13:00:19 <Qiming> #startmeeting senlin 13:00:19 <openstack> Meeting started Tue Aug 9 13:00:19 2016 UTC and is due to finish in 60 minutes. The chair is Qiming. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:00:20 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 13:00:23 <openstack> The meeting name has been set to 'senlin' 13:00:45 <Qiming> hello? 13:02:59 <elynn> o/ 13:03:07 <Qiming> hi, elynn 13:03:23 <guoshan_> hi, everyone~ 13:03:25 <elynn> evening Qiming 13:03:29 <Qiming> hi, guoshan_ 13:04:06 <Qiming> pls feel free to add items to the meeting agenda 13:04:20 <Qiming> #link https://wiki.openstack.org/wiki/Meetings/SenlinAgenda#Weekly_Senlin_.28Clustering.29_meeting 13:04:43 <Qiming> due to firewall problem, adding meeting items is not easy now, we have to use proxy 13:05:17 <Qiming> yanyan cannot join us today, I heard nothing from xinhui or haiwei 13:05:25 <Qiming> let's get started? 13:05:29 <guoshan_> ok, It's my first time in IRC meeting 13:05:37 <Qiming> thanks, guoshan_ 13:05:49 <Qiming> #link https://etherpad.openstack.org/p/senlin-newton-workitems 13:06:09 <Qiming> let's go thru the etherpad and see where we are 13:06:47 <lixinhui_> hi 13:07:01 <Qiming> yanyan has updated the etherpad for Rally support 13:07:05 <Qiming> hi, lixinhui_ 13:07:13 <Qiming> we just got started 13:07:43 <Qiming> the Rally side support was just merged, Roman has help review another patch in senlin code base, which is good ... 13:08:13 <Qiming> I'm not gonna copy/paste yanyan's update here 13:08:17 <Qiming> please read it 13:08:36 <Qiming> Will delete those text later 13:08:44 <Qiming> integration test 13:08:54 <Qiming> test job has been merged at gate side 13:09:11 <Qiming> we found things to be fixed related to neutron set up 13:09:42 <Qiming> these tests are pretty important for us to ensure we are not breaking any contracts with any other services 13:10:29 <Qiming> at the same time, I'm working on revising the docs about testing infrastructure 13:10:36 <Qiming> #link http://git.openstack.org/cgit/openstack/senlin/tree/doc/source/developer/testing.rst 13:10:47 <Qiming> the current doc is only about unit tests 13:11:13 <Qiming> while we have added api test, functional test, integration test and stress test (ongoing) 13:11:37 <Qiming> need to update the docs so we all know how to run them in our local env 13:12:05 <Qiming> pls help review the patches and see if there things to be fixed 13:12:21 <Qiming> https://review.openstack.org/347654 13:12:22 <guoshan_> sure, pleasure to do that 13:12:29 <Qiming> https://review.openstack.org/348110 13:12:41 <Qiming> moving on ... 13:12:46 <Qiming> health management 13:13:17 <Qiming> lixinhui_, the LB member status one is fixed? 13:14:02 <lixinhui_> 2 of 3 patches have been approved 13:14:19 <lixinhui_> the root one is pending for cores' approval 13:14:19 <Qiming> great 13:15:03 <Qiming> next thing is about fencing 13:15:21 <Qiming> I know you have added some support into python-openstacksdk side 13:15:52 <Qiming> those are already well shaped, just address brian's concern would be fine 13:15:56 <lixinhui_> neutron-lbaas community is not very active 13:15:57 <lixinhui_> I added Armando as reviewer 13:15:57 <lixinhui_> hope he can help 13:15:57 <lixinhui_> in future, Octavia will replace lbaas 13:16:06 <lixinhui_> Brian just add comments there 13:16:23 <Qiming> yep 13:16:39 <lixinhui_> okay 13:16:39 <lixinhui_> I will followup soon 13:16:48 <Qiming> for those of you who are not aware of these patches: 13:16:49 <Qiming> https://review.openstack.org/352723 13:16:56 <Qiming> https://review.openstack.org/351061 13:17:20 <Qiming> these are the ones added to sdk so that we can invoke nova service operations later -- for fencing's purpose 13:17:46 <Qiming> moving on to next item 13:17:59 <Qiming> documentation, not much update from me 13:18:12 <Qiming> but I did submitted some patches about api docs 13:18:41 <Qiming> one thing to notice is that the global version