15:01:59 <enriquetaso> #startmeeting cinder_bs
15:01:59 <opendevmeet> Meeting started Wed Jan 25 15:01:59 2023 UTC and is due to finish in 60 minutes.  The chair is enriquetaso. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:01:59 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:01:59 <opendevmeet> The meeting name has been set to 'cinder_bs'
15:02:03 <enriquetaso> Report of bugs
15:02:04 <enriquetaso> #link https://lists.openstack.org/pipermail/openstack-discuss/2023-January/031900.html
15:02:11 <enriquetaso> #topic Security issue with VMDK format
15:02:14 <enriquetaso> <happystacker> I have a question around it but we talk in our bug review
15:02:21 <rosmaita> o/
15:02:24 <whoami-rajat> ganso, ack, i see it's in merge conflict
15:02:31 <whoami-rajat> but we can discuss later after the BS meeting
15:02:37 <whoami-rajat> hi
15:02:43 <enriquetaso> hi
15:02:50 <hemna> hi
15:02:56 <enriquetaso> i think happystacker is not here lol
15:02:59 <rosmaita> whoami-rajat: ganso: probably because it touches image_utils, and we just merged a big change in there
15:03:18 <ganso> whoami-rajat: oh I didn't spot the merge conflict in this new gerrit interface
15:03:22 <whoami-rajat> ah yes
15:03:27 <ganso> whoami-rajat: I will rebase it, thanks
15:03:32 <whoami-rajat> cool thanks
15:03:36 <rosmaita> ok, we can return to it if happystacker shows up later
15:03:45 <enriquetaso> ++
15:03:52 <enriquetaso> #topic Group actions enable, disable and failover replication can leave volume's replication status in transient states enabling, disabling and failing-over respectively.
15:03:55 <enriquetaso> #link https://bugs.launchpad.net/cinder/+bug/2003804
15:04:00 <enriquetaso> "Consider a group(grp_1) has 3 volumes(vol_1, vol_2, vol_3) and initially all the volume's replication_status is 'disabled'. When a user executes below use-cases concurrently, there is a probability that volume's replication_status gets stuck in 'enabling' status."
15:04:06 <enriquetaso> Fix proposed to master:
15:04:06 <enriquetaso> #link https://review.opendev.org/c/openstack/cinder/+/871649
15:04:29 <enriquetaso> Looks like a valid bug but the patch needs some update
15:05:03 <happystacker> sorry, I got disconnected, the question was more on why do we need to convert some format to .raw?
15:06:02 <enriquetaso> hey!
15:06:33 <happystacker> I have one bug affected by this conversion.
15:06:47 <rosmaita> happystacker: i think it's because volumes are usually in raw format, that's what nova expects (except for some weird cases)
15:07:29 <happystacker> Ok, but it may harms some operation like a server resize where the server expects a qcow2 but have a .raw instead
15:07:30 <happystacker> https://bugs.launchpad.net/cinder/+bug/2002535
15:07:41 <rosmaita> but we convert in the other direction too, when we upload a volume as an image
15:08:04 <Roamer`> happystacker, if I understand your question correctly, in most (I believe) shared storage (at least) backends it is better for the volume to be raw, so the storage backend can expose it directly as a disk device, so that read/write operations are, well, fast
15:08:06 <happystacker> ok which makes sense
15:08:17 <rosmaita> yeah, NFS is a different beast
15:08:31 <happystacker> yeah as you substract one layer of presentation
15:08:45 <rosmaita> you're really "faking" a raw volume as a filesystem entity
15:09:27 <happystacker> but in case of NFS, it's preventing the server resize as you can see in the bug. server is expected one format but have another one
15:09:44 <eharney> yeah, we need to understand what is happening in that bug
15:10:23 <happystacker> I did a very basic investigation, we probably need to go deeper
15:11:49 <happystacker> I'm here again...
15:11:49 <happystacker> I did a very basic investigation, we probably need to go deeper
15:11:59 <happystacker> having some network issues at my office
15:12:08 <happystacker> sorry for that
15:12:14 <enriquetaso> i havent reproduced the bug yet, sorry about that, but ill try to add some input after finishing the encryption part
15:12:20 <enriquetaso> no worries happystacker
15:12:24 <happystacker> cool thanks
15:13:33 <enriquetaso> not sure how can we track all the NFS problems besides launchpad
15:13:49 <eharney> launchpad bugs should work?
15:14:02 <enriquetaso> i think so, using the tag `nfs`
15:14:09 <rosmaita> enriquetaso: ++
15:14:12 <enriquetaso> okay, moving on
15:14:43 <enriquetaso> The last 2 bugs are driver related and both have patches proposed to master
15:14:47 <enriquetaso> #topic [HPE] cinder 3par FC driver not connecting to 3PAR storage
15:14:52 <enriquetaso> #link https://bugs.launchpad.net/cinder/+bug/2003245
15:14:57 <enriquetaso> I think this is a misunderstanding about how to deploy FC with HPE driver and not really a cinder related bug. However, since the bug has been really active and HPE are still having discussion there, I've left the question so they can reply with input.
