14:00:11 <abhishekk> #startmeeting glance
14:00:12 <openstack> Meeting started Thu Dec 12 14:00:11 2019 UTC and is due to finish in 60 minutes.  The chair is abhishekk. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:00:13 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:00:15 <abhishekk> #topic roll call
14:00:16 <openstack> The meeting name has been set to 'glance'
14:00:24 <abhishekk> #link https://etherpad.openstack.org/p/glance-team-meeting-agenda
14:00:27 <abhishekk> o/
14:00:30 <rosmaita> o/
14:01:01 <abhishekk> lets wait 2-3 minutes for others to join
14:01:33 <rosmaita> ok
14:01:47 * tosky lurking
14:02:00 <abhishekk> tosky, o/
14:03:03 <abhishekk> lets start
14:03:06 <abhishekk> #topic Updates
14:03:31 <jokke_> o/
14:03:35 <abhishekk> This is the M1 release week and as decided in last week we are skipping M1 for glance
14:03:40 <abhishekk> jokke_, o/
14:03:54 <rosmaita> \o/
14:04:09 <rosmaita> (we skipped for cinder, too)
14:04:38 <abhishekk> We didn't have anything concrete merged in glance since T release, all the work targeted for M1 is now expected to be done before M2
14:04:49 <abhishekk> rosmaita, :D
14:05:08 <abhishekk> I have priorities patch uploaded, kindly review the same
14:05:10 <rosmaita> i don't think it would be illegal to release a milestone-1.5
14:05:19 <abhishekk> #link https://review.opendev.org/#/c/696017/
14:05:31 <abhishekk> haha
14:06:18 <rosmaita> i really like "Permanent solution for Subunit parser error"
14:06:23 <abhishekk> We need reviews on specs, because I can not approve it unless it is reviewed by all the cores
14:07:00 <abhishekk> rosmaita, yeah, I am working on the possibilities, started with removing registry test cases
14:07:26 <abhishekk> Moving ahead
14:07:40 <abhishekk> #topic release/periodic jobs update
14:07:55 <abhishekk> Sean has proposed glance-store release patch for M1
14:08:15 <abhishekk> I have voted -1 as we didn't merged anything in glance store either
14:08:23 <jokke_> ++
14:08:43 <jokke_> we have bunch of stuff under review 'though
14:08:45 <abhishekk> I have added comment that if we want to release it then we should merged sheepdog driver removal patch before
14:08:52 <abhishekk> jokke_, yes
14:08:55 <rosmaita> abhishekk: ++
14:09:29 <abhishekk> So, IMO we should skip this release and revisit the same around M2
14:09:46 <jokke_> yep, good plan
14:09:57 <abhishekk> cool
14:10:00 <rosmaita> smcginnis: dyk the current thinking about removing py2 support from libraries?
14:10:19 <rosmaita> oops, just realized that he isn't here
14:10:26 <rosmaita> thought i saw him come in
14:10:36 <abhishekk> rosmaita, yeah, he has headed out
14:10:57 <abhishekk> Periodic jobs all green, from last 3 weeks
14:11:09 <rosmaita> \o/
14:11:12 <jokke_> that is good news
14:11:15 <rosmaita> those are all py3 now, right?
