14:01:26 <jokke_> #startmeeting glance 14:01:30 <openstack> Meeting started Thu Jun 28 14:01:26 2018 UTC and is due to finish in 60 minutes. The chair is jokke_. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:01:31 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:01:33 <jokke_> #topic roll-call 14:01:34 <openstack> The meeting name has been set to 'glance' 14:01:36 <abhishekk> ./ 14:01:38 <McClymontS> o/ but in and out 14:01:39 <jokke_> o/ 14:01:46 <rosmaita> o/ 14:02:24 <jokke_> cool, we have almost full room :) 14:02:37 <jokke_> #topic updates 14:03:12 <smcginnis> o/ 14:03:29 <jokke_> 10mile high view of PTG schedule is out 14:03:34 <jokke_> #link http://lists.openstack.org/pipermail/openstack-dev/2018-June/131881.html 14:04:07 <abhishekk> great 14:05:13 <jokke_> That's all of non-release related updates I have atm. 14:05:16 <rosmaita> i need to start reading the ML again 14:05:41 <rosmaita> looks like foundation travel support closes tomorrow, if anyone needs it 14:05:48 <jokke_> oh, that 14:05:55 <jokke_> thanks rosmaita 14:05:57 <McClymontS> Cost is high for the PTG in general 14:06:02 <jokke_> yup! 14:06:10 <McClymontS> 599 is the base price right? 14:06:52 <jokke_> $399 is the registration price atm. 14:07:05 <McClymontS> Thought that was only early bird, not a bad deal 14:07:19 <jokke_> the $599 is after Aug 23rd/on-site 14:07:24 <McClymontS> Ah okay my bad 14:07:25 <abhishekk> early bird was $199 i guess 14:07:30 <jokke_> yeap 14:08:09 <jokke_> ok, lets move on 14:08:16 <jokke_> #topic release updates 14:08:38 <smcginnis> PTG does include two Summit passes, so if you are going to go to Summits that does make the price a little nicer. 14:08:47 <rosmaita> not much has changed since last week 14:09:07 <rosmaita> one failure on glance periodic jobs, but just once and this morning 14:09:21 <rosmaita> i will keep an eye on it over the next few days, info is on the agenda 14:09:30 <rosmaita> client and store are both green 14:09:34 <jokke_> rosmaita: ok, lets keep eye on it and dig deeper if it keeps failing 14:09:37 <jokke_> thnx 14:09:43 <rosmaita> #link https://etherpad.openstack.org/p/glance-team-meeting-agenda 14:10:03 <rosmaita> upcoming deadlines are on the agenda too 14:10:10 <rosmaita> same as last week except one week closer 14:10:18 <rosmaita> that's all from me 14:10:45 <jokke_> About the deadlines 14:11:10 <jokke_> lets try to aim releasing 0.25 glance_store at least week early 14:11:31 <rosmaita> ok, makes sense 14:11:40 <rosmaita> also that addresses the next item 14:11:43 <jokke_> so we have bit leverage _if_ for any reason the multi back-end stuff breaks the world 14:11:51 <jokke_> I don't see it happening but just in case 14:11:58 <abhishekk> makes sense 14:12:09 <jokke_> which means we need to get those patches reviewed 14:12:20 * rosmaita hides 14:12:46 <jokke_> I did my first round of reading them through instead of just eyeballing them yesterday 14:12:56 <abhishekk> I have tried to keep it simple (atleast I think so :P) 14:12:58 <jokke_> they look pretty good fwiw 14:13:23 <rosmaita> i did some eyeballing also, but need to buckle down and actually review 14:13:38 <rosmaita> i agree with jokke_ , code looks good at first glance 14:13:40 <jokke_> which leads us to next topic 14:13:50 <jokke_> #topic glance_store version 14:14:01 <jokke_> I guess I answered the question already ;) 14:14:07 <abhishekk> yeah 14:14:09 <rosmaita> yeah, so we have some flexibility with version numbering on the store 14:14:18 <jokke_> So the release containing the multi store stuff will be .25 14:14:19 <rosmaita> basically, we can do whatever we want 14:14:29 <abhishekk> and I kept it 0.25 in the doc so no need to change 14:14:37 <rosmaita> ok, i will update launchpad 14:14:45 <rosmaita> it makes sense to me 14:14:56 <rosmaita> unless smcginnis sees a problem? 