13:00:49 <Qiming> #startmeeting senlin
13:00:50 <openstack> Meeting started Tue Dec  8 13:00:49 2015 UTC and is due to finish in 60 minutes.  The chair is Qiming. Information about MeetBot at http://wiki.debian.org/MeetBot.
13:00:51 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
13:00:54 <openstack> The meeting name has been set to 'senlin'
13:01:09 <Qiming> good evening/morning/...
13:01:16 <haiwei> hi
13:01:25 <Yanyanhu> hello
13:01:26 <elynn__> o/
13:01:31 <lixinhui> hi
13:01:53 <Qiming> please check agenda and see if you have things to add
13:02:02 <Qiming> https://wiki.openstack.org/wiki/Meetings/SenlinAgenda
13:02:35 <Qiming> as usual, lets start with the etherpad
13:02:44 <Qiming> #topic Mitaka work items
13:02:51 <Qiming> #link https://etherpad.openstack.org/p/senlin-mitaka-workitems
13:03:08 <Qiming> Resource type -- elynn
13:03:53 <elynn__> hi
13:03:56 <Qiming> previously, it was delayed by the api change
13:04:15 <elynn__> no progress this week, api change is done?
13:04:17 <Qiming> I think the API modification is drawing near its end
13:04:38 <Qiming> we will do some modifications next to the senlinclient
13:04:50 <elynn__> good to know, I will update these patches this week
13:05:09 <Qiming> okay, the latest patch to sdk is about profile type
13:05:14 <elynn__> I would like to add a template to heat-templates, what do you think?
13:05:14 <Qiming> ref: https://review.openstack.org/#/c/253817/
13:05:31 <Qiming> elynn__, we can do that later
13:05:46 <Qiming> first we need to make sure things work stable
13:06:04 <elynn__> Qiming: you mean after other resources added to heat?
13:06:14 <Qiming> yes, elynn__
13:06:20 <elynn__> like policy or webhook
13:06:34 <Qiming> for example, the update operation is not fully supported I think
13:06:45 <Qiming> policy and webhook are midterm goals
13:07:00 <elynn__> ok, I will focus on cluster and node resources now.
13:07:01 <Qiming> will need specs on that before working on the code
13:07:11 <Qiming> yes, focus on the basics please
13:07:33 <elynn__> Qiming: yes, I haven't fully tests update now, I might wait for senlin to get stable.
13:07:49 <Qiming> the last missing resource type in sdk is policy-type
13:07:56 <Qiming> that won't be a big deal
13:08:08 <Qiming> after that senlinclient can be pretty simplified
13:08:31 <Qiming> elynn__, please help get it stable instead of waiting if you have cycles
13:08:44 <elynn__> Qiming: ok , I can start to test the codes now.
13:08:48 <Qiming> we need at least CRUD working properly
13:09:02 <Qiming> great
13:09:08 <Qiming> #topic senlinclient unit test
13:09:19 <elynn__> Qiming: yes, I will do some test on senlin.
13:09:25 <haiwei> I didn't do it this week
13:09:28 <haiwei> the unit test
13:09:28 <Qiming> https://review.openstack.org/251798
13:09:35 <Qiming> this one was reviewed
13:09:39 <haiwei> focused on the api side
13:09:49 <Qiming> okay
13:10:07 <haiwei> honestly I don't know why jenkins failed
13:10:08 <Qiming> please check if changing 'MagicMock' to 'Mock' makes sense to you
13:10:42 <haiwei> every single test case can be passed but when running all the tests some of them failed
13:11:00 <haiwei> replacement of MagicMock is Ok to me
13:11:07 <Qiming> alright, when help is needed, pls let the team know, :) we can share tears
13:11:28 <Qiming> okay, another concurrency problem
13:11:28 <haiwei> I will update the patch soon
13:11:37 <Qiming> good luck
13:11:47 <haiwei> yes
13:11:47 <Yanyanhu> :)
13:11:58 <Qiming> #topic api modification
13:12:23 <haiwei> you and me, right?
13:12:25 <Qiming> 201/202 response code is already done, right?
