14:00:05 #startmeeting cinder 14:00:05 Meeting started Wed Nov 15 14:00:05 2023 UTC and is due to finish in 60 minutes. The chair is rosmaita. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:05 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:00:05 The meeting name has been set to 'cinder' 14:00:08 Sorry for not being present sooner, but I was (I still do) firefighting... 14:00:18 o/ 14:00:30 #topic roll call 14:00:30 #link https://etherpad.opendev.org/p/cinder-caracal-meetings 14:00:32 o/ 14:00:41 i see some of you have anticipated the roll call! 14:00:43 o/ 14:00:43 o/ 14:01:07 Happy Diwali to anyone who celebrates, Rajat is off today, so I'll be running the meeting 14:01:17 o/ 14:01:22 o/ 14:01:35 o/ 14:02:22 ok, let's get started 14:02:28 #topic announcements 14:02:37 o/ 14:02:38 gerrit offline for about an hour on Friday beginning 15:30 UTC 14:02:53 #link https://lists.opendev.org/archives/list/service-announce@lists.opendev.org/thread/XT26HFG2FOZL3UHZVLXCCANDZ3TJZM7Q/ 14:03:05 those ^^ are the details if you want 'em 14:03:21 but supposed to be a quick gerrit update 14:03:48 next, call for proposals for FOSDEM 14:03:58 #link https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/AFQ2NKJFG3J64XE54JQQ22SXS2DWLT4U/ 14:04:13 o/ 14:04:17 there's info in that email ^^ and also on todays meeting agenda 14:04:36 key date is submission deadline is 8 December, so basically 3 weeks away 14:05:02 next, results from the OpenStack user survey 14:05:10 #link https://docs.google.com/spreadsheets/d/1ehlabuqOnNK4M7xtN1GXvhLoeHn4JBaISih6RuzLLyQ/edit#gid=0 14:05:26 i haven't looked myself, but Rajat made some notes 14:05:34 Most deployments seems to be using Ceph RBD storage -- also saw one instance of Ceph iSCSI 14:05:48 Might be useful to discuss in the video meeting end of the month 14:06:20 so take a look when you have a chance and we can talk about it in 2 weeks 14:06:24 Which line shows the driver usage? 14:06:39 don't know, i haven't looked 14:07:02 if it's anything like the past survey results, the data is all over the place and you have to dig to find it 14:07:11 oh - i see it - Colmn EF 14:07:13 (but i may be maligning the foundation there) 14:07:51 ok, next item 14:08:01 tomorrow is the caracal-1 milestone 14:08:09 #link https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/ZRRI3A5PJFIHTM6GPUSYMJSTWMTWKYAI/ 14:08:42 this would be a good time to go back over the PTG summary and look for priorities/action items that were supposed to be done by M-1 14:09:05 i think the sqlalchemy 2.0 patches are one item, jbernard has an item on the agenda for later 14:09:33 the next item is a concern for all of openstack 14:09:40 Eventlet not compatible with python 3.12 14:09:57 #link https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/YO5CZDVAJ6QSF734ALWSGNOQDDAIOXKI/ 14:10:40 something will have to be done by someone! 14:11:23 at this point, i imagine that openstack will encourage contributions to eventlet to whip it into shape 14:11:46 but we will probably also be looking for a better future solution 14:12:17 it didn't come up at the TC meeting yesterday, so there's not an official position yet, other than start worrying 14:12:43 something to keep an eye on 14:12:53 * jbernard activates worrying now 14:13:00 :) 14:13:12 it might be helpful if someone can do some PoC but I am (and I know we all) aware it's a huge effort to replace it now :-( 14:13:32 it may have to be done in oslo.services first so it may take a while until the topic is actionable for cinder, though 14:13:47 yes, i think repair + plan ahead for a replacement is the approach that we'll take 14:14:32 ok, last item is about Unmaintained branches 14:14:42 #link https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/ZYAZG43BLJJVXCYZVPYQX5733BYDVVNL/ 14:14:57 Rajat put a summary of the key points on the agenda etherpad 14:15:44 so basically, that email just summarizes how we're going to actually do the Unmaintained transition 14:16:02 except for one point, it's the same proposal that was passed in July 14:16:42 the one point is proposal that there be an openstack-wide group that has +2/+W powers on all unmaintained/* branches 14:16:52 i mean, that's the change 14:17:14 the original was that every project would have its own for example cinder-unmaintained-core group 14:17:45 but it seems better to have a single group that can take care of this 14:17:52 (at least to me) 14:17:52 question: what about releases on which OSP is based like wallaby for OSP 17.x which is before yoga 14:18:09 can we still do backport? 