14:00:13 #startmeeting glance 14:00:16 Meeting started Thu Mar 30 14:00:13 2017 UTC and is due to finish in 60 minutes. The chair is rosmaita. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:17 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:00:20 The meeting name has been set to 'glance' 14:00:24 o/ 14:00:29 \o 14:00:46 o/ 14:00:47 guess i don't need to actually #topic roll call 14:01:29 ok, hello everyone 14:01:38 #link https://etherpad.openstack.org/p/glance-team-meeting-agenda 14:01:45 light agenda today 14:01:57 my plan is to get through the first few items fairly quickly 14:02:15 and spend the remainder (except for open discussion) talking about image import 14:02:45 #topic updates - spec proposal freeze 14:02:56 the glance spec proposal freeze for pike is now in effect 14:03:20 #topic updates - openstack mission statement revision 14:03:33 #link https://review.openstack.org/447031 14:03:53 just in case you haven't seen the notice on the ML, there's a revision proposed to the openstack mission statement 14:03:59 to make it more "app-aware" 14:04:20 so, if you have strong feelings about this kind of thing, please take a look at that patch and comment 14:04:41 and that's all the updates 14:05:02 #topic specs review strategy 14:05:22 ok, the spec proposal deadline is passed, now we have to review and give feedback 14:05:42 what i'd like to do is to get each core to adopt at least one spec 14:05:49 -2 eevrything :D 14:05:54 but let's move this discussion to the spec list 14:06:06 #link https://etherpad.openstack.org/p/glance-pike-specs-review 14:06:21 jokke_ and his -2 happiness :) 14:06:27 but seriously, we should -2 anything that doesn't look reasonable 14:06:49 ok, so please open your browser to the etherpad link i posted above 14:07:10 hemanthm: I just try to promote more relaxed attitude towards -2 ... cure by exposure :) 14:07:33 these were all the open patches at the deadline 14:08:02 as i say in the comment about the first one, that's sort of a meta-spec -- it's a change to the template 14:08:43 we can argume more about the template change later, but for now, use it to guide thinking about those specs that may require database changes 14:09:45 ok 14:10:13 i volunteered dharini and stevelle to review 451560 14:10:20 np 14:10:35 it's a small change, and a code patch is up already 14:10:50 Thanks rosmaita 14:10:53 it's a full spec instead of lite-spec because it has possible api impact 14:11:06 i should stress *minimal* impact 14:11:11 minimal and legitimate 14:11:14 but i digress 14:11:41 450621 is about how to deal with removing a deprecated option 14:11:53 there are some weird interactions with the policy layer 14:12:09 but once we get the behavior sorted out, it shouldn't be bad code-wise 14:12:22 and dharinic already has a code patch up for some of it 14:12:40 206120 has been around for a long time 14:12:54 sort of a glance all-star list of past cores have reviewed it 14:13:29 priority-wise, i'm not so sure about it 14:13:40 does not mean that you need to agree with them 'though ;) 14:13:57 I disagree, jokke_ ! :) 14:14:18 well, i mentioned it primarily for anyone who wanted to take a stroll down memory lane 14:14:44 :) 14:14:57 i guess i should mention what the expectations are here 14:15:15 i'd like particular cores to be assigned to each spec 14:15:20 to give early feedback 14:15:37 Good idea rosmaita 14:15:39 then it's up to the proposer to be prompt in making revisions 14:16:02 but there's no pressure that these things *must* be approved 14:16:21 we just want early feedback so revisions can happen before the spec freeze 14:16:34 ok, moving on 14:16:58 442863 is awaiting response, so that one's ok for now 14:17:09 448680 is unreviewed 14:17:22 it's a small change though, and one operators need 14:17:36 but, it impacts the API 14:17:47 and there's an opportunity for some bikeshedding 14:18:08 key thing is we need something to supplement the current 'checksum' in glance 14:18:17 something cryptographically secure 14:18:41 jesse's proposal is to use a "multihash", a self-identifying hash 14:18:57 that way, you're not stuck with a particular algorithm forever 14:19:09 but some details need to be worked out 14:19:33 ok the rest are all lite specs 14:19:41 need some volunteers ... 