| opendevreview | Masayuki Igawa proposed openstack/cinder master: Doc: Fix way to update default quota value for a new project https://review.opendev.org/c/openstack/cinder/+/845069 | 01:07 |
|---|---|---|
| opendevreview | Keigo Noha proposed openstack/os-brick stable/train: multipath/iscsi: iSCSI connections are not reinitiated after reboot https://review.opendev.org/c/openstack/os-brick/+/829047 | 04:22 |
| *** abhishekk is now known as akekane|home | 05:01 | |
| *** akekane|home is now known as abhishekk | 05:01 | |
| whoami-rajat | rosmaita, hey, i think you missed out this patch in the DB series https://review.opendev.org/c/openstack/cinder/+/837538/ | 09:15 |
| opendevreview | Merged openstack/cinder-tempest-plugin master: Remove stable/victoria job and fixing job description https://review.opendev.org/c/openstack/cinder-tempest-plugin/+/843685 | 10:11 |
| opendevreview | Merged openstack/cinder master: Remove redundant line from host_manager.py https://review.opendev.org/c/openstack/cinder/+/844920 | 10:11 |
| stephenfin | rosmaita: Can you revisit https://review.opendev.org/c/openstack/cinder/+/837538/ please? | 10:16 |
| stephenfin | rosmaita: I also suspect this can be bumped to a +2 now as I've addressed the concern in a follow-up https://review.opendev.org/c/openstack/cinder/+/837540 | 10:17 |
| opendevreview | Merged openstack/cinder master: db: Migrate online upgrade helpers to enginefacade https://review.opendev.org/c/openstack/cinder/+/837537 | 10:43 |
| opendevreview | Brian Rosmaita proposed openstack/cinderlib master: Add requirements-check job https://review.opendev.org/c/openstack/cinderlib/+/845026 | 10:50 |
| *** dviroel|afk is now known as dviroel | 11:19 | |
| opendevreview | Jean Pierre Roquesalane proposed openstack/cinder master: WIP: Add Cinder NFS driver for Dell EMC PowerStore https://review.opendev.org/c/openstack/cinder/+/797608 | 11:20 |
| opendevreview | Arun KV proposed openstack/cinder master: Reintroduce DataCore driver https://review.opendev.org/c/openstack/cinder/+/836996 | 12:12 |
| opendevreview | Brian Rosmaita proposed openstack/cinderlib master: Add requirements-check job https://review.opendev.org/c/openstack/cinderlib/+/845026 | 12:44 |
| opendevreview | Jean Pierre Roquesalane proposed openstack/cinder master: Add NVMe/TCP support to Dell EMC PowerStore driver https://review.opendev.org/c/openstack/cinder/+/819149 | 12:44 |
| opendevreview | Vlad Gusev proposed openstack/cinder stable/yoga: RBD: Don't flatten temporary resources https://review.opendev.org/c/openstack/cinder/+/845039 | 12:53 |
| opendevreview | Brian Rosmaita proposed openstack/cinderlib master: Update doc/requirements.txt https://review.opendev.org/c/openstack/cinderlib/+/845126 | 12:55 |
| opendevreview | Brian Rosmaita proposed openstack/cinderlib master: Add requirements-check job https://review.opendev.org/c/openstack/cinderlib/+/845026 | 12:59 |
| opendevreview | Vlad Gusev proposed openstack/cinder stable/xena: RBD: Don't flatten temporary resources https://review.opendev.org/c/openstack/cinder/+/845130 | 13:04 |
| whoami-rajat | Cinder meeting in #openstack-meeting-alt at 1400 UTC | 13:59 |
| whoami-rajat | jungleboyj rosmaita smcginnis tosky whoami-rajat m5z e0ne geguileo eharney walshh_ jbernard sfernand enriquetaso hemna fabiooliveira yuval tobias-urdin | 13:59 |
| *** enriquetaso_ is now known as enriquetaso | 14:02 | |
| *** whoami-rajat__ is now known as whoami-rajat | 14:16 | |
| whoami-rajat | rosmaita, i might have done a bad copy paste, can you please check the indentation of your topics if they're correctly aligned | 14:27 |
| rosmaita | whoami-rajat: ack | 14:28 |
| whoami-rajat | sorry about that | 14:28 |
| rosmaita | np, looks fine | 14:30 |
| *** dviroel is now known as dviroel|lunch | 14:47 | |
| enriquetaso | #startmeeting cinder_bs | 15:02 |
| opendevmeet | Meeting started Wed Jun 8 15:02:01 2022 UTC and is due to finish in 60 minutes. The chair is enriquetaso. Information about MeetBot at http://wiki.debian.org/MeetBot. | 15:02 |