14:18:31 yes, those will still be open, and the usual backport rules apply 14:18:40 ok 14:18:44 (so no features, just low-risk bug fixes) 14:19:04 if you want more than that, then you will have to raise a BZ 14:19:04 Ericsson has volunteered to keep unmaintained/victoria open 14:19:49 it is a crucual version for their 5G deployments 14:20:02 s/crucual/crucial 14:20:11 thks 14:20:11 happystacker: simondodsley is correct, downstream projects will continue to follow their own rules 14:20:26 ok perfect 14:20:32 so for OSP, you'd have to do the Red Hat process 14:20:49 backport first, then a BZ 14:21:03 in any case, we'll soon have victoria->yoga in Unmaintained status 14:21:12 will be interesting to see how this works out 14:22:14 oh, i should mention as a follow up, i abandoned the EOL patches for victoria, wallaby, and xena for cinder 14:22:25 but train and ussuri are still ready to be EOL'd 14:22:51 that's all the announcements ... anyone have something else to share? 14:23:42 I have another question about OSP 14:24:10 sure 14:24:28 If I want to backport a patch to OSP17.X, which branch should I commit, wallaby or master? 14:25:22 well, ideally you will fix something in master, and then backport it as far back as you can, which would be to wallaby 14:25:31 all upstream 14:25:48 I see. Thank you! 14:25:52 but to actually get the change into OSP, you would have to file a BZ with Red Hat 14:26:29 if you have more questions, we should discuss offline, because our focus in this meeting is the upstream community 14:26:33 Thank you, rosmaita 14:26:59 oh, one final item 14:27:18 i just noticed that this friday, 17 November, is the third friday of the month 14:27:29 does anyone know why that is important? 14:27:55 festival of review? 14:27:57 reviews! 14:28:03 lol 14:28:04 yes!!! 14:28:10 you beat me eharney 14:28:15 #link https://meetings.opendev.org/#Cinder_Festival_of_XS_Reviews 14:28:32 so, please mark your calendar, and I'll see you there! 14:28:56 ok, on to the actual topics 14:29:13 #topic Nova removing VMWare support 14:29:36 Rajat added the topic but because I proposed the topic initially so I can provide some cntext 14:29:39 * context 14:29:49 thank you, go ahead 14:30:03 Nova has maintained several virt drivers, but they are now removing unmaintained ones 14:30:23 vmware driver was marked experimental and is being removed this cycle. 14:31:22 there are some features for VMware platform in the other projects and I'm wondering if we can deprecate these all in this cycle and remove it after 2024.1 release 14:31:44 for context, the vmware volume driver is only useful if using vmware compute, right? 14:32:19 i think you are right - only used in VMware Integrated OpenStack (VIO) 14:32:28 I believe so 14:32:34 we should check with hemna 14:32:47 i think he runs openstack on vmware? 14:33:06 i think that is correct 14:33:43 i have the same question as eharney, if the storage driver is only useful with the compute driver, and the compute driver is being removed, then this is a no-brainer 14:33:57 but otherwise, this may need some brain work 14:34:04 I already raised the same topic for glance during the previous PTG and we agreed to send an email to find anyone using it without VMWare. so far I've not heard any feedback to it. 14:34:32 I beleive that VIO may be on the verge of demise anyway with the Broadcom buyout 14:34:40 wondering if we can do the same, and wait for some time (like m-2) and then deprecate the driver if we hear no objections. 14:35:47 I mean, I can send an email to ask if anyone needs the vmdk driver without vmware, and if we hear nothing then we can deprecate the driver. 14:35:58 tkajinam: that would be great 14:36:05 I believe the driver is useless without vmware virt driver but just in case I'm wrong... 14:36:10 the driver requires a vcenter server to use 14:36:24 yeah that's what I read 14:36:48 i always thought of a "vcenter server" as being a kind of storage appliance 14:37:00 but, i do not know what i am talking about 14:37:12 VIO only officially supports that one driver if I read their docs correctly 14:37:13 it's the vmware virt platform (which probably has a different name by now) 14:37:17 i don't think Walt is here? hemna?? 14:37:30 that's a kind of centralized endpoint managing multiple ESXi, afair > vCenter 14:37:52 tkajinam: correct 14:38:05 If anyone here knows someone familiar with this driver or VIO then that would be great, as I can double check my observation with him/her 14:38:23 but anyway I'll send out an email and the bring this topic after some time, surely during this cycle. 