Wednesday, 2026-07-08

opendevreviewjayaanand borra proposed openstack/cinder master: NetApp ZAPI: Replace urllib with requests library in NaServer  https://review.opendev.org/c/openstack/cinder/+/99228905:07
opendevreviewNilesh Thathagar proposed openstack/cinder master: Dell PowerFlex Driver: Fix NVMe/TCP host creation with long host names  https://review.opendev.org/c/openstack/cinder/+/99639205:52
opendevreviewSusmita Poddar proposed openstack/cinder master: HPE 3par - support for autoSynchronous as part of RCG creation  https://review.opendev.org/c/openstack/cinder/+/96715805:57
opendevreviewjayaanand borra proposed openstack/cinder master: NetApp ZAPI: Replace urllib with requests library in NaServer  https://review.opendev.org/c/openstack/cinder/+/99228907:57
opendevreviewjayaanand borra proposed openstack/cinder master: NetApp ZAPI: Replace urllib with requests library in NaServer  https://review.opendev.org/c/openstack/cinder/+/99228908:05
opendevreviewNilesh Thathagar proposed openstack/cinder master: Dell PowerFlex Driver: Fix NVMe/TCP host creation with long host names  https://review.opendev.org/c/openstack/cinder/+/99639208:12
opendevreviewAnoop Kumar Shukla proposed openstack/cinder master: NetApp: enrich periodic Cinder EMS payloads  https://review.opendev.org/c/openstack/cinder/+/99620510:08
Anoop_Shukla@rosmaita, Jon, Eric. I recently looked at a patch which changes driver sleep from green thread.sleep() to time.sleep(). https://review.opendev.org/c/openstack/cinder/+/950695/11/cinder/volume/drivers/netapp/dataontap/client/api.py11:31
Anoop_ShuklaAlthough OpenStack documentation recommends using greenthread.sleep(): https://docs.openstack.org/nova/14.0.4/threading.html. So I am confused on how this change affects the driver code. Since the changes in above patch are across drivers, and after reading some documents, it ends up pausing the OS thread - what is the recommendation here? Since the changes are on api.py, it would affect all our concurrent operations which 11:33
Anoop_Shuklauses REST APIs on platform.11:33
opendevreviewRajat Dhasmana proposed openstack/cinder-specs master: Propose replication v3 spec  https://review.opendev.org/c/openstack/cinder-specs/+/98595212:11
opendevreviewAnoop Kumar Shukla proposed openstack/cinder master: NetApp: enrich periodic Cinder EMS payloads  https://review.opendev.org/c/openstack/cinder/+/99620513:12
opendevreviewNilesh Thathagar proposed openstack/cinder master: Dell PowerFlex Driver: Fix NVMe/TCP host creation with long host names  https://review.opendev.org/c/openstack/cinder/+/99639213:39
rosmaitaCinder meeting in #openstack-cinder at 1400 UTC13:59
rosmaitajungleboyj rosmaita smcginnis tosky whoami-rajat m5z e0ne geguileo eharney jbernard hemna fabiooliveira yuval tobias-urdin adiare happystacker dosaboy hillpd msaravan sp-bmilanov Luzi sfernand simondodsley  zaubea nileshthathagar flelain wizardbit agalica lutimura kaisers aloke_dev Anoop_Shukla erlon Jyotsna 13:59
whoami-rajathi14:00
simondodsleyo/14:00
jungleboyjo/  Lurking.14:00
rosmaitawait, let me start14:00
Jyotsnao/14:00
auniyalo/14:00
hvlcchao1o/14:00
rosmaita#startmeeting cinder14:00
opendevmeetMeeting started Wed Jul  8 14:00:24 2026 UTC and is due to finish in 60 minutes.  The chair is rosmaita. 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 'cinder'14:00
whoami-rajathi (again)14:00
simondodsleyo/14:00
rosmaita#link https://etherpad.opendev.org/p/cinder-hibiscus-meetings14:00
rosmaita#topic roll call14:00
rosmaitaok, now you can say hi!14:00
rosmaitahi14:01
jungleboyjo/  Hello14:01
jayaanando/14:01
fqio/14:01
eharneyhi14:01
rosmaita#chair jbernard14:02
opendevmeetCurrent chairs: jbernard rosmaita14:02
rosmaitasomething came up and Jon will be delayed today14:02
agalica_o/14:03
rosmaitaok, guess we should get started14:04
rosmaita#topic announcements14:04
rosmaitaJust a reminder that the Festival of Reviews is this friday, at the end of this week14:04