| opendevmeet | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 15:02 |
| opendevmeet | The meeting name has been set to 'cinder_bs' | 15:02 |
| enriquetaso | Full list of bugs | 15:02 |
| enriquetaso | #link http://lists.openstack.org/pipermail/openstack-discuss/2022-June/028887.html | 15:02 |
| rosmaita | nothing like the old cinder BS | 15:02 |
| enriquetaso | :P today's a small one | 15:02 |
| enriquetaso | #topic [IBM Storwize] lsfcportsetmember is being called in the wrong SVC code level | 15:02 |
| enriquetaso | #link https://bugs.launchpad.net/cinder/+bug/1976499 | 15:02 |
| enriquetaso | Fix proposed to master already, i've already reviewed, feel free to check it | 15:02 |
| rosmaita | sounds like a pure storwize problem | 15:03 |
| enriquetaso | it is | 15:03 |
| enriquetaso | #topic When creating volume from an image, the image is fetched twice | 15:03 |
| enriquetaso | #link https://bugs.launchpad.net/cinder/+bug/1977868 | 15:03 |
| enriquetaso | I know it's not the best idea to download the same image twice, but that's how we currently do it. However, I am not quite sure how to respond to this bug report. I wonder if whoami-rajat or another interested fellow could help me with this. | 15:03 |
| whoami-rajat | don't think that's true but need to check ... | 15:03 |
| rosmaita | launchpad is really slow for me today | 15:03 |
| eharney | yeah i can't even load the bugs | 15:03 |
| rosmaita | enriquetaso: you broke the bug tracker!!! | 15:04 |
| eharney | i think it is true that it can happen twice, but we do limit it w/ locking | 15:04 |
| enriquetaso | true.. was working half an hour before .. | 15:04 |
| enriquetaso | hahaha | 15:04 |
| enriquetaso | eharney, good question, i'm not sure | 15:05 |
| enriquetaso | i have to check | 15:05 |
| rosmaita | launchpad is apparently in the middle of an update | 15:05 |
| enriquetaso | oh | 15:05 |
| enriquetaso | well, that was the last bug for todays meeting anyway | 15:05 |
| geguileo | btw we kind of have 6 broken os-brick unit tests | 15:05 |
| eharney | only 6? | 15:06 |
| geguileo | it doesn't affect the gate since they use an old OS | 15:06 |
| whoami-rajat | I can take a look (if i get time from the pile of pending reviews) | 15:06 |
| geguileo | whoami-rajat: I have a local patch | 15:06 |
| geguileo | whoami-rajat: I'll submit it shortly | 15:06 |
| enriquetaso | #action (enriquetaso): check when happen twice and if we limit it with locking | 15:06 |
| whoami-rajat | geguileo, no, it was for the previous bug ... sorry | 15:06 |
| enriquetaso | whoami-rajat, thanks! | 15:06 |
| geguileo | lol | 15:06 |
| enriquetaso | lol | 15:06 |
| * geguileo realizing the world doesn't revolve around him | 15:07 | |
| enriquetaso | do we have patches for the os-brick unit test failures? geguileo | 15:07 |
| rosmaita | geguileo: we always assume that if you file a bug, you already have a fix | 15:07 |
| rosmaita | :D | 15:07 |
| whoami-rajat | :D thanks geguileo for finding and fixing it | 15:07 |
| geguileo | enriquetaso: I have it locally, but it's the first one of the new rebased nvmeof series, so will send it in an hour or so | 15:07 |
| eharney | enriquetaso: here's a starting point re: locking... https://review.opendev.org/c/openstack/cinder/+/556607 | 15:07 |
| geguileo | rosmaita: lol | 15:07 |
| whoami-rajat | what rosmaita said is true! | 15:08 |
| enriquetaso | thank eharney | 15:09 |
| enriquetaso | #topic open discussion | 15:09 |
| enriquetaso | do we have more bugs to keep an eye on geguileo ? | 15:09 |
| geguileo | enriquetaso: not that I have filed | 15:10 |
| enriquetaso | cinder team: feel free to proposed bugs now | 15:11 |
| opendevreview | Gorka Eguileor proposed openstack/os-brick master: Support extending encrypted volumes https://review.opendev.org/c/openstack/os-brick/+/836059 | 15:13 |
