13:01:14 <Qiming> #startmeeting senlin
13:01:15 <openstack> Meeting started Tue Sep 27 13:01:14 2016 UTC and is due to finish in 60 minutes.  The chair is Qiming. Information about MeetBot at http://wiki.debian.org/MeetBot.
13:01:16 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
13:01:18 <openstack> The meeting name has been set to 'senlin'
13:01:28 <Qiming> evening
13:01:34 <yanyanhu> o/
13:02:08 <yanyanhu> I think ethan is in vacation travel
13:02:21 <Qiming> yes, he mentioned that
13:02:47 <Qiming> let's see if anyone else is joining
13:02:51 <yanyanhu> ok
13:02:57 <yanyanhu> I see xinhui in senlin channel
13:03:04 <lixinhui_> hi
13:03:18 <yanyanhu> hi, xinhui :)
13:03:28 <lixinhui_> congratulations, Yanyan
13:03:36 <yanyanhu> :)
13:03:43 <lixinhui_> :)
13:04:00 <lixinhui_> and happy for Qiming
13:04:12 <Qiming> for what?
13:04:14 <yanyanhu> just try to take over some work of Qiming :)
13:04:29 <lixinhui_> to find good successor
13:04:43 <lixinhui_> and release more time for baby
13:04:47 <Qiming> everyone deserves an opportunity to learn, to practice
13:04:53 <yanyanhu> release more time for more important work, haha
13:05:01 <lixinhui_> yes, haha
13:05:02 <Qiming> let's get started?
13:05:04 <yanyanhu> or more chance
13:05:09 <yanyanhu> yes
13:05:12 <yanyanhu> lets start
13:05:12 <lixinhui_> yes
13:05:18 <Qiming> could be a short one
13:05:24 <yanyanhu> yep
13:05:27 <Qiming> #link https://etherpad.openstack.org/p/senlin-newton-workitems
13:05:39 <Qiming> newton work items
13:05:56 <yanyanhu> no progress in performance test I think
13:06:07 <Qiming> okay
13:06:16 <yanyanhu> I feel further work on rally test could be started in next cycle
13:06:18 <Qiming> no update on health management either
13:06:30 <lixinhui_> no, sorry
13:06:38 <Qiming> yanyanhu, that is fine
13:07:04 <Qiming> ruijie started a poc of batching policy, you may want to review it
13:07:21 <yanyanhu> sure, will check the patch
13:07:27 <Qiming> https://review.openstack.org/#/c/376363/
13:07:36 <yanyanhu> I think ruijie did some excellent work :)
13:07:45 <Qiming> it looks not bad
13:07:45 <yanyanhu> he got familiar with senlin very fast
13:08:01 <Qiming> agreed
13:08:18 <Qiming> documentation side
13:08:35 <Qiming> I will check the 'cloud_backend' thing
13:08:49 <yanyanhu> ok, that is an important item
13:08:59 <Qiming> as for wiki page, I'll add some pointers to our developer/user documentation
13:09:09 <yanyanhu> ok
13:09:17 <Qiming> we don't need to copy/paste all the docs everywhere
13:09:28 <yanyanhu> sure
13:09:37 <Qiming> and there have been some disputes about whether wiki should be avoided
13:10:05 <Qiming> the main reason is that it is difficult to shield it from spams
13:10:14 <yanyanhu> that's true...
13:10:21 <Qiming> keeping it up to date is also a burden
13:10:44 <Qiming> so ... I'd prefer leave some links there
13:10:50 <yanyanhu> but I feel wiki is still the first place that most people get information about a project
13:10:51 <lixinhui_> why we always spend time on document
13:10:55 <yanyanhu> yes
13:11:12 <lixinhui_> is not good if users find IRC to ask questions
13:11:21 <lixinhui_> instead of read wiki or document
13:11:30 <Qiming> for most users, docs are the starting point
13:11:43 <lixinhui_> we believe so
13:11:58 <lixinhui_> but seems most content already there
13:12:14 <Qiming> especially for the other semisphere, most of the developers are offline when they are trying to ask questions
13:12:41 <yanyanhu> agree. At least for myself, I will be very happy if there is a document has step-by-step guide when I learn something new
13:12:42 <Qiming> yes, so ... just some pointers would suffice, I won't copy/paste things that may get outdated
13:12:44 <lixinhui_> cool, where you learned this word semisphere...