rosmaitait got postponed due to a holiday in the usa14:05
rosmaitaand, the periodic weekly jobs are all green for cinderclient and cinderclient-brick-ext14:05
rosmaitaso, hooray14:05
whoami-rajat\o/14:06
jungleboyj\o/14:06
rosmaitaand that's all i've got, though i imagine there are some deadlines lurking14:06
rosmaitawell, scrolling down to june 10 in the agenda, looks like the third-party-CI-must-be-working-for-all-protocols deadline was July 314:07
rosmaitaso hopefully everyone has their mojo working14:07
rosmaitathe new driver deadline is coming up pretty soon, July 2414:08
rosmaitaanyone else have an announcement they would like to share?14:09
rosmaitaapparently not ... let's move along14:11
rosmaita#topic Follow up on Strategy for security issues14:11
rosmaitaok, so last week we took an initial look at14:12
rosmaita#link https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/44MLENT3AXUML7CYYLNWO6Z7LO6CQT6W/14:12
rosmaitawe've got several things going on, i will try to summarize the issues14:13
rosmaita1. cinder needs a document telling operators about "safe" vs. "unsafe" deployment configurations14:13
Anoop_Shuklao/14:14
rosmaita2. cinder needs a document for driver developers outlining expectations for security-conscious driver development14:14
rosmaita3. maybe we need a driver security matrix, or a column in the current matrix if we can jam all the info in there14:14
rosmaita4. we need a policy to mark drivers that are unsafe security-wise as 'unsupported' (though maybe that's part of #2)14:16
rosmaita5. we need to encourage vendor drivers to fix the "bad" bugs, e.g., hard-coded 'http' right away14:16
rosmaita6. we need to encourage vendor drivers to change default settings for options to secure values14:17
rosmaitai think that's all?14:17
rosmaitabefore opening this up to discussion, my recommendation is that we do #4 first14:18
rosmaitathat way we will have some encouragement for vendors to prioritize 5 and 614:18
jungleboyjThat would make sense as the starting point and to identify where the problems are.14:19
rosmaitaand it will give us a preliminary document for #214:19
rosmaitaok, let's discuss14:19
whoami-rajatmaybe can use the driver review skill to document safe vs unsafe practices that drivers are doing and document them https://review.opendev.org/c/openstack/agentic-workflows/+/99550214:20
rosmaitai think that is a good idea14:20
whoami-rajatit uses the cinder coresec persona for finding out potential vulnerabilities https://review.opendev.org/c/openstack/agentic-workflows/+/994081/4/personas/cinder-coresec.md14:21
simondodsleyagreed - i'll have a look at the skill and add any comments14:21
rosmaitai don't want to get too far off topic, but whoami-rajat have you mentioned to the cinder community that you have those patches up for the agentic-AI stuff?14:21
whoami-rajatthanks simondodsley 14:21
whoami-rajatrosmaita, no i didn't, can do it quickly to not take a lot of time14:22
whoami-rajatI've proposed series of patches to add cinder specific knowledge, personas and skills https://review.opendev.org/q/topic:%22cinder-agentic-work%2214:22
whoami-rajatsorry14:22
whoami-rajat#link https://review.opendev.org/q/topic:%22cinder-agentic-work%2214:22
whoami-rajatsurely they are long md files as of now but will try to use/modify/iterate to optimize them further14:23
whoami-rajatfeel free to take a look14:23
whoami-rajatthat's all! thanks14:23
rosmaitathanks!14:23
rosmaitanote to the community -- you may not have seen them because they're not proposed to the cinder repo14:24
rosmaitabut please take a look14:24
simondodsleyon it14:25
rosmaitaeven if you don't use AI agents, the md files are pretty human-readable, so you can see what they are doing and comment14:25
whoami-rajatyeah, overtime I'm planning to migrate the knowledge to cinder docs (that is also beneficial for human readers) and try to use links as much as possible in the agentic workflow project14:25
