19:01:51 <dtroyer> #startmeeting openstackclient 19:01:52 <openstack> Meeting started Thu Mar 24 19:01:51 2016 UTC and is due to finish in 60 minutes. The chair is dtroyer. Information about MeetBot at http://wiki.debian.org/MeetBot. 19:01:53 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 19:01:55 <openstack> The meeting name has been set to 'openstackclient' 19:02:01 <dtroyer> anyone here? 19:02:08 <rtheis> hi 19:04:12 <dtroyer> this may be a quick meeting... 19:04:48 <singhj> o/ 19:05:23 <dtroyer> well, lets get started 19:05:29 <dtroyer> #topic reviews 19:05:43 <dtroyer> Reviews to discuss? 19:05:58 <rtheis> https://review.openstack.org/#/c/290849/9/doc/source/command-objects/port.rst 19:06:30 <rtheis> dtroyer: question in there for you on using --clear versus --no versus an unset actin 19:06:52 <dtroyer> is that in the create command? 19:07:20 <singhj> dtroyer: set command 19:07:39 <rtheis> dtroyer: clear would be for set only 19:07:56 <singhj> dtroyer: specifically ex. port set --clear-security-groups <port> 19:08:15 <dtroyer> so a port can haz only one security group? 19:08:31 <dtroyer> nm, I see it 19:08:31 <singhj> it may have multiple 19:09:09 <dtroyer> for multiple I think it's easy to stick with the precedent of —clear-XXX 19:09:23 <dtroyer> unset could do it too though 19:09:39 <dtroyer> port unset —security-group <name> <port> 19:09:54 <dtroyer> to clear all would be a pita though 19:10:06 <dtroyer> should we do both? otherwise to remove just one it's a clear and re-load 19:11:07 <singhj> well, right now, I've got --clear-xxx working, i'd be convenient to be able to unset 19:11:16 <singhj> it'd* 19:11:32 <singhj> I'm fine with either way, rtheis? 19:11:41 <rtheis> singhj: does the REST API allow unsetting a specific group? 19:12:47 <rtheis> maybe not, you probably just have to pull it out of the list 19:12:50 <singhj> it does not, it seems to be all or none. there is one similar command "--extra-dhcp-opts" that is based on individual unsetting 19:12:50 <dtroyer> it might take a GET, remove from dict, PATCH cycle to do 19:12:57 <rtheis> yeah 19:13:02 <singhj> but to achieve that you basically set a property to null 19:13:05 <singhj> its a bit hacky 19:13:58 <dtroyer> I would like to see unset support, it need not be in the first cut though. 19:14:03 <rtheis> I like --clear option and possibly adding unset command support 19:14:05 <dtroyer> should —clear-NNN be in set or unset? 19:14:20 <dtroyer> it feels like an unset option to me 19:14:32 <dtroyer> but I could defend either choice 19:14:45 <singhj> I can port it over, it's currently in set 19:14:47 <rtheis> yeah, I could go either way 19:15:15 <dtroyer> where is it in the route commands? 19:15:20 <rtheis> set 19:15:39 <dtroyer> would it be absurd to do both? 19:15:46 <singhj> where do we have unset working so far? 19:15:53 <singhj> I havent encountered it 19:16:03 <rtheis> I think it would be okay to do both 19:16:31 <rtheis> singhj: no unset for network commands 19:16:41 <dtroyer> a lot of places: flavor, server, project, etc 19:16:43 <rtheis> singhj: but you'll find it used in others 19:16:48 <dtroyer> ok, maybe not a lot, I count 10 19:17:04 <singhj> ah okay, sure 19:17:07 <dtroyer> usually for removing properties 19:17:11 <singhj> I'm fine doing it in both places 19:17:30 <singhj> when we say both do we mean --clear-* in unset and set 19:17:37 <dtroyer> lets do that then. it feels just a tiny bit odd, but I can live with it 19:17:38 <singhj> is that the "both"? 19:17:44 <dtroyer> yes 19:18:11 <singhj> sure, sounds good to me 19:18:12 <dtroyer> so if —clear- and —security-group both appear in set, what should we do? 19:18:21 <dtroyer> is that the reason to not do clear in the set command? 19:18:30 <dtroyer> or should it clear first, then set? 19:18:39 <rtheis> I think that would be mutually exclusive group 19:18:42 <singhj> well right now they're mutually exclusive 19:18:46 <dtroyer> ok 19:19:06 <dtroyer> that'll work 19:19:21 <singhj> okay cool 19:19:23 <rtheis> thanks 19:19:35 <dtroyer> any others? 19:19:40 <rtheis> nothing from me 19:19:56 <singhj> same 19:20:03 <dtroyer> sheel and I were kicking around the volume transfer stuff this morning 19:20:27 <dtroyer> unfortunately the implementation in cinder is nothing like how image member/sharing works in glance so we can't take the same approach 19:21:04 <dtroyer> I think we can make it better but haven't convinced him yet ;) 19:21:43 <rtheis> dtroyer: this review https://review.openstack.org/#/c/292043/ ? 