14:00:14 #startmeeting cinder 14:00:14 Meeting started Wed Sep 21 14:00:14 2022 UTC and is due to finish in 60 minutes. The chair is whoami-rajat. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:14 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:00:14 The meeting name has been set to 'cinder' 14:00:19 #topic roll call 14:00:23 o/ 14:00:39 o/ kind of. 14:00:41 mep 14:01:10 o/ 14:01:17 o/ 14:01:31 hello 14:02:21 0/ 14:02:21 o/ 14:02:46 #link https://etherpad.openstack.org/p/cinder-zed-meetings 14:02:51 o/ 14:03:31 hello 14:04:16 I can see the usual people 14:04:19 let's get started 14:04:26 #topic announcements 14:04:34 first, RC-1 Released 14:04:40 #link https://review.opendev.org/c/openstack/releases/+/857102 14:04:48 we released RC1 this monday 14:05:06 so now the master is updated to 2023.1 (or as we would like to call it Antelope) 14:05:10 and a stable/zed branch is created 14:05:18 any fix merged now will be in the antelope release 14:05:39 and if it's critical for the zed release then needs to be backported to stable/zed 14:05:45 which brings me to the second announcement 14:05:47 Cinder RC-2 14:05:57 We're planning to have an RC-2 release 14:06:01 deadline is next week 14:06:24 I've created an etherpad for pontential fixes that would be good to get in into the zed release (RC-2) 14:06:26 #link https://etherpad.opendev.org/p/cinder-zed-rc2-bugfix 14:06:43 if you think you've a patch that needs to get in the zed release, please add it to the etherpad 14:07:21 I've already added some patches that i think are good for the release 14:07:43 next, Ubuntu Summit 14:07:52 #link https://lists.openstack.org/pipermail/openstack-discuss/2022-September/030533.html 14:08:11 I'm not sure how many people are interested in this but they're planning to have a physical as well as virtual summit 14:08:21 read the email thread for more details 14:08:40 next, Election results: TC and PTL 14:08:51 #link https://lists.openstack.org/pipermail/openstack-discuss/2022-September/030541.html 14:09:03 we've the list of newly elected TC members and all the PTLs 14:09:32 that's all i had for announcements, anyone has anything else? 14:10:27 congrats whoami-rajat ! 14:10:42 thanks :) 14:10:48 and thanks for PTL again 14:11:28 it's a good opportunity! 14:11:29 ++ 14:12:12 ++ 14:12:12 okay, let's move to topics then 14:12:34 * whoami-rajat feels glad to have the team support 14:12:47 #topic cinder tempest plugin release 14:12:55 #link https://review.opendev.org/c/openstack/releases/+/858401 14:13:01 congrats Rajat! 14:13:18 this is more kind of an announcement but wanted to discuss this 14:13:31 does anyone have any patches they would like to get in for the cinder tempest plugin release? 14:13:35 the deadline is this week 14:13:38 thanks HappyStacker 14:14:12 tosky, ^ you've any patch in mind that would be good for the release? 14:14:50 i have one, but i'd like to know what you think about 14:15:00 sure 14:15:08 * enriquetaso looking for the enabling RBD min client to mimic 14:15:31 uhm, not on top of my head 14:16:01 maybe https://review.opendev.org/c/openstack/cinder-tempest-plugin/+/852675 ? 14:16:05 ok, let me know if something comes up and we can try to get that in else I will +1 the release patch on friday 14:16:08 tests a functionality introduced in zed 14:16:13 enriquetaso: cinder-tempest-plugin is branchless, so it could be enabled later in a job if all pieces are in place in cinder 14:16:20 we can also release it 14:16:23 #link https://review.opendev.org/c/openstack/cinder/+/848013 14:16:29 rosmaita, oh yeah that is a good one 14:16:34 true :P ok 14:16:48 we have a few SRBAC pending patches, at least one which refactors the code a bit 14:17:21 enriquetaso, that is a cinder patch. we are referring to changes in cinder-tempest-plugin 14:17:30 #link https://review.opendev.org/c/openstack/cinder-tempest-plugin/+/823719 14:18:06 this one is failing zuul as of now 14:18:24 tosky, are you referring to this? https://review.opendev.org/c/openstack/cinder-tempest-plugin/+/842954 14:19:39 I can see i also have a patch that fixes a problem with idempotent IDs, if anyone would like to review https://review.opendev.org/c/openstack/cinder-tempest-plugin/+/839301 14:20:06 yes, that refactor patch 14:20:28 and yes, the idempotent ID has a review priority+1 from me 14:21:23 ack, will keep an eye on the 3 patches 14:22:02 that's all from my side 14:22:04 moving on 14:22:16 #topic Outreachy December 2022 to March 2023 internship round 14:22:18 enriquetaso, that's you 14:22:39 Outreachy provides internships in open source and opens science to people impacted by underrepresentation in the technology industry. 