13:12:51 <haiwei> I uploaded two patches in the afternoon
13:12:53 <Qiming> current focus is to make sure we are returning a 'location' header when appropriate
13:13:06 <Qiming> yep, saw that, progress is good
13:13:22 <haiwei> most of the APIs have been done, I think
13:13:38 <haiwei> I will check if anything is missing
13:14:01 <haiwei> 201/202 are done
13:14:12 <Qiming> okay, when you are checking it again, please cross verify it with the docs
13:14:19 <elynn__> Is there any change for senlinclient?
13:14:24 <Qiming> #link https://review.openstack.org/#/c/252245/
13:14:31 <haiwei> ok, I will check the docs
13:14:41 <Qiming> this is the doc patch, finally got some commens from cores
13:14:56 <haiwei> the long doc
13:14:59 <Qiming> the PDF version is easier to read: http://docs-draft.openstack.org/45/252245/10/check/gate-api-site-tox-doc-publish-checkbuild/d1f26a1//publish-docs/api-ref-guides/bk-api-ref-clustering-v1.pdf
13:15:35 <Qiming> the HTML version is more interation oriented: http://docs-draft.openstack.org/45/252245/10/check/gate-api-site-tox-doc-publish-checkbuild/d1f26a1//publish-docs/api-ref/api-ref-clustering-v1.html
13:16:06 <Qiming> 400/404 rework is completed
13:16:20 <Qiming> release notes commited as well
13:16:46 <Qiming> #topic health management
13:17:04 <Qiming> I've being thinking about it
13:17:09 <Qiming> need another brain
13:17:42 <Qiming> lixinhui, want to work on this?
13:17:52 <lixinhui> Sure
13:18:14 <Qiming> I was told that your brain quality is good, :)
13:18:30 <haiwei> can you explain the basic goal of health policy
13:18:34 <lixinhui> really?
13:18:40 <haiwei> health management
13:18:41 <Yanyanhu> :)
13:19:08 <zhangguoqing> explain please :)
13:19:09 <Qiming> network problem again
13:19:37 <Qiming> i heard something, then I was disconnected
13:20:02 <lixinhui> Powerful ears :)
13:20:15 <Qiming> lixinhui, let's start with some use cases
13:20:25 <haiwei> seems someone want Senlin to control like : if a vm is dead senlin will start a new one
13:20:28 <Qiming> we are not about to solve all problems
13:20:56 <Qiming> yes, haiwei it sounds pretty simple a scenario
13:21:04 <Qiming> but it is not that straightforward
13:21:06 <lixinhui> yes, whole picture is so complicated
13:21:09 <haiwei> but it seems not easy
13:21:22 <haiwei> maybe need a new command for it?
13:21:36 <Qiming> yes, there are at least three factors for consideration:
13:21:50 <Qiming> 1. failure detection; 2. failure notification; 3. failure recovery
13:22:42 <Qiming> senlin will have to do recovery, either manually or automatically
13:23:00 <Qiming> because it is MANAGER of the groups
13:23:17 <haiwei> seems there is a long way to go
13:23:23 <Qiming> senlin will provide receivers for notifications
13:24:14 <Qiming> it should allow recovery to be triggered by any external alert/alarm/event/monitoring systems
13:25:14 <Qiming> finally, we will look into some naive (builtin) detection mechanisms
13:25:14 <Qiming> there is no easy answer to any of these three questions
13:25:24 <Yanyanhu> is that possible the failure detection and notification is from senlin service inside?
13:25:59 <haiwei> off line again?
13:26:01 <Yanyanhu> ...my question make him disconnected again?
13:26:08 <haiwei> haha
13:26:12 <lixinhui> haha
13:26:49 <Qiming> sigh ... I was out again
13:26:57 <Qiming> did I miss something?
13:27:04 <Yanyanhu> nope
13:27:11 <lixinhui> :)
13:27:22 <Qiming> hexchat is very unstable
13:27:23 <haiwei> <Yanyanhu> is that possible the failure detection and notification is from senlin service inside?