13:13:00 <Qiming> bible
13:13:06 <lixinhui_> :)
13:13:15 <Qiming> next, container profile
13:13:33 <Qiming> there is a patch dealing with dependencies
13:14:22 <Qiming> after some rethinking on this, I have left some comments there: https://review.openstack.org/#/c/368539/
13:15:18 <Qiming> the dependency between one cluster/node and another cluster/node is not that straightforward
13:15:33 <yanyanhu> so it's about the handling for dependency between container cluster and VM node?
13:15:36 <Qiming> and the current practice is to record the actual dependency
13:15:45 <Qiming> yes
13:16:13 <Qiming> that dependency is getting updated each time a dependent cluster/node is created/deleted
13:16:40 <Qiming> I was trying to revisit the issue from a different angle
13:17:17 <Qiming> even if a container cluster has 0 nodes, and it has specified host_cluster to be another VM cluster
13:17:26 <Qiming> the VM cluster cannot be removed
13:17:51 <yanyanhu> Qiming, yes, since the container cluster could be scaled out again later
13:18:02 <Qiming> because when you are adding nodes to the container cluster, you will find no hosting cluster/node for it
13:18:25 <Qiming> exactly
13:18:59 <Qiming> hope that comment can help haiwei to handle that dependency thing
13:19:11 <Qiming> Receiver
13:19:22 <yanyanhu> https://review.openstack.org/#/c/373004/
13:19:28 <yanyanhu> this patch is for notification handling
13:19:35 <yanyanhu> which closes the loop
13:19:52 <yanyanhu> the basic workflow now works as expected
13:20:30 <yanyanhu> also have started to work on the doc: https://review.openstack.org/377330
13:20:30 <Qiming> okay, will jump onto that tomorrow
13:20:42 <yanyanhu> Qiming, great, looking forward to your comment
13:20:48 <Qiming> great, definitely I'll need a doc to test it
13:21:00 <yanyanhu> really hope can finish it before final release of newton cycle
13:21:24 <Qiming> TBH, I never got zaqar up and running
13:21:36 <yanyanhu> then we have a workable(althoug basic) message type of receiver
13:21:42 <Qiming> okay, remind me if I forget it tomorrow
13:21:55 <yanyanhu> Qiming, I will work with you together to test it tomorrow :)
13:22:05 <Qiming> great
13:22:14 <Qiming> event/notification
13:22:31 <Qiming> obviously some work items for Ocata
13:22:36 <yanyanhu> yes
13:22:57 <Qiming> but I have already started playing with versioned requests
13:23:06 <yanyanhu> cool
13:23:40 <Qiming> I hope I can get one workflow landed as a show case
13:23:50 <yanyanhu> ok.
13:24:06 <Qiming> it will be above wraping api request params into a versioned object
13:24:39 <Qiming> then our engine interface will be stabilized for quite a long time, while our api interface can evolve in a controlled manner
13:25:05 <yanyanhu> great
13:25:23 <yanyanhu> that is part of 'versioning everything' :)
13:25:31 <Qiming> next time when you are adding/removing a paramter for a cluster_create call, for example, you don't have to change the rpc client
13:25:53 <lixinhui_> cool
13:25:54 <Qiming> yes, versioned notification is another one
13:26:38 <Qiming> the json schema conversion from a versioned object is easy
13:26:58 <Qiming> and the converted json schema can be used to validate api parameters
13:27:29 <Qiming> e.g. https://review.openstack.org/#/c/377266/1/senlin/objects/fields.py
13:27:42 <yanyanhu> I see
13:27:44 <Qiming> this one is about the 'name' field used in many requests
13:28:14 <Qiming> by referencing this field, we can check the length and contents of the name value
13:28:40 <Qiming> it can be treated as a smarter data type
13:29:08 <Qiming> so the ad-hoc param validation at api layer can be completely dropped
13:29:34 <yanyanhu> hope that work can be done in next cycle
13:30:06 <Qiming> yes, some basics can be started today, because we are not changing any workflow yet
13:31:01 <Qiming> after newton is released, we can start massging workflow
13:31:11 <yanyanhu> ok
13:31:19 <yanyanhu> the finally release is oct. 9
13:31:40 <yanyanhu> and 29 this month for last rc?
13:31:40 <Qiming> #topic high priority bug fixes for last rc
13:31:56 <Qiming> https://releases.openstack.org/newton/schedule.html
13:32:18 <Qiming> this is our rc2 week
13:33:02 <Qiming> open reviews: https://review.openstack.org/#/q/project:openstack/senlin+status:open
13:33:20 <yanyanhu> I think there is no critical ones left?