| opendevreview | Gorka Eguileor proposed openstack/os-brick master: NVMe-oF: Consolidate code paths https://review.opendev.org/c/openstack/os-brick/+/836060 | 15:13 |
| opendevreview | Gorka Eguileor proposed openstack/os-brick master: NVMe-oF: Disconnect subsystems https://review.opendev.org/c/openstack/os-brick/+/836062 | 15:13 |
| opendevreview | Gorka Eguileor proposed openstack/os-brick master: Support shared_targets tristate value https://review.opendev.org/c/openstack/os-brick/+/836063 | 15:13 |
| opendevreview | Gorka Eguileor proposed openstack/os-brick master: Fix Unit Test failures due to host's ANA support https://review.opendev.org/c/openstack/os-brick/+/845149 | 15:13 |
| enriquetaso | cool | 15:13 |
| enriquetaso | guess that was all for today! | 15:13 |
| geguileo | enriquetaso: rosmaita whoami-rajat ^ that last one is the fix for unit tests | 15:13 |
| enriquetaso | geguileo++ | 15:13 |
| rosmaita | geguileo: ack | 15:13 |
| whoami-rajat | ack | 15:13 |
| enriquetaso | thank you for attending! | 15:13 |
| geguileo | it's a oneliner, so easy review | 15:13 |
| enriquetaso | #endmeeting | 15:14 |
| opendevmeet | Meeting ended Wed Jun 8 15:14:31 2022 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 15:14 |
| opendevmeet | Minutes: https://meetings.opendev.org/meetings/cinder_bs/2022/cinder_bs.2022-06-08-15.02.html | 15:14 |
| opendevmeet | Minutes (text): https://meetings.opendev.org/meetings/cinder_bs/2022/cinder_bs.2022-06-08-15.02.txt | 15:14 |
| opendevmeet | Log: https://meetings.opendev.org/meetings/cinder_bs/2022/cinder_bs.2022-06-08-15.02.log.html | 15:14 |
| whoami-rajat | thanks enriquetaso ! | 15:14 |
| eharney | cinder-tempest-plugin? | 15:14 |
| enriquetaso | oh true, let's wait for gmann | 15:15 |
| gmann | o/ | 15:15 |
| enriquetaso | hello gmann | 15:16 |
| enriquetaso | it's not clear what the correct process is when we are adding tests to cinder-tempest-plugin that require newly landed changes in tempest | 15:16 |
| enriquetaso | just saying "assume it's all running master" means that people who want to install tempest and our plugin themselves from pip will just break | 15:16 |
| gmann | about compatibility of tempest and tempest plugins are 1. we test all supported stable branch (Wallaby -> yoga) with tempest master and all tempest plugin master so we can 100% confirm the compatibility of both with that time period. | 15:16 |
| * enriquetaso copy-paste from the cinder meeting | 15:17 | |
| eharney | my concern isn't really about stable branches necessarily | 15:17 |
| eharney | cinder-tempest-plugin says in requirements.txt that it requires tempest>=27.0.0 currently | 15:17 |
| gmann | if there is old tempest is pinned in stable brnach like stable/victoria then yes cinder-tempest-plugin master and Tempest old pin in stable/victotria cannot be guaranteed as compatible | 15:17 |
| gmann | eharney: tempest in requirements.txt is not sure how well that is maintained. | 15:18 |
| eharney | if you add new functionality to tempest and then write new test code in cinder-tempest-plugin that needs that functionality, don't you need to bump that version? | 15:18 |
| eharney | i'm just asking how it should be done. | 15:18 |
| eharney | because what we are doing now seems incorrect | 15:18 |
| eharney | and will break people consuming cinder-tempest-plugin outside of the gate | 15:18 |
| gmann | but we can say anyone that current tempest or plugins master are all good to use for stable/wallaby-> master as we test them in upstream | 15:18 |
| gmann | eharney: we can bump but in our upstream testing we use master version only. tempest is not in requirements repo upper constraints | 15:19 |
| gmann | so we can keep up to date version in requirements.txt on plugin side but that needs to be best effort. | 15:20 |
| eharney | i'm not talking about upstream testing, i'm talking about the fact that we cut releases of cinder-tempest-plugin, and presumably somebody is expecting those to work | 15:20 |
| gmann | eharney: yes you can do that for that reason. | 15:20 |