13:27:32 <Yanyanhu> so the last question is
13:27:45 <Qiming> I'd say yes
13:28:03 <Yanyanhu> I also think so
13:28:26 <Qiming> but reliable failure detection is a domain by itself
13:28:40 <Qiming> I'm using mIRC now
13:28:50 <Qiming> just bought a license
13:29:06 <Qiming> okay, moving on?
13:29:14 <Yanyanhu> yes
13:29:15 <Qiming> or there are questions?
13:29:16 <elynn__> ok
13:29:24 <haiwei> no
13:29:29 <Qiming> #topic update operation
13:29:45 <Yanyanhu> actually didn't have time to work on it this week...
13:29:53 <Qiming> I think yanyan was completely trapped by the DB concurrency problem
13:29:59 <Yanyanhu> yea...
13:30:17 <Qiming> finally, functional tests are passing now
13:30:26 <Yanyanhu> I'm a little worried there could be no perfect solution for this issue...
13:30:28 <Qiming> except for the webhook one
13:30:30 <haiwei> saw the patch , but can't help
13:30:52 <Qiming> I spent some time reading sqlalchemy docs as well
13:31:03 <Qiming> maybe this is the only workaround we have today
13:31:15 <Yanyanhu> yes, I also have the same feeling...
13:31:29 <Yanyanhu> maybe we have to make a tradeoff here
13:31:39 <Qiming> listen to the developers, they may cheat you, but they are the most honest people you are talking to, :)
13:31:50 <Yanyanhu> yes
13:31:52 <Qiming> so we will follow the docs
13:32:14 <Yanyanhu> and I found some most useful info sometimes came from stackoverflow :)
13:32:17 <Qiming> okay, moving on
13:32:24 <Qiming> #topic Receiver
13:32:40 <Qiming> I don't think Julio will have cycles in the coming weeks
13:32:46 <Qiming> so I'm gonna take over that
13:33:06 <Yanyanhu> ok
13:33:14 <Qiming> #topic lock breaker
13:33:35 <elynn__> I will
13:33:49 <elynn__> find a way to add related tests this week
13:33:50 <Qiming> haven't do some real tests on the lock breaker thing
13:34:10 <Qiming> cool, thanks, god
13:34:27 <Qiming> #topic parser test case
13:34:30 <Yanyanhu> elynn__, great
13:34:38 <lixinhui> cool
13:34:47 <Qiming> em ... anyone want to claim this?
13:35:05 <elynn__> I can do that
13:35:16 <Qiming> thanks, god^2
13:35:38 <elynn__> ;)
13:35:53 <Qiming> #topic functional tests
13:36:16 <Yanyanhu> I think after DB sync issue is resolved, we can have a clean basement for new coming test cases
13:36:20 <Qiming> we are almost there?
13:36:25 <Yanyanhu> especially for those ones about lock
13:36:27 <Yanyanhu> yes
13:36:49 <Yanyanhu> hope can finish it this week
13:36:56 <Qiming> it is rebased on the webhook patch?
13:37:07 <Yanyanhu> yes, I rebased it
13:37:11 <Qiming> cool
13:37:27 <elynn__> good to hear!
13:37:34 <Yanyanhu> lets wait for the result :)
13:37:45 <Qiming> I was thinking maybe not just mark_succeed needs this, other similar functions will needed the fix as well
13:37:49 <Qiming> for completeness
13:38:19 <Qiming> #topic bug clearance
13:38:46 <Qiming> I did spend some time on reviewing the bugs, but ... I need to spend more
13:39:04 <Qiming> manually mark them as 'fix released'
13:39:09 <Yanyanhu> it's a hard work...
13:39:18 <Qiming> this will be done automatically in future, fortunately
13:39:20 <lixinhui> robot model
13:39:26 <haiwei> there are many ones which are open?
13:39:42 <Yanyanhu> Qiming,  http://lists.openstack.org/pipermail/openstack-dev/2015-December/081612.html this mail link you shared
13:39:56 <Qiming> haiwei, you got to do an 'advanced' search, to find the open ones
13:40:13 <Qiming> currently, 'fix commited' ones are still not treated as closed
13:41:04 <Qiming> let's take a look at the blueprints?