14:15:06 <jokke_> I think once we get to removing the old configs, will be when we push that to 1.0.0 due to backwards incompatibility 14:15:58 <smcginnis> rosmaita: I'll double check, but sounds reasonable. 14:16:09 <jokke_> Unless we _want_ to release the multi back-end stuff stable in Stein as 1.0 14:16:33 <rosmaita> we can decide that in Stein 14:16:41 <smcginnis> We have some flexibility until we get to a 1.0 release, but I do think it's a good idea to get there soon. 14:16:49 <jokke_> but the changes are definitely big enough that 0.24.1 is not an option 14:17:10 <smcginnis> ++ 14:18:00 <jokke_> abhishekk: you had something under this topic as well 14:18:04 <abhishekk> ok, I have send a mail to driver maintainers to make them aware about our work 14:18:17 <abhishekk> #link http://lists.openstack.org/pipermail/openstack-dev/2018-June/131877.html 14:18:56 <abhishekk> I asked for help in testing cinder, vmware and sheepdog drivers or way to configure them using devstack so that I can do that testing 14:19:28 <abhishekk> I am 99% sure that it will not break though 14:19:44 <smcginnis> Cinder and sheepdog are probably reasonable for us to test, but we would probably need VMware to set up third party CI to get coverage there. 14:20:19 <abhishekk> smcginnis, yes 14:20:49 <abhishekk> that's it for this topic 14:21:36 <jokke_> So the VMWare driver maintainer has changed (need to update the wiki once I get into contact with the new person) I'll forward your mail directly to the new maintainer as well 14:21:52 <abhishekk> jokke_, thanks 14:22:05 <rosmaita> abhishekk: nice job on that email 14:22:33 <abhishekk> rosmaita, you only suggested that 14:22:49 <jokke_> ok 14:22:52 <rosmaita> yes, but nice implementation! 14:23:06 <jokke_> indeed 14:23:06 <abhishekk> thank you :d 14:23:15 <jokke_> #topic open discussion 14:23:30 <rosmaita> sorry, back to sean's point 14:23:45 <rosmaita> i think VMware already has 3rd party cli on glance 14:23:50 <rosmaita> or is it on glance_store 14:23:53 <rosmaita> it's on something 14:24:09 <rosmaita> so maybe won't be too hard for them to config multistore test 14:24:57 <abhishekk> ok 14:24:59 <jokke_> I think it's glance as we really did not have any functional tests in glance_store when that was built 14:25:37 <rosmaita> you are correct, it's on glance 14:25:45 <jokke_> which is one of the reasons I'd like to get the release out bit ahead of time as I'm not sure if they use release of glance_store or tip of the repo 14:26:05 <rosmaita> it's showing failure on https://review.openstack.org/#/c/574582/, btw 14:27:13 <jokke_> it's been on/off broken for a while ... I need to figure out if the new driver maintainer is also looking after that or if I need to find someone else stabilizing it 14:28:03 <jokke_> I hope/assume they have someone maintaining it as it does get fixed when there is unrelated breakages 14:28:20 <abhishekk> me too 14:28:40 <abhishekk> I have couple of questions related to hide old images 14:28:46 <jokke_> shoot 14:29:07 <abhishekk> 1. In specs it is mentioned that for queens update call with hidden attribute will rejected with 400 Bad request, but as per current code it is returning 409 conflict saying "Property 'hidden' does not exist", so do I change it to return 400? 