rosmaitaok, back on topic ... i think whoami-rajat's agentic AI stuff will help us identify problematic drivers14:25
rosmaitabut we still need to have a specific policy about what to do about problematic drivers14:26
whoami-rajatI can provide some input by next week using that (but don't rely too much on it!)14:26
rosmaitai will take an action item to get something posted before the Festival of Reviews14:27
simondodsleyreal security issues like http only or lack of TLS support should be, IMHO, be flagged as `unsupported` until fixed14:27
rosmaitai think discussion will be easier once there is something posted to criticize14:27
rosmaitasimondodsley: agree14:28
simondodsleyi don't think lack of CHAP for iSCSI is an immeadiate `unsupported` as i don't think people really use it14:28
aloke_devHi Reviewers, my apologies if it is off topic, but I wanted to check regarding PR: https://review.opendev.org/c/openstack/cinder/+/986072 (Introducing Cinder Driver for InfoScale). We have enabled the 3rd party CI and it is green. Also this PR has got +1. Waiting for feedback from reviewers to move this toward +2.  14:28
simondodsleyit should be the next iteration though14:28
rosmaitayeah, i think we will start with the real bad stuff and then iterate14:29
simondodsleyflipping CHAP to default on is a risk without a lot of notice14:29
rosmaitaaloke_dev: let's hold that for open discussion (in about 15 minutes)14:29
rosmaita#topic RBAC enforcement in devstack14:30
simondodsleyoh yeah - that was fun...14:30
rosmaitalooks like there are some changes that need to be reviewed; luckily we have a Festival at the end of the week14:30
rosmaitasimondodsley: yes, thank you for identifying that barbican issue14:31
rosmaitai would prefer not to talk about RBAC right now unless absolutely necessary14:32
simondodsleylol14:32
rosmaitaok, we will get that stuff taken care of on Friday14:32
rosmaita#topic Pure storage reports group replication as reported but the respective methods are not implemented14:33
rosmaitawhoami-rajat: that's you14:33
whoami-rajathey14:33
simondodsleyyes - i'll take a look at that  and report back14:33
whoami-rajatthanks, that's all i wanted14:33
rosmaitaexcellent, that was easy14:34
whoami-rajatyep, simondodsley \o/14:34
simondodsleyi assume this is due to the new replication temeest tests14:34
rosmaita#topic HPE Alletra MP driver needs reviews14:34
whoami-rajatno, i found out while i was enabling group replication in RBD14:34
simondodsleyok14:34
Anoop_ShuklaI had one topic about the greenthread.sleep vs time.sleep - https://docs.openstack.org/nova/14.0.4/threading.html document suggests and recommends to use greenthreads..while we recently have changed to time.sleep.14:34
whoami-rajat#link https://review.opendev.org/q/topic:%22rbd-group-replication%2214:34
rosmaita#link https://review.opendev.org/c/openstack/cinder/+/96937314:35
rosmaitaok, we have 3 discussions happening at once14:35
Anoop_ShuklaSorry..my bad..we can take it at last14:36
rosmaitaanything more on the Pure group replication stuff?14:36
rosmaitaok, that has been discussed, thank you simon and Rajat14:37
whoami-rajatsorry and thanks, nothing from my side14:37
rosmaitaok, next item is the HPE alletramp driver14:37
rosmaitaHPE has been really good about revising and responding to requests14:38
JyotsnaThanks <rosmaita>14:38
rosmaitawhoami-rajat: i wonder if your agentic driver stuff is ready to let loose on that driver?14:38
rosmaitamy perception is that the driver is really close, and we should get it completed as soon as possible to make room for the new drivers this cycle14:39
JyotsnaWe need attention again from reviewers so that this time we dont loose chance to get the driver. 14:39
rosmaitaexactly14:39
JyotsnaWe have been chasing this from almost a year and i would appreciate all the reviwers patience and efforts they spent in reviweing!!14:40