requirement of os-api-ref has been bumped to 0.4.0 13:19:00 <Qiming> which means it has a better support to success code and error code now 13:20:03 <Qiming> if you are interested in it, you can check this subsection: http://git.openstack.org/cgit/openstack/os-api-ref/tree/doc/source/usage.rst#n185 13:20:38 <Qiming> moving on to next one 13:20:46 <Qiming> profile/policy version control 13:20:58 <Qiming> yanyan has proposed a patch which I forgot to review 13:21:11 <Qiming> please jump on to it when you have time 13:21:40 <Qiming> it is very important a feature, though in future we plan to generalize versioning control of many things 13:22:11 <Qiming> https://review.openstack.org/#/c/348709/ 13:22:18 <elynn> will review it later. 13:22:19 <Qiming> ^ is the patch 13:22:29 <Qiming> thx elynn 13:23:07 <Qiming> next topic is container cluster support 13:23:28 <Qiming> haiwei has a new patchset for review: https://review.openstack.org/#/c/349906 13:23:53 <Qiming> though it still not comprehensive, as a working prototype, it looks not bad 13:24:09 <Qiming> also need reviews from you 13:24:31 <Qiming> haiwei is not online today, seems 13:24:41 <Qiming> so we move on to next one 13:24:47 <Qiming> zaqar receiver support 13:25:06 <Qiming> yanyan has been pushing hard on zaqar support at sdk side 13:25:28 <Qiming> due to api doc and some implementation nits, the patches are still pending for review/merge 13:25:42 <Qiming> but we are pretty close to get it done 13:26:06 <Qiming> next one 13:26:11 <Qiming> events/notifications 13:26:37 <Qiming> not update from me on this, probably we won't have time to get it done in newton 13:27:40 <Qiming> some updates on cluster-collect 13:28:02 <Qiming> the senlin side has a fix to the docs merged: https://review.openstack.org/350982 13:28:40 <Qiming> and finally we got API micro-versioning support merged in openstacksdk: https://review.openstack.org/343992 13:29:17 <Qiming> following that, we got a new API (collect_cluster_attrs) merged: https://review.openstack.org/350978 13:29:50 <Qiming> I'm now working at senlinclient side on the micro-versioning and cluster-collect operation support 13:31:03 <Qiming> the output is something like this: 13:31:04 <Qiming> $ senlin cluster-collect -p details.addresses.private[0] c1 13:31:04 <Qiming> WARNING (shell) "senlin cluster-collect" is deprecated, please use "openstack cluster collect" instead. 13:31:04 <Qiming> 10.0.0.12 13:31:04 <Qiming> 10.0.0.13 13:31:04 <Qiming> 10.0.0.14 13:31:48 <elynn> looks great! 13:32:27 <Qiming> and also this: 13:32:28 <Qiming> $ openstack cluster collect -f value -c attr_value --path details.addresses.private[0] c1 13:32:28 <Qiming> 10.0.0.12 13:32:28 <Qiming> 10.0.0.13 13:32:28 <Qiming> 10.0.0.14 13:32:56 <Qiming> need to add test cases then commit the patch 13:33:13 <elynn> a question, how would the users know what kind of attributes they can get? 13:33:34 <Qiming> it is a JSON path 13:33:47 <Qiming> when you do senlin node-show <node_id> 13:33:55 <Qiming> you get the attributes 13:33:58 <elynn> oh, I see. 13:34:32 <Qiming> then you think to yourself, how can I get all the values from all nodes instead of doing cluster node-show one by one 13:34:52 <Qiming> well, that is when you decide to use senlin cluster-collect or openstack cluster collect 13:35:27 <Qiming> by the way, just found a bug in sdk and proposed a patch 13:35:28 <Qiming> https://review.openstack.org/352853 13:35:31 <elynn> Glad that we could provide this API. 13:36:02 <Qiming> when senlin is retrieving the details about a nova server, the 'image' property disappeared, so does the 'flavor' property 13:36:15 <Qiming> hope it can be merged soon 13:37:17 <Qiming> besides that, I don't have more updates 13:37:38 <Qiming> questions/comments about items on etherpad? 13:38:20 <Qiming> okay, that is a no 13:38:23 <elynn> Not from me, newton-3 will be end this month? 