14:29:23 <abhishekk> https://github.com/openstack/glance/blob/master/glance/api/v2/images.py#L245 14:29:44 <rosmaita> no, let's change the spec to reflect reality 14:29:53 <abhishekk> ok 14:30:01 <jokke_> what rosmaita just said 14:30:04 <rosmaita> though 409 seems weird for that 14:30:10 <rosmaita> but, too late 14:30:21 <jokke_> indeed 14:30:32 <abhishekk> ok 14:30:37 <abhishekk> 2. if hidden=true is passed to GET v2/images call then it should return only hidden images or it should return all images (hidden + other)? 14:30:41 <jokke_> well it's kind of justifiable as it's filtering 14:30:48 <McClymontS> I don't mind a 409 there 14:30:54 <jokke_> abhishekk: only hidden 14:31:07 <abhishekk> great, 14:31:18 <rosmaita> no objection from me 14:31:37 <jokke_> it's filtering option (I don't think we allow wild card filtering) 14:31:49 <abhishekk> 3. For show image call if it is hidden then is it required to show hidden=True in the output (this is not considered in the specs, and also I don't think it makes sense to show it) 14:32:13 <rosmaita> i think we do want to show it 14:32:24 <rosmaita> otherwise, how do you know it's hidden? 14:32:24 <jokke_> I mean we do not allow hidden=[*|any|all] type of filtering at which point it would be reasonable to return all images 14:33:15 <rosmaita> i mean, if someone does an update image call setting hidden=true, the response should show that 14:33:24 <jokke_> 3. yes 14:33:31 <rosmaita> so i think hidden=true on regular image-show is correct 14:33:40 <abhishekk> yes response shows it 14:33:55 <rosmaita> ok, maybe i misunderstood your question 14:34:12 <jokke_> in general we do not hide image properties from the user without good reason and I don't see good reason for hiding this 14:34:25 <abhishekk> but image show does not show it as of now so I need to make that change 14:35:13 <jokke_> abhishekk: ok, is it server not returning it or client not showing it? 14:35:16 <rosmaita> someone went through at some point and revised the schema so that "empty" core image properties have null JSON value if they are empty 14:35:40 <jokke_> and was that something you did on implementation or why does that happen? 14:36:13 <abhishekk> jokke_, server is returning this but client is not showing it 14:36:46 <jokke_> rosmaita: no, that is fine. If it's not set we can omit it. If it's set 'though we definitely should show it 14:36:59 <rosmaita> agreed 14:37:26 <abhishekk> ok, got it I will modify my patch accordingly 14:38:38 <jokke_> I'm gonna be completely hones with this. I'm not gonna spend time reviewing this feature before the multi back-end and v1 removals are in 14:38:41 <abhishekk> Also we need to document this right, means I also need to change docs and api-ref 14:39:14 <abhishekk> jokke_, makes sense (I will just keep it ready in spare time) 14:39:20 <jokke_> ohhh ... regarding documentation. I think we need to be clear about the Experimental status of multi back-end in the docs 14:39:41 <abhishekk> yeah, I have added a note in the docs mentioning the same 14:39:46 <rosmaita> good point, i missed that 14:40:05 <jokke_> also I did not check if the doc patch was depending the implementation, but lets not merge it before the feature lands 14:40:07 <abhishekk> https://review.openstack.org/#/c/576075/6/doc/source/admin/multistores.rst@19 14:40:28 <abhishekk> jokke_, its depending on the implementation 14:40:45 <jokke_> abhishekk: indeed, thanks 14:41:05 <rosmaita> i don't think the api-ref mentions experimental? 14:41:14 <abhishekk> :D 14:41:22 <abhishekk> rosmaita, no 14:41:27 <jokke_> that likely was where I was missing it 14:41:49 <rosmaita> we can dig up the old import patch and see how we did it for pike 14:41:53 <abhishekk> do we need to add it there as well, if yes please point out the exact position 14:41:58 <jokke_> also we need to do the experimental dance on the glance/api/versions.py on the glance patch of that feature which is not there 14:42:31 <abhishekk> for pike we have added it in the docs only I guess not in the api-ref 14:43:19 <jokke_> the api ref should pick it up from the versions.py, right? 