rosmaitalooks like agalica and erlon have looked at it most recently14:40
rosmaitaand eric14:41
Jyotsnayes they have reviewed it and we have addressed all the comments from them . All CIs are green and Zuul verified 14:41
rosmaitaok, well, reviewers please try to take a look today and tomorrow14:42
JyotsnaWill it be possible for reviewers to provide +2 and move this towards merge. 14:42
rosmaitaotherwise, i think we will have to do it at the Festival on friday14:42
rosmaitawhich will pretty much mean that no other patches will get reviewed14:42
rosmaitabut, it that's what it takes ...14:42
JyotsnaThanks <rosmaita> for bringing everyone's attention on patch. 14:43
rosmaitanp, thank you for your patience and willingness to revise14:43
whoami-rajatI can't guarantee it but will report if the skill finds anything useful worth flagging14:44
rosmaitaok, let's move to open discussion .. we have 2 things already, the greenthread.sleep question, and i forget what the other one was14:44
rosmaita#topic open discussion14:44
eharneyAnoop_Shukla: there is almost no situation where we should be calling greenthread.sleep instead of time.sleep14:44
Anoop_Shuklahttps://docs.openstack.org/cinder/newton/devref/threading.html The documentations suggest otherwise though14:45
eharneyas we are moving toward removing eventlet and running in threaded mode, we will be removing eventlet-specific code14:45
Anoop_ShuklaOkay..so the follow up question was, is there any performance implication that the drivers need to test for?14:46
eharneylooks like we need to put it on the todo list to rewrite that doc14:46
Anoop_ShuklaSince time.sleep would pause the OS thread?14:46
whoami-rajatisn't that doc Pro-eventlet rather than in favor of removing it, also I'm not sure if it's for the latest release14:46
rosmaitai thought time got monkey-patched so that you didn't have to worry about using the greenthread.sleep directly14:46
whoami-rajat#link https://docs.openstack.org/nova/latest/reference/threading.html#native-threading14:47
eharneycorrect14:47
whoami-rajatthis is the latest one14:47
eharneyso there isn't really a reason to want to do it14:47
Anoop_ShuklaYes..that seems to be the problem. There are documentations that suggest otherwise.14:47
Anoop_ShuklaOkay good14:47
Anoop_ShuklaSo the latest documentation is updated then..14:47
rosmaitawe are at a transition point, because eventlet is not supported in modern pythons14:48
rosmaitaand python has evolved so that hopefully eventlet is no longer necessary14:48
Anoop_ShuklaIn current code, time.sleep(0) does the same thing as greenthread.sleep(0) if time module is patched through eventlet.monkey_patch(). To be explicit, we recommend contributors use greenthread.sleep() instead of time.sleep()14:48
Anoop_ShuklaThe latest documentation still talks about recommending greenthread.sleep14:49
whoami-rajatit's in the eventlet section and not in native threading14:49
whoami-rajatBefore the Flamingo release all OpenStack services used the green thread model of threading, implemented through using the Python eventlet and greenlet libraries.14:49
Anoop_Shuklaokay14:50
Anoop_ShuklaSo we expect no impact of the change on performance?14:50
rosmaitaAnoop_Shukla: thank you for bringing this up14:50
rosmaitawell, we don't actually know about performance14:51
eharneydriver maintainers will want to check for performance issues (and ensure that any dependencies are actually using thread-safe python libs etc)14:51
rosmaitai think short term, we should put a warning box on the pages that mention working with eventlet to say it's no longer relevant14:52
rosmaitaunless someone else want to do it, i will take an action to do that right after this meeting14:53
Anoop_ShuklaThanks @rosmaita as it may confuse the new driver enthusiasts..14:53
Anoop_ShuklaWe will try to performance test the change and evaluate if there are any bottleneck observed with the patched code14:54