14:11:24 <abhishekk> rosmaita, yes
14:11:28 <rosmaita> cool
14:11:48 <abhishekk> next topic
14:11:58 <abhishekk> #topic Multi-store import plugins
14:12:16 <abhishekk> We have two specs up for new import related work
14:12:27 <abhishekk> #link
14:12:27 <abhishekk> https://review.opendev.org/#/c/667132/
14:12:47 <jokke_> Just wanted to throw this out there. The specs and corresponding WIPs are up
14:13:06 <jokke_> So this is something we can potentially move quite quickly off our plates
14:13:14 <abhishekk> jokke_, ++
14:13:20 <abhishekk> #link https://review.opendev.org/669201
14:13:39 <abhishekk> This link is for Import image in multiple stores
14:13:50 <abhishekk> #link https://review.opendev.org/694724
14:13:51 <rosmaita> ok, i will try to get through the specs at least today
14:14:05 <abhishekk> This link is for Copying existing images in multiple stores
14:14:19 <abhishekk> rosmaita, cool, it will really help to get going
14:14:55 <abhishekk> Related code is also up for reference and is in good shape
14:15:26 <abhishekk> moving ahead
14:15:29 <abhishekk> #topic Delete from single store
14:15:36 <abhishekk> jokke_, stage is yours :D
14:16:06 <jokke_> Another of mine for attention. So wrote quick spec as the approach I took would introduce new API endpoint
14:16:25 <jokke_> I have skeleton code up as well for it
14:16:58 <abhishekk> I went through the specs and it is written based on our discussion during PTG
14:17:07 <abhishekk> so thank you very much :D
14:17:09 <jokke_> #link https://review.opendev.org/#/c/698018/
14:17:26 <jokke_> The spec is failing test because we do not have index merged yet
14:17:38 <abhishekk> yes
14:17:40 <jokke_> #link https://review.opendev.org/#/c/698049/
14:17:50 <jokke_> that's for the code part
14:18:23 <abhishekk> cool
14:18:38 <jokke_> I pulled the code together quick to kind of POC it based on the proposed spec. If we get to agreement of the approach I start writing the rest of it
14:19:13 <abhishekk> rosmaita, Put your glance reviewers hat as early as possible :P
14:19:14 <jokke_> there is still all testing (surprise, it's me), docs and some of the verifications missing
14:19:28 <jokke_> but that should be quick to finish after we're in agreement of the approach
14:19:45 <abhishekk> jokke_, I like the idea, specially we are not going down each of the onion layer for the same
14:20:30 <jokke_> but TL;DR basically introducing "v2/stores/{store_id}/{image_id}" endpoint that accepts only DELETE method
14:21:14 <jokke_> just to make sure people don't accdentally kill the whole image because we would hve just appended the store_id to the current call
14:21:23 <abhishekk> only concern is if user has delete right and don't have right to remove the location then it will now allow deleting the location (I guess this is happening at the moment as well)
14:21:27 <jokke_> but that's the other way to approach this
14:22:35 <rosmaita> keeping in mind that i am way behind on my specs reading ...
14:22:36 <jokke_> If I recall correctly, the policy is checked on the image update, but that's something I'd (or someone else) need to test and verify
14:22:52 <rosmaita> what is the api call to upload an image to a particular store?
14:23:06 <abhishekk> its put
14:23:06 <jokke_> the image delete calls the location deletes in the background so it should be affected as well, but I'm not 100%
14:23:33 <abhishekk> location delete call has policy check
14:23:35 <rosmaita> what does the path look like?
14:23:48 <jokke_> rosmaita: in the upload it's the same call, you just add the store as header
14:23:56 <abhishekk> correct
14:24:12 <abhishekk> in case of  new import workflow, its POST
14:24:13 <jokke_> as the body is occupied by the image data
14:24:29 <abhishekk> in case of traditional create call it will be PUT
14:24:39 <abhishekk> for both the calls we are passing store as a header
14:24:41 <jokke_> the multi-store imports it's gonna be part of the body of the import call
14:24:43 <rosmaita> that works for an existing image? i would do multiple PUT /v2/images/uuid/file with different headers?
14:24:58 <jokke_> rosmaita: nope
14:25:06 <rosmaita> ok ,that's waht i thought
14:25:16 <rosmaita> that's why you need the copy to multistore spec
14:25:21 <jokke_> rosmaita: you upload the image once, ever. We don't allow uploading to active images
14:25:39 <abhishekk> rosmaita, yes
14:26:12 <rosmaita> so the alternative was DELETE /v2/images/uuid/store_id ?