14:38:26 #action tkajinam to send an email to the ML similar to his glance_store email, but for cinder 14:38:50 note that the feature in os-brick may be actually useless once we deprecate/remove vmdk driver so I don't feel like there may be any discussions needed for it. 14:39:35 ok, thanks tkajinam 14:39:48 maybe we'll know something in time for the first midcycle 14:39:56 anything else? 14:40:11 nothing from my end 14:40:56 #topic Cinderlib 2023.2: Cycle trailing release deadline 14:41:08 #link https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/RPTPVHSFSK6PXJGRV775EFVZ4Y4DX2SM/ 14:41:18 release deadline is 7 December 14:42:01 just a reminder, this version of cinderlib that will be released is the one that works with Bobcat (2023.2) 14:43:00 the CI jobs are running against stable/2023.2, so hopefully we would pick up incompatibilities 14:43:41 anyway, if you are a cinderlib user, it would be good to try out cinderlib master to make sure it works with Bobcat 14:44:05 i will leave a marker here for geguileo ^^ because cinderlib is kind of his baby 14:44:33 #topic SQLAlechemy 2.0 14:44:47 jbernard has been tracking this one 14:45:05 i have put a list of the remaining patches in the etherpad 14:45:18 this was a milestone-1 priority, so let's try to get this done in the next day or so! 14:45:34 but tldr; 1 is waiting on recheck, 5 could use a person to review 14:46:03 ok thanks, and thanks for the list in the etherpad 14:46:07 i think 1 of those 5 i need to respond to, but if anyone has time or db knowledge, we could finish these up quickly 14:46:16 line 147 in https://etherpad.opendev.org/p/cinder-caracal-meetings 14:46:28 (though it may not stay at line 147 for long!) 14:47:05 thanks jbernard ... like i said, it's a priority for M-1, so please make it a priority to review these 14:47:22 #topic windows support deprecation 14:47:26 tkajinam: that's you again! 14:47:52 this is another "kill unmaintained item" topic but an easy one 14:48:47 there are a few Windows related drivers in cinder, which are dependent on os-win. As you know Winstackers project who maintained os-win has been retired and these should be removed 14:49:10 (another option is os-win should be replaced or picked up but I don't think anyone may do these 14:49:49 I've proposed the patches to deprecate windows support in os-brick and cinder and would like to get these merged in 2024.1 14:49:50 yeah, microsoft is involved with the Foundation, but apparently they have no interest in openstack running on windows 14:49:53 so that we can drop these in 2024.2 14:50:33 ok, thanks for posting those 14:50:45 one note about my patch is that I didn't add warning for every single os.name == 'nt' to avoid showing multiple multiple warnings 14:51:26 but I added warnings to core parts (I believe) so these would be enough 14:51:32 along with the release note 14:51:44 ok, that sounds reasonable 14:52:16 tkajinam: what's the patch? 14:52:29 https://review.opendev.org/c/openstack/cinder/+/894237 14:52:29 https://review.opendev.org/c/openstack/os-brick/+/900913 14:52:30 these two 14:52:51 (I added these to the etherpad as well 14:53:07 there's some code in cinder/cmd/* to remove too 14:53:11 tkajinam: thanks 14:53:24 eharney: that's why I was asking, but I see the patch is not removing code 14:53:36 It's just giving warnings in preparation of removal 14:54:07 yes. my intention is to only deprecate these during this cycle 14:54:15 and then remove it after 2024.1 release 14:54:19 to follow SLURP 14:54:37 2024.1 deprecation 2024.2 removal I mean 14:55:59 right, the deprecation must happen in a SLURP release 14:56:25 +1 14:57:06 tkajinam: anything else? 14:57:51 ok, next item, everyone's favorite 14:57:56 #topic review requests 14:58:21 4 requests this week, looks like most have good feedback so far, so please put them on your list to review 14:58:36 and don't forget about the Festival of Reviews on Friday, 1400 UTC 14:58:50 #topic open discussion 14:58:55 ok 90 seconds left 14:59:00 anyone? 14:59:05 The primary patches under review are related to SQLAlchemy 2.0. Regarding to the structure of [Review Requests] section, it seems that patches related to new features can only be added after M2 is passed, right? I have quite a few feature-related patches waiting for review. Do you have any suggestions on how I can expedite this process? 15:00:12 Hi, forgot to add it to the doc, but please remember to send the mail about the metadata length rosmaita 15:00:27 drencrom: thanks for the reminder 15:01:11 inori: maybe attend the Festival on friday, traditionally each attendee can nominate a patch, so that can get one reviewed kind of quickly 15:01:24 ok, we are out of time 15:01:28 thanks everyone! 15:01:31 #endmeeting