Thursday, 2023-04-20

pdeore#startmeeting glance13:59
opendevmeetMeeting started Thu Apr 20 13:59:36 2023 UTC and is due to finish in 60 minutes.  The chair is pdeore. Information about MeetBot at http://wiki.debian.org/MeetBot.13:59
opendevmeetUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.13:59
opendevmeetThe meeting name has been set to 'glance'13:59
pdeore#topic roll call13:59
pdeore#link https://etherpad.openstack.org/p/glance-team-meeting-agenda13:59
pdeoreo/13:59
mrjoshio/13:59
pdeorewe have very short agenda for today, let's wait for few mins for others to join14:00
dansmitho/14:00
pdeorelet's start, others may join in between :)14:01
pdeore#topic release/periodic jobs updates14:01
pdeoreM1 is 3 weeks away from now14:02
pdeorePeriodic job all green except failure on some functional py38 tests may be due to castellan version conflict14:02
pdeore#link https://5e09181bcc1a50499619-17764b56a5c622705c872e3c7dca2597.ssl.cf2.rackcdn.com/periodic/opendev.org/openstack/glance/master/glance-tox-functional-py38-cursive-tips/32bbbc2/tox/functional-py38/functional-py38-siblings.txt14:02
pdeoreNot exactly sure but seems that castellan failure might be bcz of moving py3.8 support to py3.1014:02
pdeore#link https://review.opendev.org/c/openstack/castellan/+/87993214:02
pdeoreso do we need to move these periodic jobs to py3.10 ?14:03
croelandtPy3.8 sounds like a thing of the past to me :)14:03
pdeore#link https://review.opendev.org/c/openstack/governance/+/87223214:03
dansmith3.8 is in focal yeah?14:03
croelandtbut I think people still use 3.814:03
pdeoreAs per this, within 2023.2 python version 3.9 and 3.10 are the supported python runtimes 14:04
croelandtpdeore: oh that is a good point14:04
dansmithyeah, was just looking for that because I couldn't remember if we dropped focal yet14:05
pdeoreyeah it's in focal 14:05
dansmithright, so nova is only running 39 and 310 now14:06
pdeoreohk ok, so can we move these jobs to py310?14:07
dansmithactually, nova has a 38 functional job, but I'm not sure why14:08
dansmithbut yeah the governance patch means you don't *have* to test 3.8 anymore, so if something is broken there specifically, we can drop it14:08
croelandtIf it's a setup issue, that should not really affect distros either, right?14:09
croelandtas they'll be using their own packages rather than recreating virtual envs14:09
dansmithI'm not really seeing *why* castellan is breaking on 38 though, but if it's confirmed it did something to break support (like using importlib.resources or something) then it makes sense to move on14:10
croelandton master, python_requires is now set to >=3.914:11
croelandtin setup.cfg14:12
croelandtI'm not sure whether that is actually enforced by setuptools, but it shows they won't try to support 3.8 anymore anyway14:12
dansmithmaster castellan right?14:13
dansmiththat might be why it's just failing to install then yeah14:13
dansmithand yeah, it will be enforced, AFAIK, I just expected a better error message :)14:13
croelandtmaster castellan, yes14:13
croelandtdansmith: yeah I've given up on error messages being useful :-(14:14
pdeoreyeah exactly14:15
croelandtSo, let's move to py310?14:15
pdeoreyeah doesn't seem any other option for now :)14:15
croelandtIt's settled then14:16
pdeoreit would be nice if other members would also ack on this :)14:17
abhishekkwhat about our default jobs in tox should we drop py38 from there as well?14:17
croelandtWe might want to keep unit tests as some distros are lagging behind (Debian cough cough)14:19
pdeoreI think thats what other projects have done ?14:19
croelandtbut functional tests 1) are a pain to debug 2) will require installing a lot more stuff (like castellan)14:19
abhishekkIdeally we should move14:20
croelandt+114:21
abhishekkI think rosmaita already submitted one patch, https://review.opendev.org/c/openstack/glance/+/88068514:21
pdeorecan we move these py38 job to 310 in the same patch then ?14:22
dansmithah yep, that's the patch we need14:22
abhishekkpdeore, no that should be different one14:22
dansmithalthough it kept 38 fips14:22
croelandtdo we get rid of all the 38 tests?14:23
dansmithoh and several others14:23
dansmithhmm, I wonder why he didn't convert those14:23
dansmithah fips requires 38 still, in the commit message14:23
croelandtwell Brian's patch is keeping functional-38 though14:23
abhishekkhe just moved functional tests14:23
dansmithright but there are other 38 functional jobs that can probably go14:24
croelandtoh but the functional-38 is only local ok14:24
dansmithfunctional-py38-glance_store-tips for example14:24
abhishekkack14:24
croelandtdansmith: do you wish to review his patch and suggest a second patchset?14:24
dansmithoh are those defined in glance_store?14:24
dansmithI think we should just approve this and follow up with more14:25
abhishekkI am just confused, with his patch14:25
dansmithunless it will fail on the job that requires castellan now?14:25
abhishekkwhat is the difference between line 45~47 and 49 onwards14:26
abhishekkif I run tox -e functional-py39 it will refer line 46 or 56?14:26
croelandthm14:27
dansmith4514:27
