16:02:27 #startmeeting Cinder 16:02:27 hi 16:02:27 is smcginnis on vacation? 16:02:27 Meeting started Wed Mar 30 16:02:27 2016 UTC and is due to finish in 60 minutes. The chair is e0ne. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:02:28 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:02:30 hi 16:02:31 The meeting name has been set to 'cinder' 16:02:33 hi 16:02:37 Hello! 16:02:38 hi 16:02:39 xyang1: I think so 16:02:41 o/ 16:02:43 smcginnis is on vacation, yes 16:02:46 o/ 16:03:00 hi 16:03:04 I was going to start the meeting but was running a couple of minutes late 16:03:08 I was not ready, so no anouncments, I think 16:03:14 <_alastor_> hi 16:03:19 e0ne can have it instead 16:03:19 o/ 16:03:36 DuncanT: looks like your topic in the first 16:03:47 #topic Permanently undeprecating python-cinderclient CLI 16:03:57 Oh, excellent 16:04:03 #link https://bugs.launchpad.net/python-openstackclient/+bug/1561666 16:04:03 Launchpad bug 1561666 in python-openstackclient "python-openstackclient incorrectly calls cinder metadata 'properties'" [Undecided,Won't fix] 16:04:31 #link http://lists.openstack.org/pipermail/openstack-dev/2016-March/090375.html 16:04:32 o/ 16:04:54 So the 'official' client is ignoring docs, project conventions and other libraries, and renaming things as it wishes. I think this is a terrible idea, so I propose we un-depricate python-cinderclient 16:04:54 DuncanT: I'm agree with you 16:05:07 DuncanT: but can we do it withoit TC approvement? 16:05:27 e0ne: I suggest we just do it, and let them come back to us 16:05:31 :) 16:05:38 Better to ask forgiveness than permission 16:05:39 e0ne: I'm happy to go to the TC about it too 16:05:43 DuncanT: Is this a long term fix or just short term? 16:06:00 baumann: I think the openstack-client is long-term broken 16:06:05 DuncanT: lol 16:06:08 DuncanT: we have to add this thing the TC's meeting agendaa 16:06:10 * jgriffith is confused 16:06:13 XD 16:06:21 DuncanT: Maybe we should look for consensus once again? 16:06:22 oh... wait, I see what you're saying now 16:06:23 Well we should technically be moving to the OpenStack Client anyway right? 16:06:37 baumann: I don't think having an 'official' client that doesn't match the project docs is a good idea 16:06:43 seems like we should at least keep it un-depricated until we are happy with osc as a replacement 16:06:51 DuncanT: wouldn't the best this to fix the problems with OSC? 16:06:54 diablo_rojo: yes, but it provides different UX than cinderclient does 16:06:54 patrickeast: +1 16:06:55 diablo_rojo: that's DuncanT 's point I think 16:06:56 yeah, this "Required name option" thread this week really worries me 16:07:10 patrickeast: +1 16:07:12 diablo_rojo: because they use "properties" instead of "metdata"... I think 16:07:13 erlon: They aren't interested in fixing it, they don't consider it a but 16:07:25 jgriffith: There are other little details too 16:07:35 jgriffith: That was just the most obvious one 16:07:37 patrickeast: ++ 16:07:47 I think we need to work to get the osc right. 16:07:53 patrickeast: +1 16:07:57 jgriffith: Okay but the end goal is to get to the OSC right? So we are going to have to make changes to use the 'metadata' eventually right? 16:08:04 jungleboyj: +1 16:08:06 I think we should undeprecate the client and try to discuss this again with OSC community. Maybe at the summit? 16:08:12 jgriffith: Or am I wrong? 16:08:13 diablo_rojo: yes, absolutley correct on the end-goal 16:08:17 diablo_rojo: OSC have said 'no' to metadata 16:08:21 Now with the cross-project spec OSC is far more important for every project. 16:08:27 as I understand it they have intentionally normalized things across services 16:08:27 dulek: I don't think that's a bad option for now. 16:08:30 dulek: +1 16:08:32 hence properties 16:08:40 DuncanT, diablo_rojo: the same for not-required name option 16:08:49 this is intentional as part of giving users a sane ui 16:08:49 clarkb: I think that's a bad idea 16:08:51 jgriffith: Oh okay. 16:08:56 clarkb: They don;t 16:08:56 personally I prefer properties :) 16:09:06 jgriffith: Because you hate metadata. :D 16:09:10 :) 16:09:10 DuncanT: sure you eill have to take that up with them just trying to explain what is going o here 16:09:17 dulek: Lol 16:09:29 jgriffith: I would if we didn't already have it referred to as metadata in many other places 16:09:35 dulek: +1 :) abused, poorly defined and significantly over-used :) 16:09:39 as an osc user the initial switch is annoying and some.stuff is still missing but overall works well 16:10:01 DuncanT: so does that mean we *never* evolve or fix things? 16:10:12 jgriffith: I agree with the overused part. It seems to be everywhere. 16:10:12 DuncanT: all in the name of "that's how we've always done it" 16:10:42 jgriffith: No, it means changes have to be done with some thought, not by doing them in one client and calling that the one true way. 16:10:46 I agree with clarkb; there are some transition pains, but normalizing and moving forward seems like a good thing 16:10:57 I'm just a little concerned on how this will look outside. We should make sure that the decision is clearly stated and motivated. So we don't look like we're trying to swim against the tide. 16:11:02 jgriffith: I agree. 16:11:03 jgriffith: +1 16:11:06 jgriffith: The OSC client doesn't match the cinder client, horizon, the docs or other libraries 16:11:16 I think the problem is that we haven't taken an active role in the transition that I am aware of. 16:11:23 jgriffith: +1 16:11:23 I know users are wanting to move in this direction. 