14:27:18 <jokke_> rosmaita: yes, so the import to mutiple stores allows to populate multiple stores upong creation via import and the copy job is so far only that we allow for active image and it will be glance handling the data transfers. That should minimize the risk of anyone sneaking data in that is modified
14:27:21 <abhishekk> rosmaita, that is the alternative but our router mapping is horrible and as delete image have same mapping it will always map to delete image instead of our new method
14:27:58 <rosmaita> jokke_: makes sense
14:28:02 <rosmaita> abhishekk: sorry to hear that
14:28:10 <jokke_> rosmaita: yeah, that's the baseline either what you wrote or /v2/images/uuid/store/store_id
14:28:52 <rosmaita> ok, thanks
14:29:02 <jokke_> abhishekk: rosmaita: correct, so that approach would need us to do the logic in our image deletion code rather than have it's own isolated coe path
14:29:22 <rosmaita> your proposal is slightly asymmetric, but probably ok
14:29:30 <abhishekk> jokke_, right
14:29:48 <rosmaita> kind of makes sense to think of image data as a resource owned by a store
14:30:01 <jokke_> oh there is the 3rd option as well but I think I didn't even write it into the spec
14:30:24 <jokke_> we just tell everyone to use the vulnerable locations api and do it via image-update
14:30:39 <abhishekk> :d
14:31:01 <rosmaita> PATCH seemed like such a cool idea back in 2011
14:31:08 <jokke_> ikr
14:31:39 <jokke_> I'm happy we got over that thought fairly quickly :P
14:31:46 <rosmaita> :D
14:31:48 <abhishekk> +1
14:32:06 <abhishekk> lets move to next topic
14:32:20 <jokke_> ++
14:32:36 <abhishekk> all cores, kindly review all the specs as early as possible
14:32:55 <abhishekk> #topic Native SSL removal
14:33:02 <abhishekk> jokke_, again you
14:33:42 <jokke_> So as agreed at Train we wanted to get rid of py27 support to whack the SSL and classify ourselves as py3 compatible
14:34:48 <jokke_> py27 seems to be finally beated to dead and I did propose 3 patches for this
14:34:53 <jokke_> #link https://review.opendev.org/#/c/697969/
14:35:54 <jokke_> is the key, it depends on the patch that just removes the py27 classifies without taking any side of what we're going to do with py27 testing and then on top of that is py3 classifier addition doing the same
14:36:12 <abhishekk> Should we get this in, release glance M1 so that people can test it for regression?
14:36:37 <jokke_> That i one option. the 3 patch chain should be good to go
14:37:16 <abhishekk> +1 from me on this approach
14:38:34 <rosmaita> sounds ok to me
14:39:11 <abhishekk> I will put a release patch on Monday if all these patches merged before weekend
14:39:50 <jokke_> I could have smashed them into one review but I didn't want to clean up the rebases if we happened to get those existing testing patches merged before that also changes the classifiers
14:40:17 <jokke_> the rebase would be clean this way
14:40:23 <abhishekk> agree
14:41:15 <abhishekk> so as jokke_ is owner, rosmaita you and me need to review the patches :D
14:41:25 <abhishekk> or either smcginnis
14:41:32 <rosmaita> i guess this is higher priority than specs?
14:41:45 <abhishekk> rosmaita, yes
14:41:46 <jokke_> If we want to get M1 tagged
14:42:10 <rosmaita> ok, i will hit these first and do as many specs as i can this afternoon
14:42:30 <abhishekk> ty rosmaita \o
14:42:30 <rosmaita> i think the py3 M1 is a good idea
14:42:39 <jokke_> but I agree, the earlier we tag this the easier it is to find out if I broke something in the process
14:42:48 <abhishekk> :D
14:43:33 <abhishekk> Moving ahead
14:43:40 <abhishekk> #topic Open Discussion
14:43:42 <jokke_> I can check I think glance_store has py3 classifiers already
14:44:02 <jokke_> if not might be worth of getting py3 M1 out of that as well
14:44:38 <jokke_> yes we do so no worries there it's just the service itself
14:44:38 <abhishekk> it has
14:45:05 <jokke_> abhishekk: I was looking the registry test removal patch you posted
14:45:18 <abhishekk> ok
14:45:48 <jokke_> I have couple of questions it raised and I was trying to figure out why that task test fails just on the minute till the meeting
14:46:06 <jokke_> I will have my comments in the review right after this meeting
14:46:11 <abhishekk> jokke_, have you figured that out?