13:38:23 <Qiming> :) 13:38:51 <Qiming> yep, http://releases.openstack.org/newton/schedule.html 13:39:31 <Qiming> newton-3 on august 29 13:39:40 <Qiming> rc-1 on sep 12 13:39:47 <elynn> okay.thanks 13:39:55 <Qiming> rc-2 on sep-26 13:40:28 <Qiming> since we don't have a lot of features pipeline 13:40:45 <Qiming> so we don't need to practice a feature freeze type of thing 13:41:11 <Qiming> just work to our best ... deliver what can be made stable 13:41:18 <Qiming> #topic open discussions 13:42:04 <Qiming> anything else? 13:42:16 <lixinhui_> Qiming, have you tried the vertioned_nova? 13:42:31 <lixinhui_> versioned_nova? 13:42:32 <Qiming> no 13:42:39 <Qiming> what is versioned_nova? 13:43:05 <Qiming> a software? 13:44:13 <lixinhui_> https://review.openstack.org/350364 13:45:11 <Qiming> okay, you mean versioned notification from nova 13:45:47 <Qiming> it is good to add versioned_notifications as a target 13:46:23 <Qiming> my question was whether that new target can be used for notifications with 'notifications' as topic 13:47:02 <lixinhui_> why that way? 13:47:24 <Qiming> or, a different way to ask the question ... has nova migrated all their notifications to this versioned way? 13:48:54 <lixinhui_> is that enough to migrate all instance.update? 13:48:55 <Qiming> a safer way to add 'versioned_notifications' as a Target is to keep the old one, we won't get hurt if there are not messages with that topic 13:50:04 <Qiming> but if nova is only migrating part of their notifications to the versioned way, ... that is bad 13:50:14 <lixinhui_> according to the doc, original one will be retired 13:50:27 <lixinhui_> why we receive two channels 13:50:37 <Qiming> I never trust their docs 13:50:47 <Qiming> they said they will work out a v3 api 13:50:56 <Qiming> they said they will split scheduler out 13:51:21 <lixinhui_> I tested instance.update and service.update 13:51:36 <Qiming> we can listen to two channels, if one channel is really quiet, we can delete it later 13:51:54 <Qiming> yes, I figured that 13:52:08 <Qiming> but instance.update and service.update is not the whole world, right? 13:52:13 <lixinhui_> how about noisy 13:52:23 <lixinhui_> according to its design, yes 13:52:47 <lixinhui_> whole world of nova notifications 13:52:52 <lixinhui_> just throw a question here 13:53:20 <lixinhui_> we can leave the old one till compute service failure detection 13:54:18 <Qiming> a target is constructed with a 'topic' and an 'exchange' 13:54:29 <Qiming> we already set the 'exchange' to 'nova' 13:54:43 <lixinhui_> there is no change on exchange 13:55:09 <lixinhui_> I use Rabbitmq as message provide 13:55:16 <Qiming> the combination means, we listen to 'notifications' from 'nova' 13:55:37 <Qiming> if nova has migrated all its notifications to the versioned implementation 13:55:43 <Qiming> that target should be quiet 13:55:59 <lixinhui_> provider 13:55:59 <lixinhui_> and confirm no change on exchange 13:56:01 <Qiming> if it is not, it means nova is still doing the migration 13:56:17 <Qiming> I know you didn't change exchange 13:56:36 <Qiming> I'm trying my best to explain to you why I think keeping the existing target doesn't hurt us 13:57:30 <lixinhui_> not I change the exchange 13:57:48 <lixinhui_> I just trying to explain the test result 13:57:52 <lixinhui_> anyway 13:57:52 <Qiming> if nova has migrated all notifications to versioned one, the existing target will be quiet, we won't get anything from it, we are not hurt 13:58:12 <Qiming> you are only testing instance.update and service.update, right? 13:58:48 <Qiming> do we know when an instance is stopped, or deleted? using the 'instance.update' notification? 13:59:10 <Qiming> or maybe the protocol has changed, I haven't looked into that for a time 13:59:28 <lixinhui_> yes 13:59:55 <Qiming> okay, I'll do some experiements to verify 14:00:00 <lixinhui_> stop or delete are details of instance.update 14:00:01 <Qiming> thanks for bringing this up 14:00:11 <Qiming> time is up 14:00:18 <Qiming> thanks for joining, guys 14:00:21 <Qiming> #endmeeting