| gmann | or best way to find the compatible version of both is from release page too https://releases.openstack.org/yoga/index.html#tempest-plugins | 15:21 |
| gmann | eharney: enriquetaso ^^ that page is best effort to tell production testing also what is compatible version of plugin and tempest for that release | 15:22 |
| gmann | but again I will say master are always compatible as we test them and guarantee. | 15:22 |
| gmann | for example new cinder-tempest-plugin is released 1.7.0 today. so we can say it is compatible with latest tempest released version 31.0.0 | 15:23 |
| gmann | either through release page or via requirement.txt | 15:23 |
| enriquetaso | so we need to update the tempest version in the cinder plugin | 15:23 |
| rosmaita | gmann: ok, and if we need some tempest changes for something we have in cinder-tempest-plugin, do we ask for a tempest release? | 15:24 |
| rosmaita | (so we can coordinate cinder-tempest-plugin and tempest?) | 15:24 |
| gmann | we can for record but remember that will be best effort and need to be done carefully as we do not use requirement.txt file in our upstream tresting | 15:24 |
| enriquetaso | i.e https://review.opendev.org/c/openstack/tempest/+/843542 | 15:24 |
| gmann | rosmaita: yes, that we can do always, any time in between of release or so | 15:25 |
| gmann | enriquetaso: yes, we can cut a tempest release for that and mention in requirement.txt file too as we know by knowing the changes | 15:25 |
| rosmaita | ok, so it sounds like whenever we release a new version of cinder-tempest-plugin, we should make sure it works with the most recent tempest release | 15:25 |
| rosmaita | and if it doesn't ask for a new release | 15:26 |
| gmann | rosmaita: yes. that is true | 15:26 |
| rosmaita | (i guess we should check *before* we release cinder-tempest-plugin, though) | 15:26 |
| gmann | that way release page also will be up to date and also yoou can keep up to date in cinder-temepst-plugin requiremnts.txt file | 15:27 |
| enriquetaso | that would be nice, i think we need a new tempest release for https://review.opendev.org/c/openstack/tempest/+/843542 | 15:27 |
| gmann | rosmaita: enriquetaso eharney yes, like we want to release plugin as it has good amount of changes and need tempest to be released and we will do both release | 15:27 |
| rosmaita | ok, cool | 15:28 |
| gmann | enriquetaso: right, tempest.scenario.manager is a stable interface for plugins and any change in that is all good to be released | 15:28 |
| gmann | so in current case we can release tempest now or can wait for more time if you want to wait for more things in cinder-temepst-plugin to be released together | 15:29 |
| gmann | just ping us in qa whenever you want to do and we release tempest first and them you can update requiremnts.txt and release cinder-temepst-plugin | 15:29 |
| enriquetaso | i think we could wait for some cinder-tempest-plugin so we could release them together, what do you think eharney ? | 15:29 |
| enriquetaso | gmann++ | 15:30 |
| enriquetaso | sounds good to me | 15:30 |
| rosmaita | gmann: thanks, this sounds good | 15:30 |
| eharney | well you have to release tempest first, not together, but yes | 15:30 |
| enriquetaso | ack | 15:30 |
| enriquetaso | thanks gmann | 15:30 |
| gmann | eharney: yes, whenever you are ready for cinder-tempst-plugin release ping me in qa chanel and will do tempest release | 15:31 |
| gmann | and after that you can do plugin release | 15:31 |
| eharney | i may not be me, but yes, sounds good | 15:32 |
| gmann | enriquetaso: as eharney mentioned requirement.txt file needs to have latest release tempest so order can be 1. release tempest 2. update requiremnts.txt file with new released tempest version 2. release cinder-tempest-plugin | 15:32 |
| eharney | it may not* | 15:32 |
| eharney | thanks! | 15:32 |
| gmann | sure | 15:33 |
| enriquetaso | \o/ | 15:33 |