15:17:16 <enriquetaso> #topic  [SVf] : Enable support for replication volume with mirror pool option
15:17:17 <roquej> I have another bug but I think this is driver related, even if I'm not sure that's the expected bahavior
15:17:26 <opendevreview> Simon Dodsley proposed openstack/cinder master: [Pure Storage] Add support for 3-site, trisync, replication.  https://review.opendev.org/c/openstack/cinder/+/862365
15:17:59 <enriquetaso> roquej, sure, would you mind sharing it after the svf one?
15:18:04 <enriquetaso> #link https://bugs.launchpad.net/cinder/+bug/2003245
15:18:08 <roquej> sure
15:18:09 <enriquetaso> Fix proposed to master:
15:18:10 <enriquetaso> #link https://review.opendev.org/c/openstack/cinder/+/870006
15:18:10 <enriquetaso> Unit tests are not passing, waiting for update.
15:18:11 <enriquetaso> done
15:18:19 <enriquetaso> #topic open discussion
15:18:25 <enriquetaso> roquej, feel free now
15:18:42 <roquej> https://bugs.launchpad.net/cinder/+bug/2003179
15:18:55 <opendevreview> Eric Harney proposed openstack/os-brick master: Bump bandit to release 1.7.0  https://review.opendev.org/c/openstack/os-brick/+/871260
15:19:01 <sfv880_> Hello eharney, rosmaita: I'm sorry to bother you! We are looking for a final +2 and +w on the Infinidat fix: https://review.opendev.org/c/openstack/cinder/+/864287 - could you please review it ? Thank you very much!
15:19:20 <roquej> when creating an image from volume, it turns out that a password is sent in plain text by privsep daemon in the logs
15:19:38 <enriquetaso> i forgot to add nova to 2003179
15:19:38 <eharney> yeah we need to fix this
15:19:50 <rosmaita> sfv880_: ack
15:20:02 <sfv880_> rosmaita: thank you very much!
15:20:07 <eharney> we have code that tries to automatically filter passwords like this, i haven't looked deep enough into why it's not covering this case
15:20:12 <enriquetaso> okay, maybe nova shoudnt be added
15:20:23 <eharney> maybe no identifying field to indicate that it's a password
15:21:01 <roquej> only when creating a volume from image
15:21:20 <enriquetaso> ooh okay
15:21:23 <eharney> it's because the password is fed to a CLI call via stdin?
15:21:38 <roquej> not sure, it's done via privsep
15:22:02 <roquej> I have posted the links to the code that I've identified as involved into this bug
15:22:10 <roquej> not sure where to start now
15:22:54 <eharney> oh it's reading the password from a file and returning it
15:23:00 <roquej> yes
15:23:29 <roquej> I tried with strutils to mask the password but didn't make it
15:23:41 <eharney> right
15:23:49 <roquej> maybe we need to change our way to get the password
15:25:33 <simondodsley> Cores: I just changed the commit message on https://review.opendev.org/c/openstack/cinder/+/862365. There was a typo. Our CI is currently down for maintenance and so our CI run failed with node failure.
15:25:35 <enriquetaso> okay, this looks like a bit change/new implementation.. In addition, this affects all nfs-related drivers, ill update the bug title, because i tought it was a dell-only bug
15:25:52 <eharney> there may be a short-term workaround by changing some options with how privsep is invoked for that particular call, but not sure
15:25:57 <simondodsley> Given there was no code change and the last Pure CI passed please take this into consideration when reviewing this patch
15:26:04 <roquej> this bug is not about nfs
15:26:17 <roquej> ok
15:26:48 <eharney> this is only going to affect the scaleio connector
15:27:15 <roquej> unless some other drivers used privsep to get password in the same manner, I'd say yes
15:29:00 <enriquetaso> sorry for the misunderstanding.. this is a bug about how privsep on how it handles passwords and it's going to affect the scaleio connector
15:29:37 <Sathya> Hi, https://bugs.launchpad.net/cinder/+bug/2001619 retype for volume migration with additional extra specs not entering driver specific code
15:29:47 <Sathya> geguileo had provided a general work flow for this issue
15:30:03 <Sathya> I have pushed a patch for the bug, along with ut's fix https://review.opendev.org/c/openstack/cinder/+/869999
15:30:09 <enriquetaso> okay, running out of time eharney roquej
15:30:32 <Sathya> can i get some review for this patch https://review.opendev.org/c/openstack/cinder/+/869999
15:30:32 <enriquetaso> do you mind if we continue the discussion on the launchpad bug? anyway I'll update the bug with this information
15:30:45 <enriquetaso> thanks!!
15:30:49 <enriquetaso> #endmeeting