 <yanyan> to help user/operator to evaluate the performace of their senlin deployment
13:05:37 <yanyan> the next item is about HA, lixinhui_ Qiming, any update about it?
13:05:45 <yanyan> or new plan
13:06:01 <lixinhui_> I am working octavia
13:06:07 <lixinhui_> notification system
13:06:20 <lixinhui_> to let it send out notice of member failure
13:06:37 <lixinhui_> but still trying to get its newton version work up
13:06:43 <Qiming> octavia team agreed?
13:06:55 <lixinhui_> not discussed with them
13:07:08 <lixinhui_> I wanna try on our side firstly
13:07:14 <lixinhui_> then may propose BP
13:07:16 <Qiming> member status update is stable there?
13:07:34 <lixinhui_> what do you mean by stable
13:07:45 <yanyan> so the plan is let lb send notification for lb member status changing?
13:07:59 <lixinhui_> yes
13:08:01 <Qiming> the old story was that when member fails, octavia fails to change the 'status' field
13:08:31 <lixinhui_> once the status change, octavia should send out event
13:08:49 <lixinhui_> instead of letting others read its db
13:08:53 <yanyan> lixinhui_, I see
13:09:12 <yanyan> I also recalled that bug. have it been addressed?
13:09:24 <lixinhui_> very slow there
13:09:29 <elynn> Does it worth a BP in octavia?
13:09:36 <elynn> Or just a bug?
13:09:41 <lixinhui_> not bug
13:09:51 <lixinhui_> they have mechanism to update DB
13:09:53 <yanyan> elynn, I think the notification item is worth a bp
13:10:08 <lixinhui_> so if others care enough and they can get the right status
13:10:17 <lixinhui_> but
13:10:22 <lixinhui_> it is ugly
13:10:36 <lixinhui_> there should be some mechanism to help auto-loop
13:10:42 <elynn> yes, agree with you
13:10:45 <lixinhui_> like senlin needed
13:10:56 <Qiming> status update <-- obviously a bug; status notification <-- may need a bp
13:11:36 <lixinhui_> there seems some problem with octavianewtion
13:11:46 <lixinhui_> I cna not run it up even today
13:11:58 <lixinhui_> will try more later
13:12:06 <yanyan> even with its newton stable release?
13:12:10 <lixinhui_> yes
13:12:16 <yanyan> if so, may need some help from the team
13:12:18 <lixinhui_> some thing wrong with its image
13:12:19 <yanyan> sigh...
13:12:30 <elynn> ...
13:12:52 <lixinhui_> will connect them if I confirm nothing wrong with my env
13:13:09 <yanyan> great
13:13:45 <yanyan> and you can add the workitem about notification to etherpad
13:13:53 <yanyan> if you plan to work on it recently :)
13:14:05 <lixinhui_> Okay
13:14:09 <lixinhui_> I will
13:14:11 <yanyan> thanks
13:14:16 <lixinhui_> np
13:14:35 <yanyan> any more question about this item?
13:14:45 <yanyan> if not, will move on
13:15:02 <yanyan> ok, next one
13:15:05 <yanyan> document
13:15:13 <yanyan> hi, Qiming
13:15:31 <Qiming> no progress past weeks
13:15:33 <yanyan> I guess you didn't get time to work on it recently
13:15:38 <yanyan> I see
13:15:53 <Qiming> will work on them by adding links to official docs, in case users don't know where to find them
13:16:08 <yanyan> yes, that is very important
13:16:16 <yanyan> thanks
13:16:26 <yanyan> next one is versioned request
13:16:44 <yanyan> part of versioning everything :)
13:16:53 <Qiming> that one was just added
13:16:59 <Qiming> still working on it
13:17:20 <Qiming> for cluster create request, the request and request body have been modeled
13:17:45 <Qiming> now I'm preparing the service to serve RPC request with the new request object as parameters
13:17:58 <yanyan> great, so that will be the first one as example
13:18:19 <yanyan> hi, xuhaiwei
13:18:24 <xuhaiwei> hi yanyan
13:18:31 <xuhaiwei> sorry be late
13:18:32 <Qiming> yes, that will be the first show case
13:18:37 <yanyan> no problem
13:18:45 <yanyan> ok
13:19:12 <yanyan> hope everyone can have some time to see it
13:19:39 <yanyan> since that will be a huge change for the interface of both senlin api/engine/rpc
13:20:13 <Qiming> yup
13:20:27 <yanyan> great
13:20:32 <yanyan> next one?