14:43:29 <jokke_> or do we need to do it manually? 14:43:38 <rosmaita> no, has to be done manually 14:43:53 <jokke_> ok, then we have next problem 14:44:35 <abhishekk> what is it? 14:44:49 <rosmaita> we did mention experimental in pike api-ref 14:44:51 <jokke_> if we flag 2.7 EXPERIMENTAL and it's returned only if the multi back-end configs are used, what do we do about the hidden images if it lands in Rocky? 14:44:54 <rosmaita> http://git.openstack.org/cgit/openstack/glance/commit/?id=8005ee994cbbfb25575d13cd2d243e510c5e33d1 14:45:25 <jokke_> that is API change that needs version bump 14:45:56 <rosmaita> may have to land multistore last 14:45:57 <abhishekk> can we mark that experimental as well? 14:47:10 <abhishekk> rosmaita, thank you for the link, I will modify the api-ref based on it 14:47:20 <rosmaita> may have to do hidden + multihash = 2.7, multiple-backends = 2.8 EXP 14:47:22 <jokke_> rosmaita: that's what I was thinking as well, we might need to land the glance part of multi store at the very end and provide 2.7 stable 2.8 experimental in Rocky 14:47:37 <rosmaita> kind of sucks, but i don't see what else to do 14:47:53 <jokke_> in that note, Please do not merge the glance patch yet then 14:48:05 <rosmaita> ok 14:48:09 <jokke_> also please make sure we're ok with the state of the spec 14:48:10 <abhishekk> it will not get merge unless glance_store patches get merged 14:48:18 <abhishekk> it has a dependency 14:48:28 <jokke_> we should merge the spec amendment once we're in agreement of the implementation 14:48:35 <rosmaita> we need to merge the store patches soon, though 14:49:00 <rosmaita> but that's ok, we can have support in store and not use it yet 14:49:07 <rosmaita> (in glance) 14:49:13 <jokke_> yeah, we need to merge the store patches way earlier to be able to release the glance_store in 4 weeks or so 14:49:42 <abhishekk> jokke_, its better if we manage to release it before that 14:50:19 <jokke_> abhishekk: Oh I totally agree ... I would not like to see those patches hanging on review for next month you rebasing them 14:50:19 <abhishekk> I guess most of the people will be on leave in next week (mostly who stays in US) 14:50:28 <abhishekk> :D 14:50:49 <rosmaita> i'll just be taking one day 14:50:52 <jokke_> I know how painful it is 14:51:27 <abhishekk> actually, I have decided that I will never create a chain of dependent patches :D 14:51:42 <smcginnis> :) 14:51:42 <rosmaita> it is a world of pain 14:52:05 <abhishekk> (but if patch is to large then its actually gets difficult to review it in a one go) 14:52:14 <abhishekk> s/to/too 14:52:51 <abhishekk> So in document should I mention api version 2.8? 14:53:10 <abhishekk> we have only 8 minutes remaining 14:53:11 <jokke_> Not yet 14:53:15 <abhishekk> ok 14:53:28 <jokke_> we get there once we see what is landing this cycle and if we need it 14:54:00 <abhishekk> ack 14:54:28 <jokke_> ok, anything else? 14:54:44 <abhishekk> jokke_, I will shortly make store_description related change and upload new patches (may be after an hour) 14:54:53 <abhishekk> nothing 14:55:15 <jokke_> abhishekk: amazing, I don't mind if it's tomorrow, I know it's getting late for you 14:55:54 <abhishekk> jokke_, ok will try if possible otherwise will push it tomorrow 14:56:08 <jokke_> ok, clets close up. there is always #openstack-glance if something pops up 14:56:11 <jokke_> Thanks all! 14:56:13 <abhishekk> thank you all, have a nice time ahead 14:56:17 <jokke_> -c 14:56:18 <rosmaita> bye! 14:56:23 <jokke_> #endmeeting