14:22:48 Openstack has funding for 3 interns in December 2022 to March 2023 internship round. Mentors can submit projects for Outreachy applicants until Sept. 29, 2022, at 4 pm UTC. 14:23:07 I plan to proposed myself as a mentor for this round 14:23:15 The intern goal would be "extending the API reference to handle microversions". The work was started by Gorka and Rajat so the plan would be to break what is missing (I did a quick check and I think is the newest microversions, right?) into smaller work items, maybe based on microversion. 14:24:46 Maybe we can talk about it whoami-rajat after the cinder meeting 14:24:48 sounds like a useful task, we need some suggestions for making this sound sexy, though 14:25:07 rosmaita, yes... it doesn't sounds sexy now lol 14:25:09 sure, thanks for bringing this up enriquetaso 14:25:39 maybe y can create a etherpad or a google doc and share with you to suggest and review the proposal ? 14:25:56 I can* 14:25:57 would be good to get the gaps addressed we've in our api-ref with the addition of functional tests 14:26:00 i mean, you can learn a lot doing this, just need to spell this out a bit 14:26:55 etherpad sounds good 14:26:57 yeah, we would need to divide the work items and for that we would need to see what is currently missing 14:27:19 whoami-rajat++ 14:27:25 rosmaita++ 14:27:51 OK, thanks. If another member would like to proposed a project the link is: 14:27:58 #link https://www.outreachy.org/communities/cfp/openstack/ 14:28:18 enriquetaso: thank you for doing this 14:28:46 I'm positive about it :D 14:28:49 thanks enriquetaso ! 14:29:30 so we don't have any other topics for today 14:29:34 let's move to open discussion 14:29:37 #topic open discussion 14:29:45 HappyStacker, you had an issue to discuss? 14:29:59 yes 14:30:25 so initially we have an issue with Powerflex and the way it handles the volume creation and extension 14:30:39 as you may know, it rounds the capacity by 8GB multiplier 14:31:22 up until this patch, every time a volume is creatd it gets a model update and update cinder DB accordingly 14:31:49 so if the user creates a 2GB volume, cinder will reflect 8GB as for the powerflex backend 14:32:10 We had a discussion a few months agos and we agreed this should NEVER happened 14:32:38 as cinder expects to show the size requested by the user not what happens in the backend 14:32:50 so the devs removed this piece of code 14:33:15 and now all of our CI are broken 14:33:23 DELL PowerFlex :are we talking about https://bugs.launchpad.net/cinder/+bug/1952355 ? 14:33:32 I don't think there was any problem with the create volume 14:33:37 the main problem is with extend volume which doesn't return a model_update, hence the difference between size in cinder DB and backend LUN 14:34:03 https://review.opendev.org/c/openstack/cinder/+/836973 14:34:06 this patch 14:34:22 related to this bug https://bugs.launchpad.net/cinder/+bug/1968164 14:34:25 and if we were to return the model update, we have to update our driver interface and every backend would need to have the same return type 14:34:40 #link https://bugs.launchpad.net/cinder/+bug/1968164 14:34:41 so we're kind of stucked here 14:34:49 #link https://review.opendev.org/c/openstack/cinder/+/836973 14:35:49 I think cinder should have the real size in the DB, i.e., even if user asks for 2 GB, cinder says 8GB ... because once the volume is attached, there is no way to prevent the user from filling all 8GB, right? 14:36:37 rosmaita, yes 14:37:07 I know we need to work on how to manage it but it'll take time to write the specs and code it 14:37:41 I think for now the problem is how to revert the patch to fix the CI 14:37:42 without this patch if the user asks 2, cinder says 2 but powerflex has 8 14:37:47 the patch is also part of zed RC1 14:37:52 so need to revert in stable/zed as well 14:38:17 when you want to extend to 4 , cinder says ok, but powerflex doesn't understand as the size is already 8 14:38:23 HappyStacker, i don't think that's the case, create volume should work fine, the extend volume should report different values right? 