20:00:46 <robcresswell> #startmeeting horizon
20:00:47 <openstack> Meeting started Wed Jun  7 20:00:46 2017 UTC and is due to finish in 60 minutes.  The chair is robcresswell. Information about MeetBot at http://wiki.debian.org/MeetBot.
20:00:48 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
20:00:50 <openstack> The meeting name has been set to 'horizon'
20:01:11 <robcresswell> o/
20:01:15 <lucasxu> o/
20:01:16 <ying_zuo> o/
20:02:18 <robcresswell> Hey everyone
20:02:22 <robcresswell> #topic Notices
20:02:48 <robcresswell> Only one quick announcement today; tomorrow is Pike-2 milestone, so I'll be doing some stability checking then tagging that tomorrow
20:03:41 <robcresswell> There's a few big blueprints that are making good progress; QoS, Trunks, Overview rewrite
20:04:48 <robcresswell> I think thats most of the big ticket items...
20:05:00 <robcresswell> Any other announcements to make?
20:05:49 <e0ne> hi
20:06:56 <robcresswell> #topic Open Discussion
20:07:22 <robcresswell> Open floor! Any questions, blueprints, bugs, patches, all welcome :)
20:07:45 <lucasxu> lil
20:07:51 <lucasxu> oops : https://review.openstack.org/#/c/450566/
20:08:12 <e0ne> I hope, it will be a very quick
20:08:18 <e0ne> I would like to ask about ddt https://review.openstack.org/468058 once more
20:08:52 <e0ne> should I abandon this patch or we'll go forward with it?
20:08:59 <lucasxu> I am wondering if I could do anything for that patch
20:10:30 <robcresswell> e0ne: I'm not sure its a big enough advantage to fracture the code base
20:10:48 <e0ne> lucasxu: I'm not sure that it fits openstack stable maint. policies:(
20:11:18 <robcresswell> lucasxu: It looks like we'll need something else in master and than squash it, cant backport incompatible changes
20:11:21 <e0ne> robcresswell: I won't argue more. Everything is in a patch.
20:11:37 <lucasxu> eOne, really? I thought nobody gonna use volume as the v1 endpoint
20:11:46 <e0ne> robcresswell: if community doesn't like this approach - I'll abandon the patch
20:12:20 <robcresswell> lucasxu: If I understand, the problem is that the endpoint could be 'volume' and still be v3
20:12:21 <e0ne> lucasxu: I hope nobody uses cinder v1
20:13:21 <lucasxu> robcresswell, you are right. But normally, you won't configure 'volume' to be the v3 endpoint. you have 'volumev3'  But yeah, somebody will
20:13:50 <lucasxu> e0ne, m2 hah
20:13:56 <e0ne> :)
20:14:36 <robcresswell> I didnt realise the service names could be configured, tbh, I thought it was just the endpoint names
20:14:43 <robcresswell> as in, cinderv3 is volumev3
20:14:48 <robcresswell> in devstack, for example
20:15:27 <lucasxu> yeah, so I don't know what to do for this patch now.
20:16:36 <robcresswell> its a bizarre one, because couldn't this apply for any service?
20:17:08 <robcresswell> We do checks for compute etc
20:17:18 <lucasxu> hmm, idk. You could delete the nova endpoint for example and add it back with a different name called novav2?
20:17:37 <lucasxu> okay
20:17:46 <robcresswell> amotoki: Around?
20:18:00 <david-lyle> o/
20:18:55 <robcresswell> o/
20:19:06 * david-lyle stumbled in later
20:19:08 <david-lyle> *late
20:19:26 <robcresswell> david-lyle: Know anything about configurable service names?
20:19:34 <robcresswell> Like https://review.openstack.org/#/c/450566/
20:19:58 * david-lyle looking
20:20:34 <david-lyle> robcresswell, name or type?
20:20:50 <david-lyle> type should not be configurable, name is
20:20:57 <robcresswell> type, thats what we check
20:21:02 <david-lyle> but all of those checks are for name
20:21:06 <david-lyle> err type
20:21:16 <david-lyle> names are fair game
20:21:16 <robcresswell> right, thats what I thought
20:21:21 <robcresswell> yeah
20:21:27 <robcresswell> but if we're checking type, that should be reliable
20:21:37 <david-lyle> yes, type should not change
20:21:44 <robcresswell> but apparently a cinder v3 endpoint can have a type 'volume' or 'volumev3'
20:21:55 <robcresswell> thats what amotoki is saying, at least :/
20:22:22 <david-lyle> cinder has 3 endpoints possible, I believe
20:22:35 <robcresswell> yeah, one for each major api version
20:22:53 <david-lyle> volume and volumev2 for sure, I think volumev3 maybe?