16:11:31 DuncanT: well... maybe that's because it's fixing some of the "bad" things we've done in those projects over the years? 16:11:39 jungleboyj: +1 16:11:52 jgriffith: If we're going to rename something, it needs to be coordinated, and fixed everywhere. And official client that doesn't match the docs is broken 16:11:55 I am fine with keeping cinderclient for now until we get full parity in osc. 16:12:00 s/And/An/ 16:12:11 jungleboyj: +1 16:12:16 jungleboyj: +1 16:12:31 jungleboyj: That seems to be the right answer for now 16:12:41 jungleboyj: well... depending on what you mean by *parity*, that would actually be a step backwards depending on what you envision 16:12:43 jgregor: sheel and baumann How is the OSC coming? 16:12:49 Need to take a more active role in understanding DuncanT s concerns though and get osc fixed. 16:13:00 but anyway... I don't want to wast peoples time on this 16:13:05 * smcginnis2 listens in while I have access 16:13:10 jgriffith: By parity, I mean getting all the same commands implemented in osc. 16:13:27 jungleboyj: I think that's the wrong idea... DuncanT 's example is a good one there. 16:13:28 jgriffith: Not making it look exactly the same if that if your concern. 16:13:39 jgriffith: ? 16:13:44 diablo_rojo: great 16:13:46 jungleboyj: so dump "metadata" commands and replace with the standardized "properties" naming 16:13:55 diablo_rojo: I am almost done with half of work 16:13:58 diablo_rojo: We have a topic coming up with a question we have 16:14:05 jungleboyj: see what I mean? 16:14:19 DuncanT: for me that means taht Cinder need to work in the direction to support, OSC, fix docs etc. If it is not ready right now, the deprecation should stay for a longer time, otherwise nobody will work to make thinks 16:14:33 jgriffith: You are saying that is a good or bad thing? 16:15:00 We need parity before we consider it - in the sense that users are able to accomplish the same things. 16:15:01 jungleboyj: I'm saying moving forward and changing things is a good thing 16:15:02 can we follow keystone way? 16:15:10 jgriffith: Me too! 16:15:13 jgriffith: If we do that via deprecating the name in the cinderclient, preferably the API too, updating the docs, etc, then fine. Not by just deciding to call it something new in the OSC and leaving the mismatch 16:15:14 smcginnis2: +1 16:15:23 smcginnis2: yes, that I agree with which was my caveat about "how one defines parity" 16:15:23 smcginnis2: Exactly! 16:15:24 :) 16:15:27 inplement only api v3 with microversions in OSC? 16:15:34 DuncanT: Agreed 16:15:42 I don't care if it's the same syntax, names etc... just that i can do the "same stuff" 16:15:44 e0ne: osc isn't far enough along yet for that to fit our timeline 16:15:55 DuncanT: That sounds good to me. 16:16:09 it sounds like most people are on board with keeping cinderclient around then until we are happy with osc? 16:16:14 DuncanT: +1 16:16:20 DuncanT: There is work to be done there. I am hoping my team can help with that. 16:16:26 patrickeast: +1 16:16:33 patrickeast: Yes 16:16:39 jungleboyj: I am already up on that 16:16:43 patrickeast: +1 16:16:43 patrickeast: yes 16:16:45 patrickeast: +1 16:16:47 jungleboyj: with jgregor 16:17:01 jungleboyj: Great. I'm hoping we'll get some help from the OSC team in terms of telling us what they're wanting to change and why 16:17:02 smcginnis2: jgriffith DuncanT Lets make sure to have this as a topic at the summit and we can plan the road forward to get this all right. 16:17:16 Cross project session proposition? 16:17:22 +1 16:17:32 yea definitely a good idea 16:17:34 +1 16:17:35 DuncanT: One of our major consumers wants to move to this so I should be able to get support. 16:17:36 dulek: good idea 16:17:49 There's not a lot of them now: https://etherpad.openstack.org/p/newton-cross-project-sessions 16:17:50 <_alastor_> dulek: +1 16:17:54 Not cross project. We need a cinder plan 16:18:03 smcginnis2: +1 16:18:08 smcginnis2: +1 16:18:18 smcginnis2: +1 16:18:21 smcginnis2: +1 16:18:28 smcginnis2: Well, OSC is critical for every project - it would be hard to plan only for Cinder. 16:18:33 <_alastor_> cinder cross project plan? 16:18:43 <_alastor_> to be pedantic :) 16:18:43 https://etherpad.openstack.org/p/cinder-command-support 16:18:49 smcginnis2: well, if its a matter of changing to more standard things like 'properties' it might make sense to have the cross project environment for that discussion 16:19:07 That part is true. 16:19:10 Current status of OSC command can be found here: https://etherpad.openstack.org/p/cinder-command-support 16:19:17 specific to cinder only 16:19:19 #link https://etherpad.openstack.org/p/cinder-command-support 16:19:29 dulek: yep 16:19:41 But we already know cross project the plan is to standardize on osc. 16:19:42 patrickeast: +1 16:19:55 It seems OSC needs microversion support for cinder, nova, ironic, manila... 16:19:58 https://blueprints.launchpad.net/python-openstackclient/+spec/cinder-command-support 16:20:01 just catching up on this one... 16:20:02 smcginnis2: do we have any chance to change it? 16:20:03 i think the bug may have been hastily closed 16:20:04 And OSC does not have it ATM 16:20:05 We definitely need to talk within cinder about how and when. 16:20:08 There's been several people who've appeared to say that OSC is a bad idea with big tent