| gmann | rosmaita: enriquetaso eharney whoami-rajat__ I will documents the process in tempest doc also and you can also refer that in your doc. it will be helpful for other plugins also | 15:34 |
| enriquetaso | gmann, that would be fantastic | 15:34 |
| rosmaita | gmann: thanks, that will be helpful ... it is good to know that instead of waiting for a tempest release, we can ask for one when it's needed | 15:34 |
| gmann | sure I will do in next hr or so otherwise we keep forgetting :) | 15:34 |
| gmann | rosmaita: yes. even we end up doing tempest release every week it is all good. | 15:35 |
| rosmaita | yeah, releases are cheap! | 15:35 |
| whoami-rajat__ | gmann++ thanks | 15:37 |
| whoami-rajat__ | and I've a bunch of cinder tempest plugin patches, would like to get them before the release :) | 15:37 |
| whoami-rajat__ | this one had a lot of reviews at one point but got stuck, would be good to get some reviews on this https://review.opendev.org/c/openstack/cinder-tempest-plugin/+/795753 | 15:38 |
| *** whoami-rajat__ is now known as whoami-rajat | 15:38 | |
| gmann | and feel free to add me in cinder-tempest-plugin changes review if any question from tempest side or best practice on plugins using tempest interface | 15:38 |
| whoami-rajat | sure, that would be great help | 15:39 |
| gmann | our best effort is to avoid breaking plugins due to tempest future change and we have improved a lot now which was really bad in past. so using tempest interface in right way if the key | 15:39 |
| gmann | * is the key | 15:39 |
| whoami-rajat | would be helpful to have tempest stable interface for API as well although i don't have many things that depend on it atm | 15:41 |
| gmann | and also as I talked to whoami-rajat, anytime you think this tempest plugin code is good to use in plugins but we need to make tempest a stable interface feel free to reachout to us and we can do that in best possible wa,y. and add that in plugin stable interface - https://docs.openstack.org/tempest/latest/plugins/plugin.html#stable-tempest-apis-plugins-may-use | 15:41 |
| gmann | whoami-rajat: right ^^ we both were thinking same for that | 15:42 |
| gmann | whoami-rajat: we can judge the amount of benefits it will give or so. for example if there is large amount of duplicate code in tempest.volume.base and cinder-tempest-plugin then we can think of making that code as stable interface in tempest | 15:43 |
| whoami-rajat | gmann, yes, I understand it's going to take some amount of work and resources from tempest team so better to do it if we have enough benefits | 15:44 |
| whoami-rajat | we can retrospect it sometime | 15:44 |
| gmann | yeah | 15:50 |
| *** dviroel|lunch is now known as dviroel | 16:05 | |
| Roamer` | Hi, here's a possibly dumb question. We have Glance configured to use Cinder as a backend, and we have specified in the glance-api.conf file that it should use a specific project to keep its images in. If we create a Glance image via a file upload, everything is fine and good, it is created in the correct project. If, however, we try to create a Glance image from a Cinder volume in another | 16:38 |
| Roamer` | project, openstackclient issues a os-volume_upload_image API call (as it should) and... the new volume is cloned correctly *into the same project as the original Cinder volume*... so then Cinder tries to tell Glance "here's the location for your new image", Glance says "let me just do a quick check, give me the volume details for this volume *in my project*" and Cinder goes "nah, there is no | 16:39 |
| Roamer` | such volume in your project" | 16:39 |
| Roamer` | from what I can see in the os-volume_upload_image API call's parameters, there does not seem to be a way to tell it to create the volume in another project | 16:39 |
| Roamer` | so..... how can we create a Glance image from a Cinder volume in a different project? | 16:39 |
| Roamer` | this is all on Xena, BTW | 16:40 |
| Roamer` | when I said "is cloned correctly" above, I meant "the clone operation completes successfully with the parameters it was given" | 16:40 |