13:20:42 <Qiming> its advantage is still to be exhibited
13:20:51 <yanyan> yes
13:21:13 <xuhaiwei> what is the current topic?
13:21:22 <yanyan> I think it's definitely very helpful to improve the stability and compatibility of senlin service
13:21:27 <xuhaiwei> big change to senile api/engine/rpc?
13:21:33 <yanyan> xuhaiwei, about versioned request
13:21:42 <Qiming> xuhaiwei, versioned everything
13:21:50 <yanyan> xuhaiwei, oh, sorry for my inaccurate expression
13:22:00 <yanyan> user invisible change it is
13:22:04 <Qiming> make the code stable and smooth upgradable
13:22:21 <yanyan> yes
13:22:24 <xuhaiwei> ok
13:22:31 <yanyan> the interface between different senlin modules
13:22:51 <xuhaiwei> if everything is versioned, it's easy to be upgraded?
13:22:58 <yanyan> xuhaiwei, yes
13:23:24 <Qiming> a versioned object is tasked to masking the differences
13:23:48 <Qiming> if you add a new field to a request, for example, and you only understand an old version of request
13:24:04 <Qiming> you can ask that object to make an old version representation for you
13:24:37 <Qiming> I'm learning it by doing it
13:25:14 <xuhaiwei> that is not conflicting with api micro-version?
13:25:32 <Qiming> ideally, you can have two senlin-engines working on 1.2 version, one working on 1.1 version
13:26:05 <Qiming> good question, still trying to make that 'version' into the same whole picture without further complicating things
13:26:35 <yanyan> nice
13:26:47 <yanyan> I  believe we can get better understanding on it by reading the code
13:26:52 <Qiming> when we have some api changes we want to land, we can group that into a sprint, and bump api version after that
13:27:07 <Qiming> api version is only dealing with user visible things
13:27:32 <Qiming> versioned object, ideally, can be used to version anything inside the engine, e.g. DB objects
13:29:00 <yanyan> ok, lets move on
13:29:11 <yanyan> container profile, hi, xuhaiwei, your turn
13:30:03 <xuhaiwei> recently I am working on the demo for our company cloud platform
13:30:03 <yanyan> any update or plan?
13:30:13 <xuhaiwei> no update for it
13:30:16 <yanyan> ok
13:30:25 <xuhaiwei> I can share something about the summit
13:30:32 <yanyan> I think that will be another important workitem in Ocata cycle
13:30:42 <yanyan> great, you guys can have f2f talk :)
13:31:04 <xuhaiwei> that is for tacker design summit, I have added a talk about senile and tacker integration
13:31:12 <yanyan> cool
13:31:22 <elynn> cool +1
13:31:29 <yanyan> hope to see more cooperation between senlin and tacker
13:31:40 <elynn> Would like to listen !
13:31:51 <Qiming> xuhaiwei, many thanks on bringing that to the tacker team
13:31:55 <yanyan> would like to see the video :)
13:32:15 <Qiming> I'm confident that they will appreciate the work we offloaded from them
13:32:32 <zzxwill> yanyan, you won't be at the Summit this time?
13:32:39 <yanyan> zzxwill, yes :(
13:32:41 <xuhaiwei> tacker team doesn't know senile at all, so I will introduce senile first, and then introduce tacker to use senile for vnf auto-scaling and HA
13:32:59 <yanyan> so will catch you guys up by watching the video cap
13:33:19 <yanyan> xuhaiwei, that's great
13:33:31 <xuhaiwei> design summit doesn't have vedio, yanyan
13:33:37 <yanyan> ah, right
13:33:44 <yanyan> if time is good, maybe I can call in :)
13:33:45 <Qiming> they will soon realize that "auto-scaling" could be a tale, most of the time, they will need to manipulate their VNF clusters this and that way
13:34:04 <yanyan> to join your discussion
13:34:10 <xuhaiwei> you can't go to barcelona, yanyan ?
