14:00:16 <jokke_> #startmeeting glance
14:00:21 <openstack> Meeting started Thu Jul 25 14:00:16 2019 UTC and is due to finish in 60 minutes.  The chair is jokke_. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:00:22 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:00:24 <jokke_> #topic roll-call
14:00:25 <openstack> The meeting name has been set to 'glance'
14:00:27 <jokke_> o/
14:00:37 <jokke_> #link https://etherpad.openstack.org/p/glance-team-meeting-agenda
14:00:38 <rosmaita> o/
14:00:42 <rosmaita> welcome back
14:00:52 <jokke_> tyty
14:01:04 <zhengMa> hi
14:01:18 <abhishekk> o/
14:01:43 <jokke_> #topic updates
14:01:50 <davee_> o/
14:01:58 <jokke_> Just wanted to highlight the good work that has been going on.
14:02:22 <jokke_> Did some reviews today while catching up and happy to  see stuff moving forward, so thanks all :D
14:03:08 <jokke_> Lets take a quick rush and try to get stuff ready to go so we can get m2 release out next week for people to test the glance store 1.0.0 and stabilized glance-store
14:03:15 <jokke_> multi-store even
14:03:45 <jokke_> on that note
14:03:52 <jokke_> #topic release-updates
14:04:43 <abhishekk> hi
14:05:06 <abhishekk> So we are having Milestone release 2 this weeks
14:05:35 <abhishekk> thanks to Brian and Sean for reviewing the patches
14:05:47 <rosmaita> :)
14:06:03 <abhishekk> We have lined up glance-store v 1.0.0. lined up
14:06:18 <abhishekk> #link https://review.opendev.org/#/c/672627/1
14:06:26 <abhishekk> above is the release patch for the store
14:06:37 <abhishekk> jokke_, and rosmaita kindly have a look
14:06:49 <abhishekk> #link https://etherpad.openstack.org/p/Glance-Train-MileStone-2-Release-Plan
14:07:04 <jokke_> will do
14:07:14 <abhishekk> This etherpad will help us to track down the glance changes which are essential before M2 release
14:07:35 <rosmaita> so do we expect glance_store 1.0.0 to break anything?
14:07:38 <abhishekk> This is on release front, any word jokke_ ?
14:07:51 <abhishekk> rosmaita, I don't think so
14:07:53 <rosmaita> looks like it def breaks our functional tests
14:08:03 <rosmaita> looks like the unit tests are ok
14:08:09 <jokke_> rosmaita: we are expecting the backend->store breaking so we will need to get that change in glance asap
14:08:29 <rosmaita> yeah, but my point is when we release, upper-constraints will be changed
14:08:43 <abhishekk> rosmaita, are you looking at tips jobs?
14:08:49 <rosmaita> yes
14:08:56 <rosmaita> https://review.opendev.org/#/c/671707/3
14:09:20 <abhishekk> those were failed because in morning the git or opendev.org was down for couple of hours
14:09:24 <rosmaita> take a look at the experimental results on that
14:09:48 <abhishekk> oops
14:09:55 <rosmaita> http://logs.openstack.org/07/671707/3/experimental/glance-tox-functional-glance_store-tips/2651b8a/testr_results.html.gz
14:11:04 <jokke_> thanks rosmaita for pointing it out, we need to check that before we pull the trigger on the release
14:11:29 <rosmaita> np, will just take some careful coordination
14:11:53 <jokke_> yup, the naming change I know will break stuff but haven't been looking for other things yet
14:12:58 <jokke_> abhishekk: you had something on the periodic jobs as well
14:13:12 <abhishekk> jokke_, yeah
14:13:26 <abhishekk> we have 3-4 failures due to parser error during this weel
14:13:31 <abhishekk> s/weel/seek
14:13:36 <abhishekk> opps
14:13:39 <jokke_> :P
14:13:49 <abhishekk> Looks like I need some sleep
14:13:59 <jokke_> So any idea what's triggering those. Is it still the same issue or has something else changed?