| Roamer` | ...into the wrong project. | 16:40 |
| Roamer` | so yeah, what happens later is that Cinder says "okay, this operation failed, let's do the full attach-and-run-dd dance" and that works correctly | 16:43 |
| opendevreview | Gorka Eguileor proposed openstack/os-brick master: NVMe-oF: Consolidate code paths https://review.opendev.org/c/openstack/os-brick/+/836060 | 16:49 |
| opendevreview | Gorka Eguileor proposed openstack/os-brick master: NVMe-oF: Disconnect subsystems https://review.opendev.org/c/openstack/os-brick/+/836062 | 16:49 |
| opendevreview | Gorka Eguileor proposed openstack/os-brick master: Support shared_targets tristate value https://review.opendev.org/c/openstack/os-brick/+/836063 | 16:49 |
| Roamer` | hmmmm, okay, so I just found the image_upload_use_internal_tenant config option | 16:49 |
| Roamer` | ...is that the way to go? should we configure Glance to use the Cinder internal tenant instead of its own project? | 16:50 |
| Roamer` | so, yeah, ignore all this | 17:13 |
| Roamer` | it turned out to be a misconfiguration on our part: a coworker clearly meant to use the internal tenant, but forgot to configure it | 17:14 |
| Roamer` | now everything seems to work | 17:14 |
| Roamer` | actually... everything almost works, I'd forgotten that I'd made a small change a while ago: I think there is another bug related to the URLs that Glance uses to reference Cinder-volume-backed images: when cinder.volume.manager.VolumeManager._clone_image_volume_and_add_location() tries to give cinder://<volume-id> to Glance, Glance does not accept it, since it does not contain a Glance store | 17:42 |
| Roamer` | name (e.g. "main", something like cinder://main/<volume-id>)... and, TBH, I don't know what to do about that, because I'm not sure that Cinder has a way to know the name of the Glance backend to specify :( | 17:42 |
| Roamer` | ...I wonder if it could be made into a configuration option | 17:42 |
| Roamer` | right now, I will modify cinder/volume/manager.py and hardcode cinder://main/<volume-id>, but that doesn't really seem like a viable solution :) | 17:43 |
| Roamer` | maybe I snould file a bug about that, because Glance API v2 certainly does not want to work with cinder://<volume-id> URLs | 17:48 |
| Roamer` | of course, making it into a configuration option would imply that people only want to use a single Cinder-volume-backed Glance backend, which might not be the case - there might be two Glance backends with different settings... and I have no idea how that could possibly work when uploading a volume to a Glance image. | 17:50 |
| whoami-rajat | Roamer`, hey, thanks for finding this issue, it is indeed a valid issue | 18:02 |
| whoami-rajat | Roamer`, we added support for multiple cinder stores in i guess victoria and there have been a bunch of code paths that were not updated to take into account the new location format | 18:02 |
| whoami-rajat | Roamer`, we added a volume type extra specs parameter 'image_service:store_id' but unfortunately it only updates the location metadata and not the URL | 18:03 |
| whoami-rajat | Roamer`, if you can file a bug, i can work on a fix for your issue :) | 18:04 |
| Roamer` | whoami-rajat, okay, I will, thanks a lot! | 18:04 |
| whoami-rajat | Roamer`, np, may i know which cinder backend are you using in your deployment? and are you using multiple glance cinder stores? | 18:06 |
| Roamer` | whoami-rajat, glance-api.conf says enabled_backends = basic:file, readonly:http, main:cinder and in Cinder we use the storpool backend | 18:07 |
| Roamer` | whoami-rajat, so, if I understand you right, we should now add the image_service:store_id=main extra spec to all our volume types | 18:07 |
| whoami-rajat | ack so it's a mix of multiple stores with only one single store | 18:07 |
| Roamer` | and then fix cinder/volume/manager.py to honor that | 18:07 |