14:20:00 key thing is quick feedback about whether a spec-lite is enough 14:20:12 I do think that the current hash security issue is critical enough for us to _need_ to figure out fix for it 14:20:19 i figure i am on enough of them already 14:20:24 even if we end up breaking the API again ;) 14:20:47 i think we are ok 14:20:59 i think we have a non-breaking way to do that 14:21:32 ok, i just volunteered to look at the lite specs 14:21:46 I'll look at them too 14:21:53 any non-cores, by the way, should look at specs, too 14:21:57 444530 is one jokke_ should look at since it was largely his idea 14:22:07 in terms of the expected solution 14:22:16 i don't mean to imply that only cores should review 14:22:25 i just want to make sure there's at least one core taking a look 14:22:47 so non-cores who want to become more involved, please review! your quality comments will be noticed 14:23:14 stevelle: k 14:23:32 ok, last thing 14:24:05 at the bottom there are some proposed specs that I -1'd back in ocata, with no response 14:24:18 I may just abandon them 14:24:22 with a nice note 14:24:38 i do want us to think about 373896, though 14:24:53 we still don't have a plan for what to do about OSSN-0075 14:24:55 373896 is something we need to discuss carefully, perhaps take it up in the Forum 14:25:17 :) 14:25:17 good point about the forum 14:25:54 #action propose something around OSSN-0075 for the forum or incorporate it into a current proposal 14:26:17 ok, thanks everyone ... early feedback is important, spec freeze is in 2 weeks 14:26:27 which brings us to the next item 14:26:39 #topic releaze czar report 14:26:53 I'll keep this quick 14:27:04 I'll be taking over the realse CPL duties from sigmavirus, who has done an amazing job for us last cycle. 14:27:12 To ensure we release Glance on time, we must stick to our priorities and respect community milestones. 14:27:15 To do the necessary for each milestone, in consultation with the PTL, I'll set a deadline. 14:27:24 And, there will be close to no exceptions to sneak something in too close to/after the deadline. (blame sigmavirus, he started playing tough and I have no option but to follow :P ) 14:27:28 I'll be using the following etherpad to track our milestones and priorities 14:27:31 https://etherpad.openstack.org/p/glance-pike-ptg-roadmap-prelim 14:27:55 rosmaita already talked about spec proposal freeze and spec freeze 14:28:05 Then, we have Pike-1 coming up soon in R-20, Apr 10 - Apr 14. 14:28:05 Anything that is intended for Pike-1 must be merged no later than Thursday next week, that is April 6th. 14:28:08 I understand that rosmaita wants to merge partial image download stuff. He may talk more about that later. 14:28:58 done 14:29:07 hemanthm: thank you 14:29:21 #topic would like to get into P-1 14:29:36 dharinic has been working on some partial download stuff 14:29:46 the 3 patches are noted on the agenda 14:30:03 they are mostly ready, but will be a review priority this week 14:30:12 the other priority is, of course, image import 14:30:41 and on that note ... 14:30:47 #topic image import refactor 14:30:58 jokke_ has had some patches up 14:31:16 for the backend "stage" 14:31:50 jokke_ has some technical issues he wants to discuss 14:31:57 but first, i do want to say 14:32:22 we need to keep in mind that the image import refactor has been bikeshedded extensively 14:32:43 at this point, we are stuck with the basic design 14:32:50 so please keep that in mind 14:33:06 also, while we don't want to merge bad code 14:33:32 we should have a "patch later" policy on the non-essential stuff (e.g., exact wording of help text) 14:33:57 that's it from me 14:34:01 dharinic: thanks for pointing out that state error in the workflow ... apart from that I've tried to figure out why the gate does not like the config option checks. Any insight for that would be great. Do we overwrite the config object in tests and I need to add it up there or do I have some (re)basing issue lingering there (the staging endpoint is just pep8 issue) 14:35:10 anyone have any ideas ^^ 14:36:30 I didn't look at the import patches yet, will comment if I know anything 14:36:56 thnx hemanthm 14:37:16 You are welcome jokke_. 