13:41:08 <Qiming> #link https://blueprints.launchpad.net/senlin/+specs?show=all
13:41:13 <Yanyanhu> ok
13:41:49 <haiwei> some are started but not approved
13:41:54 <Yanyanhu> most backlogs have been marked as implemented
13:42:48 <Qiming> em ...
13:43:01 <Qiming> lock breaker patches were not linking back the the bp
13:43:36 <elynn__> Seems most of my patch didn't link back to launchpad
13:43:36 <Yanyanhu> I can see three patches there
13:44:15 <elynn__> Hmm, This BP have all my patches links.
13:44:40 <Yanyanhu> this one? https://blueprints.launchpad.net/senlin/+spec/lock-breaker
13:44:41 <Qiming> okay, maybe because it was not approved?
13:45:23 <haiwei> I don't think so Qiming
13:46:40 <Qiming> okay, maybe this has to be manually reviewed
13:47:16 <Qiming> #topic open discussions
13:48:35 <haiwei> elynn__ has submitted a patch https://review.openstack.org/#/c/254254/4
13:48:52 <Qiming> anything folks want to discuss?
13:49:07 <elynn__> ah, this one
13:49:40 <haiwei> I am a little confused about the original implement
13:49:41 <Yanyanhu> quick question about action API rework. So we still have action_id returned back in resp body right?
13:49:55 <elynn__> haiwei: You mean after policy/profile_create, we should retrieve policy again ?
13:50:08 <Qiming> Yanyanhu, it was not mentioned anywhere
13:50:15 <haiwei> Yanyanhu, I think so, client side is using it
13:50:25 <Yanyanhu> yes, just check the code and feel so
13:50:38 <Qiming> I was thinking about the same question
13:51:01 <Qiming> however, no one is preventing us from doing so
13:51:21 <haiwei> elynn__ I just don't know why we should both support policy_id and policy
13:51:48 <haiwei> Qiming, what question?
13:52:20 <Qiming> returning action id in body
13:52:31 <haiwei> elynn__  since only one way is in effect
13:52:48 <elynn__> Good question, the codes are already there before I modify it... If use policy, we can save a call to get resource back.
13:52:59 <Qiming> whether we should do that, my feeling is that it is okay, no one said 202 cannot return a body
13:53:27 <haiwei> for update, we do return a body
13:53:47 <elynn__> haiwei: It's ok to remove policy support and always retrieve resource after creation.
13:54:14 <haiwei> I feel the same way, elynn__
13:54:27 <Qiming> I was lost, seems the patch is about profiles, and you are talking about policies
13:54:59 <Yanyanhu> :)
13:55:05 <elynn__> Qiming: It's a bug report by you
13:55:23 <haiwei> policy should be done exactly as profile, but it is not
13:55:38 <haiwei> I don't know why
13:55:38 <Yanyanhu> Qiming, about the action_id. I agree to keep it in resp body
13:55:40 <Qiming> ah, that title should be changed
13:55:49 <Qiming> I mean the commit message
13:56:23 <Qiming> it is talking about both profile and policy, but the patch only fixes profile, not policy, that is the issue you are talking about?
13:56:39 <elynn__> I just make them behave the same.
13:56:54 <elynn__> this patch touches both of them.
13:57:10 <elynn__> Qiming: It also fixes policy
13:57:25 <haiwei> actually no, I don't know both policy_id and policy are both in the parameters Qiming
13:57:31 <elynn__> Qiming: https://review.openstack.org/#/c/254254/4/senlinclient/v1/client.py L90
13:58:11 <haiwei> I am lost too when reviewing this patch for the first time
13:58:42 <elynn__> heihei, Seems my commit message mislead you...
13:58:51 <Qiming> em... need some time to understand the 'extra_attrs' parameter
13:59:28 <elynn__> If extra_attrs is True ,will use response body to update resource attributes
13:59:52 <Qiming> sdk code says so?
13:59:53 <elynn__> like created_time, so that we can display them correctly.
13:59:59 <elynn__> No...
14:00:16 <Qiming> okay, let's continue in #senlin
14:00:20 <haiwei> time is up
14:00:20 <Qiming> #endmeeting