14:14:11 <abhishekk> no still the same issue
14:14:27 <abhishekk> large size of log messages are causing it
14:14:37 <jokke_> have we increased logging somewhere?
14:14:45 <abhishekk> nope
14:14:59 <rosmaita> jokke: are you going to -1 the glance_store release patch?
14:16:00 <abhishekk> rosmaita, any Idea how can I locally test this failures
14:16:26 <jokke_> just did
14:16:30 <abhishekk> I guess I need to copy glance-store in virtual environment and then run the tests against those, right?
14:17:00 <jokke_> abhishekk: locally you would nbeed to change the requirements to use the git link instead of just package name iirc
14:17:36 <abhishekk> jokke_, ack
14:17:41 <rosmaita> abhishekk: either that, or we skip-test the failures
14:17:43 <rosmaita> merge that
14:17:47 <rosmaita> release glance_store
14:17:54 <rosmaita> then you fix the failures
14:18:01 <rosmaita> we remove the skip-test
14:18:06 <abhishekk> make sense
14:18:07 <rosmaita> and hooray
14:18:21 <rosmaita> jokke_: that sound ok?
14:18:41 <jokke_> lets just check what is blowing first to make sure we don't need to release 1.0.1 tomorrow just because it was something we had to fix in glance_store
14:18:59 <abhishekk> I will give it a try tonight
14:19:18 <jokke_> if it's just the glance stuff we know needs to be changed, skipping the tests or queueing the fix first thing to merge would be fine
14:19:32 <jokke_> I'll have a look on the logs after the meeting
14:19:49 <davee_> abhishekk: after you get some sleep I hope
14:20:21 <abhishekk> davee_, If I sleep now in my dream I will see broken gate :D
14:20:29 <rosmaita> sounds good ... do we have a deadline to release glance_store ?
14:20:54 <davee_> hmm, I have dreamed and written code in my sleep that worked the next day
14:20:55 <jokke_> rosmaita: kind of today, if it needs to be pushed because it's broken then we push it to next week
14:21:03 <abhishekk> :D
14:21:44 <jokke_> ok, lets move on
14:21:59 <jokke_> #topic wording on multi-store documentation
14:23:01 <jokke_> I just wanted to highlight that I think we need to be stronger than the current "not recommend" on using same path for task_work dir, staging and possible filestore
14:23:20 <jokke_> #link https://review.opendev.org/#/c/611253/16/doc/source/admin/multistores.rst
14:23:49 <abhishekk> rosmaita, ^^^^ you are the one who can propose this :D
14:24:09 <rosmaita> ok
14:24:29 <jokke_> my reasoning is that _if_ someone uses the same path, it will give same prefix for multiple store id's and might end up to the autodiscovery tool assigning "os_glance_*" store id to the image
14:25:04 <rosmaita> yeah, that will definitely happen
14:25:06 <abhishekk> if someone uses same path for staging and fs store then it will be mess
14:25:14 <jokke_> there should not be often those cases, but I don't like the idea of possibility
14:25:29 <rosmaita> we probably need to put some kind of checking in the code
14:25:36 <rosmaita> (eventually)
14:25:37 <jokke_> so "not support" or blunt "Don't do it!" would be mnore of my liking :)
14:26:29 <rosmaita> yes, a warning box and do not do it
14:26:40 <abhishekk> sounds good
14:27:03 <jokke_> thanks, that was all from me ... just wanted to take couple of mins for brainstorming how we word that doc
14:27:20 <abhishekk> it will be kind of difficult to add that check, and if we want to add that check that needs to be added in store
14:27:24 <jokke_> rosmaita: if you have something in mind, and have time to push new rev, that would be amazing
14:27:39 <rosmaita> yeah, i will push somethinig in a few min
14:27:45 <jokke_> <3
14:27:52 <jokke_> #topic open discussion
14:27:59 <rosmaita> abhishekk: yes, probably when the map is populated
14:28:23 <rosmaita> i will file a bug so we dont forget
14:28:25 <abhishekk> rosmaita, right
14:28:56 <rosmaita> actually, should probably be a known issue in the glance_store release notes?