rosmaita#action rosmaita eventlet-removal warnings in dev docs14:54
rosmaitaAnoop_Shukla: that will be nice14:54
rosmaitaok there was another issue that i told someone to wait for open discussion14:54
rosmaitathat time is now14:55
aloke_devHi, I'd like to check on PR https://review.opendev.org/c/openstack/cinder/+/986072 (InfoScale Cinder Driver). The 3rd party CI is now active and green, and the patch has a +1. I'm looking for reviewer feedback to help move this toward a +2. Since this is my first contribution, please let me know if I need to adjust anything to align with the workflow.14:55
rosmaitathank you!14:55
Anoop_ShuklaWith greenthread.sleep(): While one green thread sleeps, dozens of other green threads continue executing I/O tasks or logic. High network throughput is maintained.   With time.sleep(): The event loop freezes entirely. No other green thread can run during the countdown period, destroying application concurrency.14:55
fqiI have updated the cinder spec for retype volume with migration including snapshots. I would appreciate it if you could please review it. https://review.opendev.org/c/openstack/cinder-specs/+/98932214:55
rosmaitaaloke_dev: do you know about the new driver review checklist?14:57
eharneyfqi: i am still unclear about how many backends can synthesize snapshots out of temporary volumes, or how that works exactly14:57
aloke_devYes, I followed the guidelines @rosmaita. 14:57
rosmaitaaloke_dev: excellent!14:57
rosmaitawell, i think we need to get the HPE alletramp out of the way, and that will free up bandwidth for infoscale14:58
rosmaitaaloke_dev: in the meantime, take a look at the bugs listed in https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/44MLENT3AXUML7CYYLNWO6Z7LO6CQT6W/14:59
rosmaitajust to make sure that they don't apply to you14:59
rosmaitafqi: we are running out of time, but if you want to discuss some more, people will probably stick around for a little while15:00
rosmaitaok, thanks everyone and don't forget the Festival of Reviews on friday15:01
rosmaita#endmeeting15:01
opendevmeetMeeting ended Wed Jul  8 15:01:06 2026 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)15:01
opendevmeetMinutes:        https://meetings.opendev.org/meetings/cinder/2026/cinder.2026-07-08-14.00.html15:01
opendevmeetMinutes (text): https://meetings.opendev.org/meetings/cinder/2026/cinder.2026-07-08-14.00.txt15:01
opendevmeetLog:            https://meetings.opendev.org/meetings/cinder/2026/cinder.2026-07-08-14.00.log.html15:01
aloke_devSure, I will have a look! Thank you <rosmaita>!15:01
fqiEric, Sorry I have to be off for a while. I will talk to you later when you are free.15:06
opendevreviewBrian Rosmaita proposed openstack/cinder master: [docs] Eventlet is not a preferred threading model  https://review.opendev.org/c/openstack/cinder/+/99649515:38
rosmaitaeharney: please do a sanity check ^^15:39
opendevreviewSimon Dodsley proposed openstack/cinder master: [Pure Storage] Implement Tiramisu group replication actions  https://review.opendev.org/c/openstack/cinder/+/99649715:58
opendevreviewMerged openstack/devstack-plugin-ceph master: Allow customizing NFS-Ganesha port  https://review.opendev.org/c/openstack/devstack-plugin-ceph/+/98310316:15
opendevreviewjayaanand borra proposed openstack/cinder master: Implement per-pool image-volume cache promotion  https://review.opendev.org/c/openstack/cinder/+/98831516:58
opendevreviewSimon Dodsley proposed openstack/cinder master: doc: Add driver security disclosure matrix  https://review.opendev.org/c/openstack/cinder/+/99655421:52
opendevreviewSimon Dodsley proposed openstack/cinder master: doc: Add driver security disclosure matrix  https://review.opendev.org/c/openstack/cinder/+/99655422:03
opendevreviewSimon Dodsley proposed openstack/cinder master: doc: Add driver security disclosure matrix  https://review.opendev.org/c/openstack/cinder/+/99655422:34

Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!