14:38:24 and thus failing 14:38:34 yes create works fine 14:39:03 in create call, we return the rounded up size in the model_update so it always returns the right size to cinder 14:39:13 gotcha 14:39:13 but extend doesn't return anything 14:39:24 well fine, if we consider having discrepencies between cinder and powerflex are ok 14:40:08 so I'm asking your recommendation on how we can proceed 14:40:24 I think we should fix it but we can't do anything about it right now 14:40:32 currently the requirement is to fix the dell CI right? HappyStacker 14:40:53 if we let it as is, PowerFlex CI will alsway fail 14:41:19 or we can skip the cases which extends a volume 14:41:39 the patch only touches powerflex, so i think revert change in master, and backport the revert to stable/zed 14:41:51 (and re-open the bug) 14:41:59 rosmaita, that's what i was thinking about 14:42:35 then fix in master soon, and backport 14:42:40 ++ 14:43:03 so plan is First: revert change in master, and backport the revert to stable/zed 14:43:32 Second: work on the implementation of this 8GB support in master then backport 14:43:45 Is that correct? 14:44:03 correct 14:44:35 after the revert, I'll reset the launchpad bug 14:44:53 ok thks I'll certainly need your help on scoping the work for the new feature 14:45:17 How can I revert the change? 14:45:23 good that it got flagged before the official release 14:45:34 HappyStacker, there is an option on the top right 14:45:46 I just saw it 14:45:47 you will need to mention the reason for revert in the commit message 14:46:20 ok 14:46:44 that's for reverting the change, what about backporting it to Zed? 14:47:02 cherry pick it after it merges 14:47:36 OK I'll sort it out and get back to you in case I have questions 14:47:40 i think there's a option on the top right menu 'cherry pick' 14:48:04 i think you may cherry pick it to stable/zed 14:48:25 but, i'm not sure 14:48:38 yep, that's it 14:48:38 So click on cherry-pick and choose stable/Zed 14:48:52 after reverting it 14:48:56 right? 14:49:07 yeah but cherry pick after the revert patch merges 14:49:14 yep roger that 14:49:38 thank you all! 14:49:49 no problem 14:49:51 I was so confused on the way to handle it 14:50:05 Glad we all agreed 14:50:25 This is a community! revert change in master, and backport the revert to stable/zed 14:50:27 HappyStacker: put this on the agenda for the PTG, and we can discuss the fix 14:50:32 ok 14:50:37 will do 14:50:54 good to sort that out, do we have anything else for open discussion? 14:50:57 rosmaita, +1 14:51:03 Hey, I have added a test to cinder-tempest-plugin for the Ceph backup driver patch (see https://review.opendev.org/c/openstack/cinder-tempest-plugin/+/855876). Some questions: How can I ensure this test only runs with Ceph driver? How can I set Cinder config options to specific values? 14:51:41 jhartkopf, it currently has some pep8 failures that needs to be addressed 14:52:10 I don't think there is a driver specific option as of now 14:52:15 tosky, ^ 14:53:11 yes, pep8 failures will be addressed 14:54:47 Problem is the test only works with Ceph and obviously only makes sense with Ceph. And it requires a specific config value. 14:54:48 jhartkopf, the way that gate works currently is you can only set devstack variables and those variables need to enable project specific config options 14:55:50 i think we have some decorators to skip drivers, but i'm not sure to only run one driver, i'll review 14:55:56 well, there's also a post_config (i think it's called) section you can put in zuul.yaml that allows you to modify the cinfog file 14:55:57 jhartkopf: there are a few tests which explicitly set a skip if the plugin is ceph, even though that doesn't work well for clusters with multiple cinder backends 14:57:26 enriquetaso: a review would be welcome, thank you 14:58:01 so basically a mixture of skip everything except Ceph (maybe) and Zuul post_config, right? 14:58:45 Before the meeting ends, is anyone running openstack redhat platform system and I can ask him few questions? 14:59:02 maybe a devstack + tempest config option would be useful for these kind of scenarios but too much work for a single test 15:00:03 yuval, the meeting is over basically but questions are welcome in #openstack-cinder 15:00:09 we're out of time 15:00:16 thanks everyone for joining 15:00:21 thanks 15:00:22 #endmeeting