14:37:24 hemanthm: maybe you can just look specifically at that config option check 14:37:34 sure, will do 14:37:40 i think you and dharinic are the most experienced with the config stuff 14:37:56 so this is the one in the chain https://review.openstack.org/#/c/443632/3 where the errors starts 14:38:26 http://logs.openstack.org/32/443632/4/check/gate-glance-python35/9f6f124/console.html#_2017-03-10_12_51_30_241611 14:38:43 we've got time, let's all look now 14:41:18 I think we need it here paste_deploy_opts 14:41:25 oops. https://github.com/openstack/glance/blob/master/glance/opts.py 14:42:05 actually not. Apologies. 14:42:23 jokke_ added the opts under an existing group. 14:42:59 yes that option should be in the DEFAULT 14:43:48 so I really think that patch is either not basing correctly or we overwrite the conf object (not the input to config parser) somewhere ... 'cause it should pick up the default even not specified 14:45:25 we don't need to import it, like show_multiple_locations do we? 14:47:20 it should come with the import from glance.common.config 14:47:32 but that said, I'm not convinced anymore 14:47:49 jokke_: what patch does that depend on? 14:48:14 hemanthm: the workflow patch that depends on the config patch 14:48:30 the config patch: https://review.openstack.org/#/c/391441/4 14:48:59 workflow patch: https://review.openstack.org/#/c/391442/ 14:49:00 so, I don't see that in my git log when I download that patch 14:49:01 depends on sha bb0acf1278b3e9cfe9ff4b0eba094a73f4534512 // WIP:Add api_image_import flow 42/391442/4 14:49:19 rather, child of 14:49:56 I mean, I don't see the patch that's adding the config opt in the git log. So, maybe the dependency chain is messed up? 14:50:01 or is it just me? 14:50:07 hemanthm: and that why I'm not convinced that it's based correctly in gerrit ... I tried to rebase it but the output was the same 14:50:35 and I'm not sure how it should work when you pull it down from gerrit 14:50:45 that's why I'm crying for help ;) 14:50:53 as I don't know anymore 14:51:07 it should pull down all commits it depends on 14:51:29 here's what I see. 14:51:29 http://paste.openstack.org/show/604865/ 14:51:34 ok ... so the dependency chain is definitely messed up in gerrit then 14:51:41 I would manually try adding dependency branch wise. 14:52:25 it looks right to me, this change is grandchild of the one that introduces config 14:52:58 hemanthm: https://review.openstack.org/gitweb?p=openstack/glance.git;a=log;h=9b2deb9a13f3a74cb694d7ad55eaeac57d393b94 14:53:03 stevelle: do you see all ancestors when you pull that change down? 14:55:38 hemanthm: yes, when I use checkout 14:55:52 git fetch https://git.openstack.org/openstack/glance refs/changes/32/443632/3 && git checkout FETCH_HEAD 14:56:22 I do not the see the ancestors when i download that patch 14:56:26 ok, so just downloading the review won't pull dependencies? 14:56:50 you say downloading the review, that's ambiguous to me :) 14:56:54 Whereas actually with just a git review -d, ideally we should be able to see it 14:56:57 git review -d 14:57:06 yeah, that's what I'd expect 14:57:16 me too 14:57:47 I did see the dependencies when I was working rolling upgrades chain 14:57:50 I can say that with confidence cos just yesterday i created a manual branch wise dependency and downloading (git review -d) the child, pulls the parent and appears on the git log 14:58:01 ok, let's continue in openstack-glance, it would be great if we could get jokke_ unstuck 14:58:05 I never use review -d, just out of habit 14:58:12 #topic open discussion 14:58:20 anyone got anything? 14:58:24 stevelle: yeah 14:58:28 -st 14:58:40 rosmaita: o/ 14:58:41 I'll be in Boston ... anyone else got confirmed yet? 14:58:44 will try to provide feedback on the doc for release stuff later today 14:58:58 thanks 14:59:01 I was thinking about organizing an on-boarding event every two weeks or so 14:59:02 Just a question about Glance v1 deprecation plans: Is that something that's going to be removed during Pike? 14:59:22 blancos: no 14:59:27 blancos: no, but it hasn't been announced yet 14:59:40 jokke_ rosmaita Okay, thank you 14:59:40 just realistically not possible 15:00:02 ok, got to stop 15:00:06 #endmeeting