13:34:09 <Qiming> this one? https://review.openstack.org/371922
13:34:39 <yanyanhu> ah, yes, that one is important
13:34:53 <Qiming> and this one is okay, imo, https://review.openstack.org/372061
13:35:49 <Qiming> other than those ones, I don't think we have high priority patches to merge
13:36:09 <yanyanhu> will check 372061 soon
13:36:18 <yanyanhu> it's interesting
13:37:15 <Qiming> the logic is simple
13:37:19 <yanyanhu> yes
13:37:36 <Qiming> the project_safe should be checked even if the requesting user is admin
13:37:55 <yanyanhu> the checking logic of is_admin and project_safe are different and should be done in different layers
13:38:02 <yanyanhu> yep
13:38:03 <Qiming> in other words, an admin user may want to check things from a single project as well
13:38:41 <Qiming> right
13:39:00 <Qiming> just checked bug list, not critical bugs either
13:39:03 <yanyanhu> it's a historical issue I think
13:39:11 <Qiming> yes
13:39:19 <Qiming> #topic summit prep
13:39:19 <yanyanhu> yes, didn't see critical bug left
13:39:36 <Qiming> already talked to eldon?
13:39:39 <yanyanhu> just started an etherpad to collect the basic idea
13:39:41 <yanyanhu> https://etherpad.openstack.org/p/barcelona-summit-senlin-topic-cluster-1000-vms
13:39:45 <yanyanhu> yes, have told him
13:39:54 <Qiming> the outline looks good to me
13:39:57 <yanyanhu> he hasn't got time to see it I think
13:40:12 <Qiming> need to work on it now
13:40:12 <yanyanhu> need some concrete result and data from them I feel
13:40:21 <Qiming> true
13:40:25 <yanyanhu> yes, only about 2~3 weeks left
13:40:43 <Qiming> let's see if we can grab him online this week
13:40:50 <yanyanhu> ok
13:41:08 <Qiming> #topic open discussion
13:41:13 <Qiming> anything else?
13:41:21 <yanyanhu> nope from me
13:41:51 <Qiming> next cycle I want to focus more on container clustering
13:42:24 <yanyanhu> yes, that's an important workitem
13:42:45 <Qiming> it is an intersection between company agenda and community work
13:43:00 <yanyanhu> right, may need more thinking here
13:43:05 <yanyanhu> especially about the use case
13:43:19 <Qiming> use case is not a problem
13:43:26 <yanyanhu> how to make the design consistent with the real demand from user
13:43:27 <Qiming> I can see a lot out there
13:43:56 <Qiming> yep, users can be end users or operators
13:44:06 <yanyanhu> yes
13:44:24 <Qiming> need to be a little bit careful about the differences
13:44:37 <yanyanhu> I see
13:44:46 <Qiming> try something useful and get it deployed
13:44:53 <lixinhui_> by company agenda, could you give more information
13:45:19 <Qiming> then move forward based on that
13:45:23 <lixinhui_> share more strategy or customer
13:45:33 <Qiming> lixinhui_, not yet
13:45:57 <Qiming> you will know it when they want you know
13:45:58 <lixinhui_> okay
13:46:11 <lixinhui_> who are they
13:46:25 <yanyanhu> they are big man LOL
13:46:32 <lixinhui_> oh
13:46:32 <yanyanhu> just kidding
13:46:46 <lixinhui_> sorry for my understanding capability these days
13:47:06 <lixinhui_> busy and English downgrade very quickly
13:47:19 <yanyanhu> lol
13:47:23 <yanyanhu> :)
13:47:27 <Qiming> if you need some additional nutrition, you can call me any time
13:47:33 <Qiming> I can watch you eat
13:47:39 <yanyanhu> haha
13:47:42 <lixinhui_> Thank you, Qiming...
13:47:49 <yanyanhu> don't steal food from baby
13:47:52 <Qiming> u r welcome
13:48:00 <lixinhui_> :)
13:48:13 <Qiming> guess that's all for today?
13:48:29 <lixinhui_> no from me
13:48:33 <yanyanhu> yes, think so
13:48:33 <Qiming> thanks for you time
13:48:42 <yanyanhu> thanks, have a good night
13:48:49 <lixinhui_> have a good night
13:48:52 <Qiming> bye, and
13:48:53 <Qiming> #endmeeting