14:29:10 <rosmaita> i mean, additionally?
14:29:49 <abhishekk> rosmaita, ok for me
14:29:51 <jokke_> _if_ we need to change something else before the release. It's really glance issue, not glance_store issue 'though
14:30:13 <rosmaita> well, i guess that's right
14:30:37 <rosmaita> anyway, we are the only consumer of glance_store multibackend right now
14:30:43 <jokke_> yup
14:31:29 <rosmaita> i will add it to that glance releasenote on https://review.opendev.org/#/c/611253/16 , though
14:31:42 <jokke_> that sounds like a good plan
14:31:52 <rosmaita> cool
14:32:04 <abhishekk> great
14:32:24 <jokke_> we can always remove it from reno if we come up with bulletproof fix before actual release
14:32:46 <rosmaita> right
14:33:16 <rosmaita> we may have to do it on the glance side before we create the store
14:33:24 <rosmaita> we can crawl the config and check
14:34:13 <abhishekk> we can do it in glance after store is initialized
14:34:37 <jokke_> or we can just prevent the func using anything starting with os_glance_
14:35:10 <rosmaita> ok, well, we are not short on ideas
14:35:17 <jokke_> but it doesn't need to be today's decision. Lets try to get the 1.0.0 release out and think it after so we get something out there for people to test
14:35:19 <rosmaita> i'll file the bug for glance and not the store
14:35:24 <jokke_> ++
14:37:22 <jokke_> anything else?
14:37:29 <zhengMa> Hi, I have purposed a spec  lite: - Add 'compressed' option to container_format - https://review.opendev.org/#/c/670454/
14:37:38 <zhengMa> It has been discussed on this meeting last week and I'm expecting more reviews
14:37:54 <abhishekk> I found the problem in functional failures
14:38:26 <abhishekk> if we add skip test and then merge https://review.opendev.org/#/c/658962 this patch then it will solve the issue
14:38:57 <abhishekk> https://review.opendev.org/#/c/658962/6/glance/api/v2/images.py look at line no #1033 to #1037
14:39:03 <rosmaita> abhishekk: cool!
14:39:06 <jokke_> so that is glance_side failure we're queued to fix anyways?
14:39:23 <abhishekk> yes
14:40:25 <jokke_> cool so not blocking the release then
14:40:35 <jokke_> was there anything else we were not aware of?
14:41:02 <abhishekk> nope
14:41:11 <rosmaita> zhengMa: i think your spec looks good, did LiangFang say in the cinder meeting yesterday that you will test to see what happens if you try to boot an image with container_format == compressed in nova?
14:41:16 <abhishekk> so should I go ahead and add a patch for skipping the tests?
14:41:21 <jokke_> great, will swap my -1 to +1 on the release patch then
14:41:52 <abhishekk> I guess it will not break the release at it will just runs the unit tests
14:42:11 <jokke_> zhengMa: yes, sorry I've been on PTO and just catching up and bit tangled with the release. I promise I will read through that spec for you
14:42:32 <zhengMa> jokke_: thank you !
14:42:38 <rosmaita> yeah, what we have to watch for is as soon as upper-constraints is updated, glance will block the gate until https://review.opendev.org/#/c/658962 is merged
14:42:51 <jokke_> mhm
14:42:57 <zhengMa> rosmaita: yes, he told me it, and we have been testing it
14:43:04 <jokke_> lets propose skip to that test and merge it right away
14:43:15 <rosmaita> zhengMa: great!
14:43:18 <abhishekk> jokke_, ack
14:43:25 <rosmaita> jokke_: sounds good
14:43:25 <jokke_> so we have bit time to go through that patch and remove the skip in it
14:43:40 <rosmaita> jokke_: ++
14:45:11 <abhishekk> will push that patch in some time
14:45:46 <jokke_> great, I'll see that having +W before swapping my release -1 to +1
14:45:51 <jokke_> anything else?