13:34:32 <yanyan> right...
13:35:03 <xuhaiwei> Qiming, in fact to use seniln in tacker has two ways
13:35:11 <yanyan> anyway, very chance to bridge these two projects
13:35:15 <yanyan> very good chance
13:35:42 <Qiming> yes, I saw your discussion with sridahar
13:35:43 <xuhaiwei> first is make tacker talk to seniln directly, instead of talking to Heat under the current design
13:36:11 <Qiming> we'll need to find out why they need heat desparately
13:36:16 <xuhaiwei> but it may very difficult to persuade them to do in this way
13:36:30 <Qiming> if it is a valid scenario, we won't get in the way
13:37:03 <Qiming> if it is not valid, or it is something senlin can fill in easily, we will do it
13:37:57 <Qiming> mindset change is very difficult for people
13:38:22 <xuhaiwei> yes, currently in openstack community Heat is the standard orchestration tool, so people will use it without any thought
13:38:48 <yanyan> understand
13:38:57 <Qiming> heat is good, but no project is designed for all use cases ...
13:39:19 <yanyan> just for some cases, using senlin could be better
13:39:20 <yanyan> yes
13:39:43 <Qiming> so, instead of doing some abstract comparison, I'd prefer understanding their use cases
13:40:04 <yanyan> totally agree
13:40:16 <yanyan> start from a real case to understand their requirement
13:41:27 <yanyan> so we may need to spend more effort here :)
13:41:31 <yanyan> xuhaiwei, thanks for starting this
13:41:34 <xuhaiwei> if tacker support seniln heat resource well, it's a quite good thing in my opinion
13:41:38 <elynn> Maybe some day you could show us how tacker work and let us get a deeper understanding and figure out the gap :) xuhaiwei
13:41:49 <yanyan> elynn, +1
13:42:23 <elynn> xuhaiwei: yes, it's a win-win for both projects.
13:42:28 <xuhaiwei> yanyan, elynn, yes, but currently tacker can hardly do anything
13:42:39 <lixinhui_> So sridahar still be the PTL?
13:42:45 <lixinhui_> for ocata?
13:42:51 <xuhaiwei> yes, lixinhui_
13:42:55 <lixinhui_> okay
13:43:16 <xuhaiwei> but he can't join ocata summit for health problem
13:43:40 <lixinhui_> ...
13:44:01 <xuhaiwei> I hope you guys can also join the discussion in tacker design summit
13:44:18 <lixinhui_> who else from broadcade the same company with him?
13:44:18 <Qiming> it is all overlapped ...
13:44:35 <lixinhui_> when is it
13:44:36 <yanyan> all our work sessions are located in friday morining
13:44:40 <elynn> What is the time slot for your topic? xuhaiwei
13:44:45 <yanyan> https://www.openstack.org/summit/barcelona-2016/summit-schedule/#day=2016-10-28&summit_types=10
13:44:52 <yanyan> start from 9 to 12:40
13:44:59 <Qiming> most of the design summits are in the same morning
13:45:17 <xuhaiwei> oh
13:45:21 <elynn> okay...
13:45:27 <xuhaiwei> my topic is on the last day
13:45:31 <Qiming> a ridiculous design
13:45:44 <xuhaiwei> the final session
13:45:57 <yanyan> xuhaiwei, in afternoon?
13:46:01 <yanyan> or morning
13:46:12 <xuhaiwei> from 11:30 on Friday
13:46:25 <yanyan> sigh
13:46:27 <xuhaiwei> the last one in the morning
13:46:31 <yanyan> conflict with our last session
13:46:36 <yanyan> the HA one
13:47:18 <yanyan> lets have further discussion later to see whether we can get the chance to talk with tacker team
13:47:21 <yanyan> in the summit
13:47:27 <yanyan> lets move on?