19:22:01 <dtroyer> yes 19:22:59 <dtroyer> I think 'volume transfer' is a very confusing resource name…it looks like an action 19:23:18 <rtheis> agreed 19:23:21 <dtroyer> but aside from that, the mechanics of the transfer are odd 19:23:46 <dtroyer> in a way that doesn't make adding it as options to the existing volume commands easy 19:24:00 <dtroyer> so, ideas are welcome 19:24:14 <rtheis> I'll put it on my review list 19:24:29 <dtroyer> also, https://review.openstack.org/#/c/297157/ could use a quick look rtheis 19:24:46 <dtroyer> its part of the release runup 19:24:51 <rtheis> ok 19:25:14 <dtroyer> How about… 19:25:18 <dtroyer> #topic bugs 19:25:54 <dtroyer> I have been doing UX studies with Piet Kruithof, and in 4 sessions found three bugs 19:26:16 <dtroyer> some may already exist, I haven't checked 19:26:28 <dtroyer> any bugs to bring up? 19:26:34 <rtheis> https://bugs.launchpad.net/python-openstackclient/+bug/1560157 19:26:34 <openstack> Launchpad bug 1560157 in python-openstackclient ""openstack network list" fails with "An SSL error occurred."" [Undecided,New] 19:26:41 <rtheis> this one caught my eye 19:27:10 <rtheis> maybe SDK related or interaction between osc and sdk auth 19:27:23 <dtroyer> that's interesting… yeah, I wonder if any other SDK commands do that? 19:27:32 <dtroyer> aren't we using the same sesison object either way though? 19:27:56 <rtheis> I thought so, but will have to check 19:28:23 <dtroyer> also, could be how the endpoint is conigured 19:28:28 <dtroyer> configured 19:28:45 <dtroyer> we should get some openssl commands to dump the endpopint info together for debugging 19:29:34 <dtroyer> #action make some notes for debugging SSL bits to get info from the server-side endpoint 19:29:52 <rtheis> sounds good 19:30:12 <rtheis> no other bugs from me 19:31:42 <dtroyer> ok, so what else is going on? 19:31:46 <dtroyer> #topic open discussion 19:32:29 <dtroyer> I want to do a release soonish, at least one more before the summit 19:32:34 <dtroyer> maybe two 19:32:47 <rtheis> sound good 19:33:05 <rtheis> any updates on summit schedule for osc? 19:33:36 <dtroyer> no schedule for sessions yet 19:33:51 <rtheis> ok 19:33:59 <dtroyer> I may be doing an on-stage update in a keynote 19:34:22 <rtheis> nice 19:34:38 <dtroyer> and I think Piet is going to do some UX sessions with operators on Tuesday or Wednesday 19:34:59 <dtroyer> we'll get some better info on commands more complidated than creating a server 19:35:18 <dtroyer> which is bad enough but is also pretty well understood and debugged 19:36:34 <dtroyer> ok if there is nothing else, lets be done 19:36:38 <rtheis> FYI: briancurtin has a sdk resource refactor https://review.openstack.org/#/c/289998/ ... the results of which will eventually impact osc 19:36:55 <rtheis> this is part of the sdk path to 1.0 19:37:07 <dtroyer> I saw a comment along those lines…and was the kind of reason I hoped we could wait until 1.0 19:37:38 <dtroyer> can we handle it in a compatible way? 19:37:50 <rtheis> yeah...I plan to dig into it further to see if we can handle in a compatible way 19:37:56 <briancurtin> the impact should be fairly minimal, mostly only affecting resources that have a child/parent relationship (server and server_interface, for example) 19:38:45 <briancurtin> i guess it could also affect you if you’re depending on resource properties that aren’t actually properties on resource classes…as in you’re just taking advantage of the fact that it previously would just send and receive arbitrary data 19:39:24 <rtheis> briancurtin: I think the later is what may impact some osc options for network commands 19:40:07 <briancurtin> rtheis: i’m going to be tackling some of that while applying the changes to the resources, as i’m doing the changes with the docs open next to the files rather than only doing find/replace…sort of an assisted find/replace 19:40:46 <rtheis> ok 19:41:33 <dtroyer> so I think we're going to need to either be able to straddle this in a single OSC release or do an osc 3.0 when we get those changes 19:41:52 <dtroyer> briancurtin: that makes its first appearance in 1.0 right? 19:42:46 <briancurtin> dtroyer: i’m not sure i understand the question — asking about the refactored resources and proxies? 19:42:57 <dtroyer> yes 19:43:37 <briancurtin> dtroyer: i haven’t exactly worked out how to release this stuff, but yeah, i think once it’s all in it will be released as 1.0 19:43:38 <dtroyer> we've talked about doing a major rev soon, it might make sense to use SDK 1.0 in our OSC 3.0 19:44:46 <dtroyer> ok, anything else? 19:44:53 <rtheis> nothing else 19:45:18 <dtroyer> ok, thanks guys 19:45:27 <dtroyer> #endmeeting