14:00:29 <rosmaita> #startmeeting cinder 14:00:29 <opendevmeet> Meeting started Wed Sep 25 14:00:29 2024 UTC and is due to finish in 60 minutes. The chair is rosmaita. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:29 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:00:29 <opendevmeet> The meeting name has been set to 'cinder' 14:00:29 <rosmaita> #link https://etherpad.opendev.org/p/cinder-dalmatian-meetings 14:00:41 <rosmaita> #topic roll call 14:01:33 <whoami-rajat> hi 14:01:40 <eharney> hi 14:01:40 <akawai> o/ 14:02:07 <kpdev> hi 14:02:29 <rosmaita> #link https://etherpad.opendev.org/p/cinder-dalmatian-meetings 14:02:44 <rosmaita> (Jon has a conflict today, so I'm running the meeting) 14:03:38 <whoami-rajat> rosmaita, should we also do the ping on the cinder channel? 14:03:45 <rosmaita> i thought i did 14:03:49 <jungleboyj> o/ 14:04:01 <rosmaita> though maybe i pinged in glance by mistake 14:04:04 <whoami-rajat> oh yes you did, i had the wrong username sorry 14:04:51 <rosmaita> well, it doesn't hurt to ping twice 14:05:05 <rosmaita> ok, i guess we can get started 14:05:06 <tosky> o/ 14:05:16 <rosmaita> #topic schedule updates 14:05:23 <rosmaita> Epoxy development starts next week 14:05:30 <rosmaita> #link https://releases.openstack.org/epoxy/schedule.html 14:05:38 <rosmaita> (actually it has already started, we'll talk about that in a minute) 14:05:49 <rosmaita> back to the present 14:05:55 <rosmaita> #link https://releases.openstack.org/dalmatian/schedule.html 14:06:04 <rosmaita> this is final-rc week 14:06:11 <rosmaita> #link https://releases.openstack.org/dalmatian/schedule.html#d-finalrc 14:06:21 <rosmaita> just a reminder of how the OpenStack development process works 14:06:32 <rosmaita> we released 2024.2 (Dalmatian) Release Candidate 1 (RC1) two weeks ago 14:06:39 <rosmaita> #link https://review.opendev.org/c/openstack/releases/+/928519 14:06:52 <rosmaita> when ^^ merged, the stable/2024.2 (dalmatian) branch was cut 14:07:00 <rosmaita> master became the epoxy development branch 14:07:15 <rosmaita> everything we work on always goes to master first 14:07:23 <rosmaita> so if you do a git log on master, you'll see a bunch of stuff has merged after the 25.0.0.0rc1 tag 14:07:31 <rosmaita> that's all Epoxy 14:07:42 <rosmaita> the only way to get it into the Dalmatian release is to backport it from master to stable/2024.2 14:07:52 <rosmaita> We will need to do a RC-2 to include https://review.opendev.org/#/q/I0b13c6351733 14:07:59 <rosmaita> (which has already merged to master and all earlier stable branches) 14:08:08 <rosmaita> the only other patches currently merged to master are test-only-changes 14:08:16 <rosmaita> so no need to backport those (at least not now) 14:08:27 <rosmaita> ok, the point of all that is that if anything else is to be included in RC2 this week 14:08:34 <rosmaita> it has to merge to master today 14:08:41 <rosmaita> and be backported to stable/2024.2 14:08:50 <rosmaita> i've seen 1 patch that i think is a good candidate 14:08:57 <rosmaita> (but people can nominate others) 14:09:06 <rosmaita> it's the StorPool patch that's on the agenda 14:09:13 <rosmaita> #link https://review.opendev.org/c/openstack/cinder/+/847536 14:09:52 <rosmaita> so it will need one more +2 and merge, and then kpdev can propose the backport 14:10:10 <kpdev> yes 14:10:47 <rosmaita> my reasons for nominating it is that it's been sitting a long time, and it has implications for glance and multiattach 14:11:07 <rosmaita> so i think it would be good to have it in the Dalmatian release 14:11:33 <whoami-rajat> are we adding iSCSI support to storpool driver? 14:12:28 <kpdev> for attach whenever possible configured using options 14:12:52 <whoami-rajat> to me this looks like a new feature, and storpool should do it like how other drivers do it 14:13:05 <whoami-rajat> i.e. create a common file for common methods 14:13:19 <whoami-rajat> and separate code for iSCSI, FC or any other protocol that they support 14:13:38 <whoami-rajat> we have examples in the netapp and pure driver for this, those support iSCSI, FC and NVMe 14:14:22 <rosmaita> i thought pure was a single-file driver 14:14:32 <rosmaita> (one of the "legacy" cinder drivers) 14:14:37 <kpdev> isn't its driver specific ? 14:14:59 <eharney> pure is single-file but different classes for iscsi/fc drivers 14:15:05 <whoami-rajat> rosmaita, pure has a single file but different classes for iSCSI and FC drivers 14:16:22 <whoami-rajat> i don't like the initialize_connection or other methods returning different protocols based on some config options, looks very error prone 14:16:26 <whoami-rajat> but that's just me 14:16:27 <eharney> re: storpool i'm also not a huge fan of adding very generic sounding options (like iscsi_cinder_volume/iscsi_portal_group) which are driver-specific 14:17:30 <whoami-rajat> eharney, +1, i think we have a precedent of driver specific config option names starting with the driver prefix, like storpool_iscsi_cinder_volume or something 14:17:39 <kpdev> but these options are in driver section 14:17:44 <eharney> yeah, other options do that there, but the new ones don't 14:17:47 <whoami-rajat> to avoid conflicts with other config options 14:18:17 <rosmaita> ok, sounds like there's some pushback, but it would be good to have a commitment to get comments on the review so it can be revised for early Epoxy 14:19:02 <whoami-rajat> sure, i can add my concerns 14:19:03 <kpdev> ok so changing options to storpool_* should be fine for this patch to get in ? 