croelandtfunctional-py39 and functional-py39-rbac are not the same thing14:27
dansmith55 is the rbac14:27
dansmithright14:27
abhishekkhmm, dumb me :/14:27
abhishekkwe can define similar for rbac as well and reduce some lines :D14:27
croelandtI kind of like the patch, but I think dansmith wants more agressive cleaning up14:27
croelandtabhishekk: right14:28
croelandtthis would be good feedback14:28
abhishekkyep anyone can add that feedback14:28
dansmithcroelandt: no, I said we should merge this now and follow up with more :)14:29
dansmiththe cleanup is not relevant to the actual work of the patch though, which is to change the jobs in zuul14:29
dansmithwhich requires the 311 to be added, which is why he's converting the existing functional targets to the pattern14:30
croelandtok ok14:30
dansmithif we're going to convert the rbac 38 to 311 as well, then sure, but I'm just saying, we might as well iterate and get this going, IMHO14:30
abhishekkworks either way for me14:30
* dansmith +2d14:31
croelandt+2+1d14:31
abhishekkcool, just remember someone has to work on migrating other tests as well14:31
pdeoreack14:31
pdeorecroelandt, is it possible for you to submit the patch ? otherwise I will start that tmrw morning :)14:33
croelandtfor rbac 38->311, yes14:33
pdeoreok14:33
pdeorelet's move ahead14:34
pdeore#topic Specs14:34
pdeorethere are few specs up for review, so kindly have a look 14:34
pdeore#link https://review.opendev.org/c/openstack/glance-specs/+/880627 14:34
pdeoreI have reproposed the image encryption spec to the current cycle14:35
pdeoreI have also submitted bobcat project priorities14:35
pdeore#link https://review.opendev.org/c/openstack/glance-specs/+/88063714:35
pdeoremoving ahead14:36
pdeore#topic Important Bugs 14:36
pdeorecroelandt, ^14:36
croelandtyep14:36
croelandtso the glanceclient gate is blocked14:36
croelandtif we could merge this https://review.opendev.org/c/openstack/python-glanceclient/+/880765 that would be great14:37
croelandtsome tox4 magic I guess14:37
croelandtThen the glance-download plugin does not work well in the glanceclient, because we're trying to get attributes with invalid names14:37
croelandtsee https://review.opendev.org/c/openstack/python-glanceclient/+/88069614:37
croelandtthere is a follow-up patch that I don't really like, but sent to get input from the bug reporter, I'll try to rework this ASAP14:38
pdeoreohh ok14:40
pdeoreso this skipsdist thing we have already removed from glance14:40
croelandtand glance_store, yes14:40
pdeoreyeah14:41
* croelandt has given up on understanding the underlying details14:41
* croelandt is shoving up code in there, pushing it with his boots until it works14:41
pdeoreohk let's wait for the reporter to test it14:42
croelandtyeah the first patch is defintily good14:42
croelandtgetattr(x, 'foo-bar') cannot work14:43
croelandtsince 'foo-bar' is not a valid attribute14:43
croelandt--foo-bar is converted to "foo_bar" by argparse14:43
pdeorenot sure how it missed while implementing this :/14:44
croelandtother than that, there were a few bugs on Launchpad with not enough info so I'm not gonna bother you with this :)14:45
croelandtyeah it happens14:45
pdeorecool, let's move to open discussions then :)14:45
pdeore#topic Open Discussions14:45
pdeoreArtem has added abhishekk & croelandt into the openstacksdk-service-core group 14:46
croelandt\o/14:46
pdeoreso yeah that's it from me for today14:46
* croelandt has nothing more14:47
abhishekkadded?14:47
pdeore#link https://review.opendev.org/admin/groups/33c03b5e000a75ffc476c51d484c5111648d399c,members14:47
pdeoreyes14:47
pdeoreI thought you must have got that notification 14:47
abhishekkack14:47
pdeoreok, let's conclude for today then :)14:48
abhishekkWait14:48
pdeoreohh 14:48
abhishekkI had something 14:48
pdeoreyeah plz go ahead14:48
abhishekkany update from the author which added new proposal last week?14:48
abhishekkwhether he is going to work on it or not14:49
pdeoreno updates yet14:49
abhishekkalso cyril is working on migrating rbac jobs,so periodic jobs will be still pending14:49
abhishekkack, I guess you can try to reach him by mail if no reply till next week14:50
pdeoreyeah that I will start working on tmrw 14:50
abhishekkAND, last is there is slight change in new Location API work, rajat will submit the spec14:50
pdeoreyeah I will try to ping him 14:50
abhishekkso all reviewers please kindly have a look at it14:50
abhishekkdone from me14:51
pdeorehe has not yet updated the spec14:51
abhishekkhe will submit the updated spec 14:51
pdeoreahh missed that :/14:51
pdeorecool14:52
pdeorelet's wrap up then14:52
pdeoreThanks evryone for joining 14:52
pdeore#endmeeting14:52
opendevmeetMeeting ended Thu Apr 20 14:52:52 2023 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)14:52
opendevmeetMinutes:        https://meetings.opendev.org/meetings/glance/2023/glance.2023-04-20-13.59.html14:52
opendevmeetMinutes (text): https://meetings.opendev.org/meetings/glance/2023/glance.2023-04-20-13.59.txt14:52
opendevmeetLog:            https://meetings.opendev.org/meetings/glance/2023/glance.2023-04-20-13.59.log.html14:52

Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!