14:46:14 <abhishekk> cool
14:46:22 <jokke_> nope ... I did not figure it out yet
14:46:38 <jokke_> I need to run that tests and try to figure out where it actually fails
14:46:49 <jokke_> hopefully it throws stack trace or something
14:47:12 <abhishekk> it tries to connect to sql database on image.save call and didn't find the connection
14:47:35 <abhishekk> and then it goes to revert to try to delete the image and it fails there as well :D
14:47:53 <jokke_> ok, then I have good idea where that happens
14:48:03 <abhishekk> great
14:48:20 <abhishekk> so next week is Christmas and I will be on leave on 26 and 27
14:48:39 <jokke_> we need to probably mock the task_repo for it
14:49:07 <abhishekk> is anyone from you available for the meeting or should we skip next weeks meeting?
14:49:17 <abhishekk> jokke_, ack, I will try the same
14:49:27 <rosmaita> next week is dec 19 ?
14:49:28 <jokke_> the gateway providing the repos does not take the simple database stuff into consideration
14:49:42 <jokke_> what really baffles me is how in earth it works now
14:49:54 <abhishekk> me neither
14:49:56 <jokke_> abhishekk: rosmaita: indeed. You're week ahead still
14:50:10 <jokke_> abhishekk: XMas in in 2 week :P
14:50:13 <abhishekk> sorry :P
14:50:56 <abhishekk> I thought we decide it now :d
14:50:58 <rosmaita> so next week, yes, but week after, no
14:51:37 <abhishekk> jokke_, rosmaita I guess we should skip Christmas week meeting which is on 26
14:51:38 <jokke_> So I will pretty much honour company shutdown. Will be at least whole christmas week offline (Telegram might reach depending of my reception) and will return early Jan
14:51:54 <jokke_> yeah I definitely won't be around on 26th
14:51:55 <abhishekk> I will be there in new years week to host the meeting
14:52:05 <rosmaita> yes, i will be offline from 24 dec until 2 jan
14:52:38 <abhishekk> cool, I will send the mail on ML for the awareness
14:52:59 <jokke_> I'll be Flying next Tue so I have bit early evening tuesday, but will work from Finland the rest of the weeks still
14:53:18 <abhishekk> safe travels
14:53:18 <jokke_> rest of the next week even
14:53:20 <davee_> I have sad news.  Funding was pulled and my contract was canceled so today is probably my last day with supporting this project until aft I can find new job
14:53:34 <jokke_> davee_: crap :(
14:53:43 <abhishekk> davee_, sorry to hear this
14:53:47 <abhishekk> :(
14:53:49 <jokke_> Sorry to hear. What a "great" christmas present
14:53:54 <davee_> not as sad as I was
14:54:22 <rosmaita> davee_: that is a real bummer, i wish you much luck
14:55:02 <abhishekk> +1
14:55:17 <abhishekk> anything else?
14:55:18 <davee_> I have a few leads, so hopefully I will bounce back quickly for new year
14:55:30 <abhishekk> davee_, great
14:55:38 <rosmaita> good to hear
14:55:47 <davee_> Just wanted to say I have enjoyed working with you all
14:55:57 <rosmaita> the feeling is mutual
14:55:59 <abhishekk> likewise :D
14:56:04 <jokke_> thanks for the heads up, likewise
14:56:29 <abhishekk> wrapping up guys
14:56:37 <abhishekk> Merry Christmas in advance
14:56:46 <abhishekk> crap
14:56:54 <abhishekk> we are meeting next week :D
14:56:58 <jokke_> yes :D
14:57:09 <jokke_> abhishekk: get some sleep every now and then :P
14:57:15 <abhishekk> Thank you all
14:57:15 <jokke_> Thanks all!
14:57:30 <rosmaita> bye
14:57:35 <abhishekk> jokke_, yes, thats why I was dreaming of chirstmas leave :D
14:57:56 <abhishekk> #endmeeting