14:19:36 <rosmaita> kpdev: sounds like there may be more refactoring involved 14:20:06 <rosmaita> so looks to me like not for RC2 14:20:26 <rosmaita> but, whoami-rajat will get comments on the patch (and maybe eharney too) 14:20:48 <rosmaita> and if you have questions about the refactoring request, you can put it on the agenda for the ptg in a few weeks 14:21:06 <kpdev> ok, isn't it possible to merge this as experimental feature support in rc2 and do refactoring as asked for epoxy ? 14:21:08 <rosmaita> but maybe Rajat's comments plus looking at the Pure driver will give you enough ideas 14:22:22 <rosmaita> that's really a question for the PTL 14:22:42 <rosmaita> i'm inclined to say no 14:22:54 <eharney> if you do that, you have to then handle compatibility because changing from an option to enable iscsi to a different driver class is a significant change 14:23:00 <eharney> would be kind of a mess 14:23:11 <kpdev> ok 14:23:31 <rosmaita> kpdev: sorry about getting your hopes up 14:23:44 <rosmaita> on the plus side, you did get some more people looking at the patch 14:24:02 <rosmaita> all right, any other nominations for RC2? 14:24:59 <whoami-rajat> not related to RC2 but i had concerns with one of the changes that storpool driver did some time ago, should i hold it till open discussion? 14:25:15 <rosmaita> yes, we will have plenty of time 14:25:22 <whoami-rajat> ok thanks 14:25:34 <rosmaita> ok, I will be able to post the RC2 patch right after the meeting, then 14:25:42 <rosmaita> moving on ... 14:26:04 <rosmaita> the Epoxy virtual PTG will be the week of October 21 14:26:20 <rosmaita> i'm sure Jon has an etherpad, but I can't find a link 14:26:25 <rosmaita> anyone got one? 14:27:30 <eharney> https://etherpad.opendev.org/p/epoxy-ptg-cinder 14:27:34 <rosmaita> thanks! 14:28:10 <rosmaita> please think about topics and add them to the etherpad 14:28:33 <rosmaita> also, feel free to propose something like a review day 14:28:40 <rosmaita> we did one last PTG for backup reviews 14:28:49 <rosmaita> and, finally, speaking about reviews 14:29:12 <rosmaita> we have a monthly "Festival of Xtra-Small Reviews" the third friday of each month 14:29:30 <rosmaita> #link http://eavesdrop.openstack.org/#Cinder_Festival_of_XS_Reviews 14:29:45 <rosmaita> attendance has been kind of ... not good 14:29:58 <rosmaita> please think about putting it on your calendar 14:30:17 <rosmaita> and if the time is no good, you could propose an alternative at the weekly meeting 14:31:37 <rosmaita> ok, that's it from me ... there's a list of reviews on the etherpad, please look at Luigi's zuul topics at line 127 on the etherpad first 14:31:55 <rosmaita> #topic open discussion 14:32:17 <rosmaita> whoami-rajat: you had something you wanted to talk about 14:32:22 <whoami-rajat> yes 14:32:30 <whoami-rajat> #link https://review.opendev.org/c/openstack/cinder/+/847131 14:32:45 <whoami-rajat> i wanted to discuss about the cross pool cloning functionality that storpool added 14:33:00 <whoami-rajat> i mean the support to report that the backend supports it 14:33:08 <whoami-rajat> I've a comment there 14:33:12 <whoami-rajat> #link https://review.opendev.org/c/openstack/cinder/+/847131/comment/8e807a49_4762ba8e/ 14:33:53 <whoami-rajat> so if we have 2 storage arrays and we configure 2 cinder backends, each pointing to a different array, this logic says you can clone the volume but in reality you cannot and the operation will fail 14:34:24 <whoami-rajat> this is a case which i think was missed, reading the reply, looks like that's not a configuration that storpool supports 14:34:38 <whoami-rajat> I'm not sure why since it should be possible theoretically 14:35:06 <whoami-rajat> anyways, the code is common to all cinder drivers so if another vendor comes in with multiple storage arrays, this logic might not behave correctly for them 14:35:39 <whoami-rajat> to be specific, this is the image volume cache functionality and we will only see this in action when we are creating a bootable volume from an image 14:37:36 <whoami-rajat> right now no other driver than storpool reports clone_across_pools so it's fine but I think NetApp was also trying to add support for the same 14:38:05 <rosmaita> yeah, i am glad you flagged that 14:38:10 <msaravan> Yes, the coding is in progress for NetApp as well.. 14:38:24 <msaravan> have some doubts, and we can connect offline next week.. 14:38:37 <whoami-rajat> sure 14:38:44 <rosmaita> also, there's the ptg 14:39:58 <rosmaita> whoami-rajat: why don't you put up a patch with a "FIXME" comment before line 65 in that file, so that if we all get hit by buses in the next week, there will be an indicator in the code that this needs to be addressed 14:41:00 <whoami-rajat> rosmaita, sure i can do that, i also have the fix but no testing environment to validate it 14:41:05 <whoami-rajat> but i like the FIXME idea 14:41:48 <rosmaita> having an actual fix is even better! 14:42:16 <rosmaita> ok, anybody else have something to discuss? 14:44:05 <rosmaita> sounds like "no" 14:44:21 <rosmaita> ok, see you here next week, and we can celebrate the Dalmatian release 14:44:26 <rosmaita> #endmeeting