Thursday, 2022-06-02

pdeore#startmeeting glance14:00
opendevmeetMeeting started Thu Jun  2 14:00:10 2022 UTC and is due to finish in 60 minutes.  The chair is pdeore. Information about MeetBot at http://wiki.debian.org/MeetBot.14:00
opendevmeetUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.14:00
opendevmeetThe meeting name has been set to 'glance'14:00
pdeore#topic roll call14:00
pdeore#link https://etherpad.openstack.org/p/glance-team-meeting-agenda14:00
pdeoreo/14:00
abhishekko/14:00
dansmitho/14:00
rosmaitao/14:00
mrjoshio/14:00
pdeorelets wait few minutes for everyone to join14:01
* abhishekk There is intermittent electricity failure at my end since afternoon, I might get disconnected in between14:01
jokke_o/14:02
croelandto/14:02
pdeorelet's start :)14:02
pdeore#topic release/periodic jobs14:02
pdeoreWe have skipped glance M1 tag due to gate issues 14:02
pdeorePeriodic Jobs, all green except the 'oslo master' and 'translation update' jobs failure due to some python version dependency conflicts14:03
abhishekkack14:03
pdeoremoving ahead14:04
pdeore#topic intermittent failure on glance-multistore-cinder-import job14:04
pdeorerosmaita, ^14:04
rosmaitathanks14:04
rosmaitaabhishekk left a comment14:04
rosmaitai'll see if we can figure out what the race condition could be14:05
rosmaitait's intermittent, so not a big deal ATM14:05
abhishekkYes, I have tried it to reproduce locally but not able to do it14:05
abhishekkhttps://github.com/openstack/glance/blob/master/glance/api/v2/image_data.py#L14014:05
abhishekkthis is where it is failing sometime14:05
rosmaitaok, that is helpful14:05
rosmaitathat's all i have14:05
pdeoreack, moving ahead14:06
pdeore#topic Spec for review14:06
abhishekkrosmaita, let me know if anything is needed 14:06
rosmaitaty14:06
pdeorereview request again for the location APIs spec (whoami-rajat) https://review.opendev.org/c/openstack/glance-specs/+/8408814:06
whoami-rajathey14:06
jokke_yes, my bad I missed the updates on that earlier14:06
jokke_on my todo list for today14:07
jokke_been bit hectic14:07
whoami-rajatjokke_, ack thanks!14:07
dansmithwhoami-rajat: is that marked as draft?14:07
dansmithoh, no that url must be wrong14:07
whoami-rajatdansmith, i think it should be active14:07
dansmithpdeore: ^14:07
rosmaitamissing a digit, i think14:07
dansmithyeah14:07
whoami-rajathttps://review.opendev.org/c/openstack/glance-specs/+/84088214:07
dansmithends in -214:07
pdeoreoops!!14:07
pdeorehttps://review.opendev.org/c/openstack/glance-specs/+/84088214:07
whoami-rajat#link https://review.opendev.org/c/openstack/glance-specs/+/84088214:08
whoami-rajatwanted to request to at least remove the -2 so it doesn't discourage other reviewers from taking a look :)14:08
abhishekkyep, I think we should follow a practice of avoiding putting -2 as per our core review guidelines14:08
whoami-rajati mean if the concerns are addressed ^^14:08
dansmithabhishekk: agree14:08
whoami-rajatabhishekk, +114:09
abhishekkwhoami-rajat, has pointed it out to me that we have these guidelines14:09
jokke_whoami-rajat: yeah, will read it through and act accordingly14:09
jokke_sorry for leaving it hanging for that long14:09
abhishekkno worries14:09
pdeoreSpec for SRBAC system admin persona support in glance - https://review.opendev.org/c/openstack/glance-specs/+/84428914:10
jokke_abhishekk: I say this again, feel free to remove me from core if you don't want me using my +-2 votes ;)14:10
abhishekkI had a look at it and given some suggestions14:10
pdeoreI have submitted a spec for whatever we discussed so far on handling the owner issue for system scope14:10
jokke_pdeore: saw that today. Will hopefully read through it too14:11
pdeoreabhishekk, I've addressed your comments, kindly have a look at new patch set14:11
abhishekkjokke_, noted, I need to see everyone's interest is maintained14:12
pdeorejokke_, thannks!14:12
abhishekkpdeore, It needs some rewording, I will have a look at it once again14:12
pdeoreabhishekk, ack14:12
pdeoreUpdate proposal for duplicate image download - https://review.opendev.org/c/openstack/glance-specs/+/734683 14:13
pdeorethis one also needs some reviews14:13
abhishekkthis is pending for long, I think  rosmaita have some concerns/suggestions about it 14:13
pdeoreok, moving ahead14:15
rosmaitai haven't looked at the latest version yet14:15
pdeore#topic Are the version tests useless?14:15
pdeorerosmaita, ^14:15
rosmaitasorry14:17
pdeorenp :)14:18
rosmaitathis is what i am talking about14:18
rosmaita#link https://review.opendev.org/c/openstack/glance/+/84302814:18
dansmithI would also add in that making the version we expose differ based on config is also not a great experience, which makes those tests far more complicated than need be14:19
