16:00:26 <jungleboyj> #startmeeting Cinder
16:00:27 <openstack> Meeting started Wed Jun  5 16:00:26 2019 UTC and is due to finish in 60 minutes.  The chair is jungleboyj. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:00:28 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:00:30 <openstack> The meeting name has been set to 'cinder'
16:00:40 <whoami-rajat> Hi
16:00:45 <jungleboyj> courtesy ping:  jungleboyj whoami-rajat rajinir lseki carloss pots woojay erlon geguileo eharney rosmaita enriquetaso e0ne smcginnis davidsha walshh_ xyang hemna _hemna eharney
16:00:50 <walshh_> hi
16:00:52 <enriquetaso> o/
16:00:53 <davidsha> hey
16:00:54 <e0ne> hi
16:00:55 <geguileo> hi! o/
16:00:58 <m5z_|mszwed|> hi :)
16:00:59 <xyang> hi
16:01:05 <smcginnis> o/
16:01:07 <jungleboyj> Agenda:  https://etherpad.openstack.org/p/cinder-train-meetings
16:01:09 <rosmaita> o/
16:01:47 <jungleboyj> @!
16:01:47 <_pewp_> jungleboyj (;-_-)ノ
16:01:54 <hemna> @!
16:01:54 <_pewp_> hemna (✧∇✧)╯
16:02:03 <woojay> hi
16:02:13 <jungleboyj> @!
16:02:14 <_pewp_> jungleboyj (*ﾟ͠ ∀ ͠)ﾉ
16:02:33 <carloss> hi
16:02:38 * jungleboyj misses _pewp_ in the cinder channel
16:02:48 <lseki> o/
16:03:42 <jungleboyj> Ok.  Lets get started.
16:03:48 <jungleboyj> #topic announcements
16:04:06 <jungleboyj> Just a friendly reminder that we have the Train Mid-Cycle planning going on.
16:04:17 <jungleboyj> #link https://etherpad.openstack.org/p/cinder-train-mid-cycle-planning
16:04:55 <rajinir> o/
16:05:00 <jungleboyj> If you can make it please add your name.
16:05:09 <jungleboyj> Start planning ahead so we can have a good turnout.
16:05:14 <_pewp_> hemna (/ .□.) ︵╰(゜Д゜)╯︵ /(.□. )
16:05:15 <_pewp_> hemna (/ .□.) ︵╰(゜Д゜)╯︵ /(.□. )
16:05:15 <_pewp_> hemna (/ .□.) ︵╰(゜Д゜)╯︵ /(.□. )
16:05:16 <_pewp_> hemna (/ .□.) ︵╰(゜Д゜)╯︵ /(.□. )
16:05:17 <_pewp_> hemna (/ .□.) ︵╰(゜Д゜)╯︵ /(.□. )
16:05:18 <_pewp_> ...
16:05:19 <_pewp_> ...
16:05:19 <_pewp_> ...
16:05:20 <hemna> shit
16:05:25 <hemna> wtf was that
16:05:32 * jungleboyj is laughing
16:05:47 <jungleboyj> _pewp_:  Is mad at you
16:05:50 <hemna> evidently
16:05:55 <geguileo> lol
16:06:12 <jungleboyj> Anyone have other announcements?
16:06:32 <e0ne> jungleboyj: will you cut train-1 for cinder tomorrow?
16:06:56 <e0ne> AFAIK, it's not required anymore
16:06:58 <jungleboyj> e0ne:  Good question.  I thought that we didn't have to do that anymore.
16:07:02 <jungleboyj> smcginnis: ^^
16:07:11 <rosmaita> pretty sure it is up to us
16:07:17 <e0ne> jungleboyj: it's up to the team
16:07:31 <smcginnis> The cycle-with-intermediary deliverables will be proposed, but no need to do beta releases of the services anymore.
16:08:00 <smcginnis> Only reason to do a x.x.x.0b1 release is if you actually know someone downstream wants to pick it up and do something with it.