13:47:36 <yanyan> we have 13 mins left
13:47:49 <yanyan> zaqar message receiver
13:47:54 <yanyan> the basic workflow is ok now
13:48:12 <yanyan> next step is keeping refactoring it to make it useful for real use case
13:48:20 <yanyan> including addressing some TODOs left
13:48:29 <yanyan> and also writing a guide
13:48:41 <Qiming> yes, guide me pls, :D
13:48:46 <yanyan> haha :)
13:48:49 <yanyan> definitely
13:48:57 <yanyan> the next one is notification
13:48:58 <yanyan> hi Qiming
13:49:25 <Qiming> no progress, pass
13:49:37 <yanyan> ok
13:49:41 <yanyan> next one is batch policy
13:49:49 <yanyan> I think Ruijie_ is working on it
13:49:53 <yanyan> hi, Ruijie_
13:50:08 <yanyan> https://review.openstack.org/#/c/376363/
13:50:24 <yanyan> here is the patch. welcome to leave comments :)
13:50:45 <Qiming> 10 minutes left, we still have 3 topics, :D
13:50:59 <yanyan> yes, need to speed up
13:51:12 <yanyan> ok, lets jump to next topic
13:51:19 <yanyan> #topic work session time slot
13:51:26 <yanyan> https://www.openstack.org/summit/barcelona-2016/summit-schedule/#day=2016-10-28&summit_types=10
13:51:27 <yanyan> here
13:51:33 <yanyan> all in the friday morning
13:52:03 <yanyan> the current sequence is, versioned policy/profile,  versioning everything, container cluster and HA
13:52:04 <Ruijie_> yes, yanyan
13:52:04 <Ruijie_> I'm working on it
13:52:16 <yanyan> Ruijie_, thanks a lot for this job :)
13:52:48 <yanyan> will leave comments offline
13:53:10 <yanyan> ok, next one?
13:53:15 <Ruijie_> no problem, yanyan :)
13:53:15 <Ruijie_> okay
13:53:20 <yanyan> :)
13:53:27 <yanyan> summit presentation preparing
13:53:48 <yanyan> we have two presentations, HA and managing cluster of 1,000 VMs
13:54:01 <yanyan> I'm now preparing the slices for second one
13:54:18 <yanyan> and will talk with eldon and CMCC colleague later about it
13:54:34 <yanyan> elynn, lixinhui_ I think you guys can start to prepare the slice for HA topic?
13:54:50 <lixinhui_> yes, yanyan
13:55:01 <yanyan> if you all free in this thursday, we can have a meetup in vmware that afternoon :)
13:55:01 <lixinhui_> we need discuss with you and Qiming
13:55:02 <elynn> I will check with lixinhui_ tomorrow.
13:55:08 <yanyan> and go through this topic together
13:55:18 <lixinhui_> great
13:55:19 <yanyan> ok :)
13:55:44 <yanyan> to fill in the content and arrange the speacking sequence
13:56:20 <yanyan> ok, lets talk more about it offline
13:56:30 <yanyan> #topic open discussion
13:56:39 <yanyan> open discussion now:)
13:56:55 <Qiming> I've added a topic
13:57:02 <Qiming> about weekly meeting schedule
13:57:14 <yanyan> oops, sorry I didn't refresh the agenda
13:57:16 <yanyan> ah, yes
13:57:17 <Qiming> yanyan and I discussed about this
13:57:21 <yanyan> that one
13:57:46 <Qiming> if the time is no good for anyone, pls speak up
13:58:10 <yanyan> yes, we can rotate between day time meeting and night time meeting(beijing time)
13:58:38 <Qiming> we can move it completely
13:58:44 <yanyan> Qiming, ok
13:58:58 <yanyan> if so, we need to find a time slot everyone is free to join
13:59:11 <yanyan> to avoid conflict with other meetings in daily work
13:59:34 <yanyan> I will send out mail to check with all you guys
13:59:54 <yanyan> ok
13:59:55 <Qiming> public vote is better
14:00:03 <yanyan> Qiming, yes, will check how to do it
14:00:17 <yanyan> and need to check the available timeslot in irc channel first
14:00:21 <yanyan> ok, time is over
14:00:32 <yanyan> lets move back to senlin channel
14:00:38 <yanyan> thanks all you guys for joining
14:00:42 <yanyan> have a good night
14:00:45 <yanyan> #endmeeting