abhishekkrosmaita, afaik, this version increment is quiet messy at this moment and we need to sort it out14:19
rosmaitayeah, and that's why i think they didn't break and detect the problem14:19
rosmaitaanyway, i guess my real point is that if you review anything that changes the version, make sure it is testing the right stuff14:20
rosmaitaok, that's all from me, unless anyone has questions14:21
abhishekk++14:21
pdeore++14:21
dansmithcould we just discuss and move towards a single linear monotonically increasing version?14:21
jokke_So say if it was just testing constants (not our first test doing so) I'd agree 100% now maybe only 90% :P14:21
dansmithand stop the variation based on config?14:21
abhishekk+1 for stopping increasing version based on config14:23
jokke_dansmith: IIRC there is nothing that is based on config that is not already deprecated to be removed. All of the versions that are config dependant should be new features to be always on, but configurable at the beginning14:23
abhishekkit is way more difficult to maintain these versions now as we have couple of config options in there14:24
dansmithjokke_: okay, I think we had a couple things recently added that kept the same "if this is not enabled, then fall back in version"14:24
abhishekkfor multi store and image cache APIs14:24
dansmithjokke_: but maybe we just need a comment in there that we should stop with the config-based versioning14:24
dansmithand let the old stuff age out14:24
jokke_well multi-store config is going away14:24
dansmithhttps://review.opendev.org/c/openstack/glance/+/843028/2/glance/api/versions.py14:25
dansmiththe image_cache_dir controls which version we expose, for example14:25
abhishekknot until we will be stopping single store support which will take couple of cycles atleast14:25
dansmithand even still, you can't request the old version and get the old behavior (AFAIK) so I don't know why we wouldn't just collapse everything that is there right now14:25
rosmaitathe version negotiation will 404 if someone requests /v2.15/whatever and it's not enabled14:25
jokke_abhishekk: my point, it's not planned to be there for ever, just ended up being there longer than we hoped for14:25
dansmithand expose 2.16 as the current14:25
dansmithrosmaita: but it doesn't mean anything right? it's silly to say 2.13 is enabled and 2.16 is enabled, but 2.15 is disabled14:26
rosmaitawell, it sort of gives you some info14:26
jokke_so I've never understood the need for minor version API entrypoints in first place as we're not microversioning anything14:26
jokke_it's just indicator 14:27
jokke_So it shouldn't be headace to anyone either just always call v2/foo/bar and you're grand14:27
abhishekkjokke_, right, the point is it is difficult to maintain and to much confusing even for me to address new version bump if required14:27
dansmithrosmaita: if you're saying they're supposed to be feature flags, I think it's super confusing for them to have linear numbers, instead of names or uuids :)14:28
jokke_I guess it's good question to the next user survey if anyone is actually looking at the version info ... iirc I was this great idea of having autogenerating client based on schemas that drove us to the api versioning in the first place14:29
dansmithand the current/supported classification would make even less sense14:29
rosmaitai'm not sure what i'm saying14:29
dansmithokay :)14:29
rosmaitabut it does help to know what version of the api you are contacting14:29
rosmaitabecause if it's train, there's stuff you won't be able to do14:29
dansmithrosmaita: right, but if the top version comes and goes based on config,14:29
dansmiththen you don't know14:29
rosmaitawell, you know for the site you are contacting, right?14:30
dansmithso if you're trying to be bug compatible with train, but you can't tell if it's train or ussuri because config changes it, then you're hosed14:30
jokke_The minor API version is there to indicate change in the API ... we have releases without API version bump and we have ones with. That's all it is, a flag "smething in the API has changed since the previous"14:30
jokke_so it's not feature flag per se14:30
dansmithright,14:30
dansmithand if we have one version in a release that is dependent on config,14:30
dansmiththen you could look like release N or N-1 depending on your config14:31
jokke_we've had quite a few of them, clearing out as the config options gets finally removed that allowed disabling the features causing the bump14:31
dansmithwell, perhaps a topic for the next ptg14:32
jokke_I think the most confusing part is we have no documentation mapping what feature caused which version bump, so people really can't figure any of that out without looking git logs14:32
dansmithjokke_: well, then let's torch it for sure :)14:33
