13:00:03 <Qiming> #startmeeting senlin
13:00:04 <openstack> Meeting started Tue Apr 25 13:00:03 2017 UTC and is due to finish in 60 minutes.  The chair is Qiming. Information about MeetBot at http://wiki.debian.org/MeetBot.
13:00:05 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
13:00:07 <openstack> The meeting name has been set to 'senlin'
13:00:38 <yanyanhu_> hi, evening
13:00:51 <Qiming> hi
13:01:34 <Qiming> don't know if others are joining
13:01:55 <yanyanhu_> maybe lets wait for a while for them?
13:02:00 <Qiming> ya
13:02:06 <ruijie> evening :)
13:03:04 <yanyanhu_> hi, ruijie :)
13:03:11 <Qiming> evening
13:03:14 <Qiming> let's get started
13:03:20 <ruijie> hi :)
13:03:29 <Qiming> #topic pike work items
13:03:39 <Qiming> evening ruijie
13:03:55 <Qiming> don't think we have elynn today
13:04:06 <Qiming> so we skip vdu topic
13:04:06 <elynn> yes, you have
13:04:10 <ruijie> :)
13:04:22 <elynn> just no update on this topic
13:04:23 <Qiming> engine improvement, CLUSTER_CHECK patch looks good to me
13:04:35 <Qiming> evening elynn
13:05:04 <Qiming> okay, hope we can have it by the end of this cycle, :)
13:05:35 <Qiming> oh, I was wrong, XueFeng's patch needs some work
13:06:00 <Qiming> node adopt patches are merged now
13:07:02 <Qiming> need to work on api support, doc and test
13:07:41 <Qiming> we will need to bump microversion of api soon
13:08:31 <Qiming> in my mind, next api microversion will include node-adopt, node-adopt-preview, service-list and some changes to cluster actions that have health_check support
13:09:25 <Qiming> the health_check parameter for scale in, patch 459090 looks good to me
13:09:30 <Qiming> please review
13:09:55 <ruijie> but for policy side, that is not easy to implement with current design
13:10:45 <ruijie> Qiming, I have a question here, when check the deletion policy
13:11:15 <ruijie> I found the data from region&zone policy only been used in deletion policy?
13:11:29 <Qiming> so far, yes
13:12:42 <XueFengLiu> Qiming, a question about cluster_check, if the node is shutoff we think it was error
13:12:50 <Qiming> yes
13:12:52 <ruijie> okay .. need to think how to interact with deletion policy
13:13:33 <XueFengLiu> then we do_recover, the default action is  delete and create.
13:13:59 <XueFengLiu> In fact we only need to start the node
13:15:20 <Qiming> sorry, network connection not stable
13:15:33 <Qiming> all nodes that are not operational should be treated unhealthy
13:16:35 <Qiming> and, ideally, we should be able to recover them based on their current status (deleted -> recreate, crashed -> rebuild, shutdown -> start, etc.)
13:17:04 <Qiming> that makes sense?
13:17:13 <XueFengLiu> yes:)
13:17:19 <XueFengLiu> think so
13:17:28 <Qiming> okay
13:17:53 <Qiming> next work item (just added), is about optimizing the runtime_data properties
13:18:28 <Qiming> they were designed to be a cache of DB entries, however, we are seeing some obvious overhead in using them
13:18:42 <Qiming> I'm working on some patches to optimize it
13:18:59 <XueFengLiu> ok, great
13:19:05 <ruijie> Qiming, you mean dumping data to action.outputs?
13:19:29 <Qiming> for example, a cluster's rt['nodes'] now contains only OVO version of Nodes, rather than engine Node objects
13:19:57 <ruijie> oh, okay
13:19:58 <elynn> Will you back port these patches back to coat?
13:20:03 <elynn> ocata
13:20:04 <Qiming> in other words, we will have array of senlin.objects.node.Node instead of senlin.engine.node.Node
13:20:17 <Qiming> it would be a little bit difficult
13:20:30 <Qiming> if there is a requirement, elynn, I can do that
13:20:54 <Qiming> I can try cherry-pick though
13:22:11 <Qiming> okay for you, elynn?
13:22:34 <Qiming> moving on
13:22:39 <Qiming> rdo progress?
13:23:05 <XueFengLiu> no update
13:23:12 <XueFengLiu> Qiming
13:23:26 <Qiming> health management
13:23:45 <Qiming> we have basic workflow (mistral) support added to Node class now
13:23:56 <lxinhui> I am drafting chart for Boston
13:24:00 <lxinhui> about workflow
13:24:04 <lxinhui> and hope
13:24:12 <Qiming> need to revise recover operation to invoke them
13:24:22 <lxinhui> renat could provide 3-5 pages for us
13:24:33 <Qiming> oh, godess is here, below the water surface
13:24:40 <Qiming> cool
13:24:41 <lxinhui> ...
13:24:55 <lxinhui> sorry, just back home from CMCC lab
13:25:04 <lxinhui> we are doing R2 integration these days
13:25:07 <lxinhui> ...
13:25:13 <Qiming> I see.