14:46:29 <davee_> I have nothing else
14:46:33 <abhishekk> we have @skip_if_disbaled decorator
14:46:59 <abhishekk> how about moving those tests on blacklisted text file created by rosmaita
14:47:34 <jokke_> there should be skip decorator you can give bug or comment that shows up on the test run
14:47:37 <zhengMa> rosmaita: nova seems unable to handle this new format now
14:47:38 <rosmaita> i think there's an unconditional skip decorator too
14:47:57 <rosmaita> zhengMa: what happens? what is the response from nova?
14:47:59 <jokke_> just put message on it that it was disabled for working around 1.0.0 release
14:48:04 <abhishekk> am not able to found it anywhere
14:48:38 <zhengMa> rosmaita: well currently no proper response and it just failed accessing the VM's console
14:49:11 <jokke_> https://docs.openstack.org/tempest/latest/library/decorators.html
14:49:13 <rosmaita> zhengMa: ok, we will need to open a discussion with nova about this
14:49:33 <zhengMa> rosmaita: right in nova, thanks
14:50:02 <jokke_> we can use @skip_because if we file a bug for it
14:50:18 <abhishekk> there is skipTest
14:50:28 <jokke_> that should work
14:50:36 <jokke_> just put todo or fixme comment on it
14:50:41 <rosmaita> zhengMa: what kind of image did you use?
14:51:14 <abhishekk> ack
14:51:55 <jokke_> zhengMa: I don't think we want to have format that nova and cinder can't handle, so we likely need to address that with nova first like rosmaita said
14:52:08 <zhengMa> rosmaita: container_format: compressed disk_format: either raw or others
14:52:19 <rosmaita> jokke_: cinder will be able to handle this
14:52:48 <jokke_> rosmaita: (nova and cider) condition fails if either fails :D
14:53:15 <jokke_> just generalizing for further discussion that I would feel the same with cinder failing
14:53:16 <rosmaita> zhengMa: the real test would be to take a raw image that does boot correctly with Nova (a VM you can connect to), and then upload a gzipped version of that raw image, and see what happens when you boot from nova
14:54:07 <rosmaita> jokke_: yes, we just want to make sure that nova either can handle the format, or knows to reject it as unknown container_format
14:54:16 <jokke_> yup
14:54:25 <jokke_> getting something meaningful out of it
14:54:27 <zhengMa> rosmaita: that is what we have done, and nova failed while directly booting from that gzipped image
14:54:48 <jokke_> and that we have documented the type somewhere not being nova bootable if that's the case
14:55:19 <rosmaita> zhengMa: ok. and the failure was that the VM booted, but was unusable?
14:55:26 <jokke_> zhengMa: have you discussed with nova if they are interested to support that?
14:55:43 <zhengMa> jokke_: not yet
14:55:47 * jokke_ sees edge benefit for this
14:56:01 <jokke_> less bits to move to the edge site
14:56:12 <zhengMa> rosmaita: right! it boots up but unusable
14:56:33 <zhengMa> because nova just takes it as the raw image
14:56:40 <jokke_> zhengMa: yeah ^^ is something that needs to be addressed before we implement this
14:56:55 <zhengMa> so I think maybe a little patch in nova would address it
14:57:25 <jokke_> I think there is usecases where lots of ops would be happy about this
14:57:30 <jokke_> lets just do it right :D
14:58:09 <zhengMa> thank you:)
14:58:30 <jokke_> last 2min
14:58:40 <rosmaita> zhengMa: can you send a message to the mailing list about this?
14:59:00 <rosmaita> we can talk in glance channel after the meeting
14:59:12 <zhengMa> rosmaita: ok!
14:59:16 <rosmaita> thanks
14:59:24 <jokke_> ok, we're running out of time
14:59:26 <jokke_> Thanks all!
14:59:33 <rosmaita> bye!
14:59:36 <abhishekk> jokke_, rosmaita https://review.opendev.org/672752
14:59:36 <zhengMa> thanks! bye!
14:59:41 <abhishekk> thank you all
14:59:45 <jokke_> #endmeeting