20:23:47 <lucasxu> volume for cinder v1 and it should be deprecated I think
20:23:57 <david-lyle> I'm not sure amotoki's concern is valid, but still reading
20:23:59 <e0ne_> v2 is also deprecated but used
20:24:13 <lucasxu> so now more likely it is volumev2 and volumev3 I believe
20:24:43 <robcresswell> v1 is dead afaik
20:25:05 <lucasxu> e0ne, yeah, you are right, even v2 deprecated
20:25:18 <e0ne_> http://paste.openstack.org/show/611765/ - cinder endpoints on devstack
20:25:45 <e0ne_> lucasxu: nit: my nickname is with a 'zero' symbol
20:26:24 <lucasxu> e0ne_, hah, I notice that but I forgot the underscore
20:26:39 <e0ne_> oops, my fault :(
20:27:39 <lucasxu> robcresswell, anyways, we could discuss this on gerrit. Thanks for your time
20:27:49 <lucasxu> dw.
20:27:52 <david-lyle> should line 548 not be checking for volume OR volumev2?
20:28:20 <david-lyle> then we don't break backward compatibility?
20:29:07 <lucasxu> david-lyle, volume is for v1 and we don't support v1 any longer so that's why I am checking v2 / v3
20:29:22 <david-lyle> sure we do
20:29:54 <david-lyle> horizon usually supports older versions even if deprecated for some time
20:30:17 <david-lyle> the only thing we've really booted support for was nova-net
20:30:30 <david-lyle> s/booted/removed/
20:30:46 <robcresswell> david-lyle: We don't support it in master any more. It's been eol for several cycles afaik.
20:31:03 <david-lyle> in ocata though?
20:31:25 <david-lyle> isn't that the target branch?
20:31:33 <david-lyle> you can edit cherry-picks
20:31:48 <david-lyle> you just have to add the justification in the commit message
20:32:03 <david-lyle> maybe I missed too much of the conversation
20:32:19 <robcresswell> Yeah, we'd need to support all 3 for ocata
20:33:09 <robcresswell> Cinder guys are saying we can rely on it being 'volumev3', FYI.
20:33:46 <david-lyle> types are fixed for openstack services
20:34:37 <robcresswell> it seems ambiguous, the cinder guys said thats only if following best practices
20:35:00 <robcresswell> Though I'm happy with that; if someone wants to go crazy with custom config thats their problem.
20:35:01 <david-lyle> if those practices aren't followed, Horizon fails to work
20:35:05 <david-lyle> for all services
20:35:05 <robcresswell> +1
20:35:12 <robcresswell> yup
20:35:14 <smcginnis> david-lyle: So will Cinder most likely.
20:35:19 <lucasxu> sounds good
20:35:51 <robcresswell> that was what I was confused about, since we do those type checks all over; I was confused about why it was suddenly an issue.
20:36:22 <lucasxu> hah. thanks guys.
20:36:24 <david-lyle> I think the random dropping of 'volume' in ocata was the issue
20:36:25 <robcresswell> anyway, thats mostly resolved :)
20:36:30 <robcresswell> yeah
20:36:43 <david-lyle> just not clearly articulated
20:37:02 <robcresswell> We shouldnt remove that support post-release, at least.
20:37:44 <david-lyle> if this is a backport, and the cherry-pick is edited, I'm not sure what's left to change on master?
20:38:25 <robcresswell> nothing
20:38:41 <robcresswell> master can stay as-is, but the cherry pick needs altering if we want to backport it
20:38:55 <david-lyle> misread you previous comment, ignore my last question
20:40:02 <robcresswell> :)
20:40:14 <robcresswell> Any other questions or issues anyone would like to discuss?
20:47:51 <david-lyle> robcresswell, #endmeeting :)
20:50:25 <lucasxu> LOL I think he's gone
20:57:22 <robcresswell> david-lyle, lucasxu: Ha, forgot to hit send
20:57:30 <robcresswell> Thanks everyone!
20:57:35 <robcresswell> #endmeeting