13:25:24 <lxinhui> for the VDU
13:25:33 <lxinhui> health management
13:25:54 <lxinhui> is more complex than what we think before
13:26:08 <lxinhui> actually tocsa has add extention spec on this part
13:26:30 <lxinhui> such as relationship bound with a VDU
13:26:32 <Qiming> alright, any link?
13:26:46 <lxinhui> I will send out doc to all of you
13:27:15 <Qiming> great
13:27:19 <elynn> thanks, we need to discuss about it after we read the docs.
13:27:21 <lxinhui> The documents are latest VDU spec
13:27:30 <lxinhui> sure
13:27:43 <Qiming> okay, looking forward to that
13:27:57 <Qiming> Testing
13:28:20 <Qiming> local tempest testing fix is done, XueFeng?
13:28:29 <XueFengLiu> yes
13:28:31 <XueFengLiu> Qiming
13:29:34 <Qiming> rally job gets problem?
13:29:49 <yanyanhu_> yes, seems so
13:29:55 <Qiming> don't understand why admin users cannot get profile object created by non-admin
13:30:03 <yanyanhu_> should be caused by this patch: https://review.openstack.org/415690
13:30:05 <XueFengLiu> yes, somebody mail me and I have discuss with yanyanhu_
13:30:11 <XueFengLiu> discussed
13:30:26 <yanyanhu_> it enforces the project_safe for admin user
13:30:35 <Qiming> right
13:30:56 <Qiming> even an admin should have a role in a project in order to operate things there
13:31:02 <yanyanhu_> so I guess we may need to find a way to address this kind of requirement
13:31:45 <Qiming> the admin user, should login using the proper OS_PROJECT_ID
13:32:02 <XueFengLiu> http://eavesdrop.openstack.org/irclogs/%23senlin/%23senlin.2017-04-25.log.html
13:32:17 <yanyanhu_> Qiming, I guess the project_id of an user is fixed once the user is created
13:32:29 <Qiming> no, that is the default project id
13:32:48 <yanyanhu_> yes, but using other project_id rather than default one won't work I think...
13:33:04 <Qiming> when you authenticate to keystone, you can still specify the target project id
13:33:08 <yanyanhu_> at least it didn't last time I tried it...
13:33:13 <Qiming> you have to have a role in that project
13:33:14 <yanyanhu_> a while ago
13:33:41 <yanyanhu_> Qiming, yes... I tried it like that, but got authentication error... not sure whether it works now
13:33:53 <Qiming> it should work
13:34:07 <yanyanhu_> oh, I tried it for trust
13:34:14 <Qiming> we only use project-scope tokens
13:34:16 <yanyanhu_> maybe that is a special case
13:34:20 <Qiming> right
13:34:26 <Qiming> trust is different
13:34:30 <yanyanhu_> for normal case, it should work
13:34:44 <Qiming> and maybe it is the trust that caused the problem
13:35:03 <yanyanhu_> yes, it could be
13:35:14 <Qiming> I'd suggest they dump the context contents and see if there are roles we can check
13:35:33 <XueFengLiu> ok
13:35:57 <Qiming> client functional test, no update?
13:36:09 <XueFengLiu> yes:(
13:36:49 <Qiming> moving on
13:37:00 <Qiming> last is policy improvement
13:37:04 <Qiming> ruijie any update?
13:37:08 <ruijie> no progress
13:37:22 <yanyanhu_> Qiming, no, I guess
13:37:47 <Qiming> okay
13:38:02 <Qiming> anything else wrt pike work items?
13:38:22 <Qiming> guess no?
13:38:48 <yanyanhu> Qiming, no I guess
13:38:57 <XueFengLiu> Yes
13:38:57 <yanyanhu> sorry just dropped
13:39:00 <Qiming> alright
13:39:06 <Qiming> #topic boston summit prep
13:39:41 <Qiming> anyone needs other's help?
13:39:53 <XueFengLiu> Qiming, one problem
13:40:20 <XueFengLiu> create are alarm, and make the vm high
13:40:35 <Qiming> vm refused to go high? :D
13:40:47 <yanyanhu> :P
13:40:49 <XueFengLiu> no
13:41:01 <XueFengLiu> can't get the alarm
13:41:35 <Qiming> could be another ceilometer/aodh problem
13:41:45 <XueFengLiu> maybe
13:41:55 <XueFengLiu> reference to the doc you write
13:42:15 <Qiming> then better debug aodh/ceilometer first
13:42:23 <XueFengLiu> ok
13:42:26 <Qiming> they change things every single day
13:42:36 <XueFengLiu> en
13:43:39 <Qiming> #topic open discussion
13:43:53 <Qiming> anything else people want to discuss?
13:44:16 <yanyanhu> no more from me
13:44:22 <yanyanhu> just dropped again...
13:44:28 <ruijie> no from me :)
13:44:48 <XueFengLiu> no more
13:44:56 <Qiming> alright, thanks for joining boys and girls
13:45:01 <Qiming> have a good night
13:45:09 <XueFengLiu> good night
13:45:16 <yanyanhu> have a good night :)
13:45:17 <Qiming> we can end this earlier
13:45:17 <ruijie> good night :)
13:45:24 <Qiming> #endmeeting