| whoami-rajat | Roamer`, yes, that seems to be the right approach | 18:08 |
| whoami-rajat | s/single/cinder | 18:08 |
| Roamer` | (of course, I could also whip up a quick Glance fix to check whether there is only a single :cinder backend defined in the config, but that seems less than ideal :)) | 18:08 |
| whoami-rajat | I think glance is moving towards defaulting multiple stores only so even if you want to configure a single store, it will be in the multiple stores format | 18:10 |
| Roamer` | yeah, I meant something like patch glance_store.multi_backend.get_store_from_store_identifier() to check for store_identifier=None and, if so, go through all the defined backends and see if there is only one that handles that specific scheme... but that seems much, much, much worse than fixing Cinder to fully honor image_service:store_id :) | 18:12 |
| Roamer` | whoami-rajat, thanks for letting me know about that extra spec, I had somehow managed to completely miss it! | 18:13 |
| whoami-rajat | Roamer`, np, thanks for reporting the issue, you found a bug in the original implementation which was targeted to address your usecase, see commit message here: https://github.com/openstack/cinder/commit/350973f3dddd8e3b1d521238c69a5bd8c8df3fec | 18:15 |
| whoami-rajat | Roamer`, will you be proposing an upstream fix or just patch it in your deployment? | 18:16 |
| Roamer` | whoami-rajat, I think I can propose a simple upstream fix, if you'd be okay with that | 18:18 |
| whoami-rajat | Roamer`, sure, let me know and i will review it | 18:19 |
| Roamer` | thanks! | 18:19 |
| Roamer` | I think I can do that a bit later tonight or early tomorrow, I really need to go grab something to eat now | 18:19 |
| whoami-rajat | sure anytime | 18:20 |
| geguileo | rosmaita: the cinderlib gate failure is weird... I can't reproduce it locally with the same dependencies, same python hash seed, and same execution order... | 18:21 |
| Roamer` | whoami-rajat, just one more thing: should I make it so that if the volume type's extra specs do *not* have this setting, then assume that Glance does not use multiple backends and send an old-style cinder://<volume-id> URL, otherwise send a cinder://<store-name>/<volume-id> one? | 18:22 |
| Roamer` | <store-id>, of course | 18:22 |
| whoami-rajat | Roamer`, yes, correct | 18:22 |
| Roamer` | okay, thanks again! | 18:22 |
| rosmaita | geguileo: i don't understand it either | 18:23 |
| whoami-rajat | np | 18:24 |
| rosmaita | geguileo: we haven't had a lot of patches posted to cinderlib, so it's hare to tell when it started: https://zuul.opendev.org/t/openstack/builds?job_name=cinderlib-tox-py39&project=openstack/cinderlib | 18:25 |
| geguileo | rosmaita: what I don't understand is how I can't reproduce the issue | 18:26 |
| geguileo | I've reproduced dozens of gate issues with this same procedure... | 18:27 |
| geguileo | mostly for flaky gate failures... | 18:27 |
| geguileo | and this is no longer intermitent, this is always failing... | 18:27 |
| geguileo | I can only think that it's the Python version, mine is 3.9.12 running on Fedora and that one is 3.9.5 running on Ubuntu | 18:29 |
| rosmaita | i'm running 3.9.5 | 18:31 |
| rosmaita | on Fedora, though | 18:32 |
| geguileo | rosmaita: this is kind of crazy :-( | 18:44 |
| geguileo | rosmaita: I was trying to run it in Ubuntu, but ubuntu focal doesn't seem to have python 3.9 as a standar package from the normal repos | 18:45 |
| rosmaita | maybe it is some kind of packaging problem, then | 18:45 |
| rosmaita | i wonder if we try centos nodes for that job | 18:45 |
| rosmaita | gimme a minute to find the nodeset name | 18:45 |
| rosmaita | i will put up a test patch real quick | 18:46 |
| opendevreview | Brian Rosmaita proposed openstack/cinderlib master: DNM: use centos-9-stream node for py39 https://review.opendev.org/c/openstack/cinderlib/+/845167 | 18:49 |
| rosmaita | geguileo: failure: https://zuul.opendev.org/t/openstack/build/7453076e4bb34676a586d89aecd1294d | 19:06 |