abhishekkI think to be short term solution we should carefully review a version bump patch, long term create a survey to check if version info is used or not and then follow some standard for version increase thereafter14:33
rosmaitai left some comments on the patch that's up to fix the version history in the api-ref14:33
jokke_I'd say if it's not useful lets get rid of the minor version in that api version all together and call it v2 like everyone using it :P14:34
jokke_and we can just put a shim in place that routes /v2*/ -> /v2/14:34
jokke_So I'm not convinced the tests are useless, the minor version numbers might be :D14:36
rosmaitawell, it would be helpful to know what's new in the API14:36
rosmaitahttps://review.opendev.org/c/openstack/glance/+/841053/2/api-ref/source/versions/index.rst#2514:36
jokke_rosmaita: true, I wish we had something like concise notes for each release highlighting new features and changes14:37
jokke_oh wait :P14:37
abhishekkrosmaita, I think if owner is not around I will work on that patch14:39
rosmaitathat's ok with me14:40
abhishekkI have noted this topic to be discussed in next PTG where probably we will meet in person 14:42
pdeorecool, shall we move to next topic ?14:43
abhishekklets move ahead for now14:43
rosmaitain fabulous columbus, ohio!!!14:43
abhishekkyep14:43
pdeoremoving to open discussion14:44
pdeore#Open Discussion14:44
croelandtCould Core take a few minutes to look at stable patches? :)14:44
croelandtI think we can abandon the Victoria one14:44
croelandtand there are a few that could use some +2s14:44
dansmithcroelandt: s/core/stable core/ :)14:44
abhishekkdansmith, I have pushed new revision to immediate caching of images14:44
croelandtdansmith: yes :)14:45
dansmithabhishekk: just recently I assume? I looked yesterday and it was still failing14:45
abhishekkits still WIP though14:45
abhishekknot fixed tests atm14:45
dansmithoh okay14:45
abhishekkjust wanted to make sure that approach is OK before moving to fix tests14:45
dansmithokay I'll look14:46
abhishekkcroelandt, we can abandon victoria patches14:46
abhishekkI will have a look at others today or tomorrow morning14:46
croelandtok ok14:46
abhishekkdasm|off, thank you14:46
abhishekkI am not able to found victor or the other member who were working on glance-download import method14:47
abhishekkalistarle, around?14:47
abhishekkdansmith, thank you14:47
* abhishekk tagged different handle by mistake 14:48
pdeoreanyone has anything else to discuss? or we are done for today ? :)14:49
jokke_I have nothing, thnx14:50
abhishekknothing from me as well14:50
abhishekkpdeore, just follow up for devstack multistore related patches with devstack team14:51
pdeoreabhishekk, sure14:51
abhishekkyeah, we need to get it merged so that we can start work towards removing single store support14:51
alistarleHi, I see you comment on the glance patch abishek14:52
abhishekkalistarle, yes14:52
abhishekkjust added it now14:52
alistarlesorry for that, indeed Pierre Samuel is currently in hollyday, he will be back next week14:52
abhishekkack, no worries14:53
abhishekklet us know if you guys need any help14:53
alistarlewe just need to write tests normally, this code is already in production on our side and seems to work well14:53
abhishekkyeah, I have also given some reference to Pierre about coding standards which needs  to be followed14:54
alistarleyeah sure, you can say in the patch if there is anything you think is not right, we will fix it right away14:54
abhishekkother than that I will have a look again at it14:54
alistarleabhishekk have you seen something which are not following the guideline ?14:54
abhishekkI don't remember it now, but will highlight it on the patch14:55
abhishekkone thing is related to import14:55
abhishekkwhere method name or exception class is imported which is not as per Openstack standard14:55
abhishekkwill highlight those on the patch14:56
abhishekklast 3 minutes14:57
alistarleyes please, we'll take care of that. And if you think the overall implementation is good we will fix and write the tests for the feature :)14:57
abhishekkalistarle, ack, thank you14:57
pdeoreThanks everyone for joining!! 14:58
rosmaitabye!14:59
jokke_thanks all14:59
pdeore#endmeeting15:00
opendevmeetMeeting ended Thu Jun  2 15:00:01 2022 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)15:00
opendevmeetMinutes:        https://meetings.opendev.org/meetings/glance/2022/glance.2022-06-02-14.00.html15:00
opendevmeetMinutes (text): https://meetings.opendev.org/meetings/glance/2022/glance.2022-06-02-14.00.txt15:00
opendevmeetLog:            https://meetings.opendev.org/meetings/glance/2022/glance.2022-06-02-14.00.log.html15:00

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