| geguileo | rosmaita: is it me, or those are even more failures? | 19:10 |
| rosmaita | i think one more? | 19:10 |
| rosmaita | there are usually 14 | 19:10 |
| rosmaita | i have seen one with 22 | 19:10 |
| geguileo | rosmaita: nothing, can't make it fail runing on stream-9 either | 19:16 |
| rosmaita | this is very frustrating | 19:16 |
| geguileo | rosmaita: Since I can't reproduce it, I'll try reading the tests and the code tomorrow | 19:21 |
| rosmaita | sounds good | 19:22 |
| rosmaita | i think i figured out the problem with the py36 job, it wants to use the latest importlib-metadata (4.11.4) and that requires python >= 3.7 ... i think it's the same problem i was having yesterday with the requirements job, cinderlib master != zed development | 19:37 |
| tosky | rosmaita: I hit that issue somewhere else (internal stuff) but is there a reason why you need 4.11.4? IIRC the version of py36 is 4.8.something | 19:38 |
| rosmaita | tosky: i don't want 4.11.4 at all, but see the bottom of https://zuul.opendev.org/t/openstack/build/430adc04b01145b3bc6b63837e915386/console#4/0/10/ubuntu-bionic | 19:40 |
| rosmaita | i think i need to override checkout openstack/requirements to stable/yoga so it uses the correct constraints file | 19:43 |
| opendevreview | Brian Rosmaita proposed openstack/cinderlib master: DNM: use stable/yoga requirements for unit test jobs https://review.opendev.org/c/openstack/cinderlib/+/845170 | 19:47 |
| opendevreview | Walt proposed openstack/cinder master: Fix and unify capacity calculations https://review.opendev.org/c/openstack/cinder/+/831247 | 19:50 |
| *** dviroel is now known as dviroel|afk | 20:27 | |
| opendevreview | Walt proposed openstack/cinder-specs master: WIP: Add new volume transaction tracking https://review.opendev.org/c/openstack/cinder-specs/+/845176 | 21:05 |
| hemna | geguileo ^^ I did a rough initial draft of the spec for the idea of volume transaction tracking. Feel free to read/modify/discuss. | 21:05 |
| hemna | ultimately I want to be able to run 'cinder top' | 21:06 |
| hemna | and see a live view of what cinder is doing right now. | 21:06 |
| hemna | first step towards graceful shutdown and restart of transactions as well. | 21:06 |
| hemna | I haven't thought everything out, but wanted to have a WIP spec up for review so we can all have a starting point. | 21:07 |
| hemna | rosmaita as well ^^ | 21:07 |
| rosmaita | hemna: nice | 21:08 |
| hemna | It would be kinda cool to have an external project for transaction tracking, then nova and others could use it as well and one could run 'openstack top' | 21:09 |
| hemna | and see what all the services are doing at any moment. | 21:09 |
| opendevreview | Brian Rosmaita proposed openstack/cinderlib master: DNM: use stable/yoga requirements for CI jobs https://review.opendev.org/c/openstack/cinderlib/+/845170 | 21:11 |
| opendevreview | Jean Pierre Roquesalane proposed openstack/cinder master: WIP: Add Cinder NFS driver for Dell EMC PowerStore https://review.opendev.org/c/openstack/cinder/+/797608 | 21:14 |
| opendevreview | Peter Penchev proposed openstack/cinder master: Send the correct location URI to the Glance v2 API https://review.opendev.org/c/openstack/cinder/+/845178 | 22:55 |
| opendevreview | Peter Penchev proposed openstack/cinder master: Send the correct location URI to the Glance v2 API https://review.opendev.org/c/openstack/cinder/+/845178 | 22:55 |
| Roamer` | (of *course* I will notice a stray blank line immediately *after* pushing to Gerrit... some things never change) | 22:56 |
| opendevreview | Peter Penchev proposed openstack/cinder master: Send the correct location URI to the Glance v2 API https://review.opendev.org/c/openstack/cinder/+/845178 | 23:10 |
| opendevreview | Brian Rosmaita proposed openstack/cinderlib master: Use stable/yoga override for CI jobs https://review.opendev.org/c/openstack/cinderlib/+/845170 | 23:21 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!