16:08:01 <jungleboyj> smcginnis:  Ok, that is what I thought.
16:08:20 <jungleboyj> eharney: geguileo Anything you guys need a beta for?
16:09:01 <geguileo> jungleboyj: I don't think so...
16:09:03 * jungleboyj hears crickets
16:09:05 <geguileo> eharney: ^
16:09:16 <jungleboyj> Ok.  Good.  So, I would say we just keep rolling forward.
16:09:29 <smcginnis> Based on what we saw, those usually were not being used.
16:09:40 <jungleboyj> We also don't have any deadlines for T-1 .
16:10:31 <jungleboyj> Though, now we need to make sure we are helping get people to T-2.
16:10:46 <jungleboyj> Doing driver reviews, reviewing spes.
16:10:49 <jungleboyj> *specs
16:10:50 <smcginnis> Gotta get those drivers reviewed.
16:11:39 <jungleboyj> smcginnis:  ++
16:12:58 <jungleboyj> So, a todo for the whole team is to please work on reviewing drivers and specs.
16:13:00 <jungleboyj> :-)
16:13:04 * jungleboyj is included
16:13:07 <eharney> i don't think we have any need for it
16:13:47 <jungleboyj> Ok.  I think we can move on.
16:14:08 <jungleboyj> #topic Driver Unsupported Patches
16:14:48 <jungleboyj> I have a list in the agenda of the drivers I have unsupported.
16:15:03 <hemna> is there a review topic for those?
16:15:44 <jungleboyj> Appreciate help getting those merged.
16:15:57 <jungleboyj> They all start with 'unsupport'
16:16:05 <jungleboyj> Guess I should have made that more general.
16:16:49 <jungleboyj> :-(
16:17:02 <jungleboyj> I could set them to a review topic.
16:17:02 <hemna> maybe change the topic on all of them, to the same?
16:17:22 <smcginnis> jungleboyj: I think I had always used "ci-unsupported" to make it easy to find all of them.
16:17:32 <jungleboyj> Ok.  I will do that.
16:19:08 <rajinir> Is there a link that can list them?
16:19:14 <jungleboyj> I am also going to try to work on getting driver removal patches up yet this week so that is all taken care of.
16:19:28 <jungleboyj> rajinir: They are in the agenda right now:  https://etherpad.openstack.org/p/cinder-train-meetings
16:19:35 <smcginnis> rajinir: If they all have the same topic, then it's easy to list the gerrit query to list them all
16:19:57 <jungleboyj> Or you can just click on the topic and get a list.
16:19:58 <smcginnis> Similar to what I set for the new drivers: https://review.opendev.org/#/q/topic:train-drivers
16:20:05 <rajinir> ++
16:21:12 <smcginnis> jungleboyj: Had the ci_unsupported wrong (-/_): https://review.opendev.org/#/q/topic:ci_unsupported
16:21:57 <jungleboyj> Ah, you do that for removal too.  Ok.  Good.
16:22:06 * jungleboyj thinks that is something to add to our documentation.  :-)
16:22:36 <rajinir> smcginnis: thanks
16:23:35 <jungleboyj> So, I will keep all the patches under that topic.
16:24:19 <jungleboyj> So, I think that is all I have there.
16:24:36 <jungleboyj> #topic Update on Ceph iSCSI driver
16:24:59 <hemna> ok I guess I'm up
16:25:12 <hemna> so I've been doing some research on an iscsi ceph driver
16:25:13 <jungleboyj> hemna:  If you don't mind.
16:25:31 <hemna> I have an etherpad here: https://etherpad.openstack.org/p/cinder-ceph-iscsi-research
16:25:50 <hemna> and a WIP of the driver here:  https://review.opendev.org/#/c/662829/
16:26:28 <jungleboyj> Awesome.
16:26:29 <hemna> the driver assumes that it has access to the rbd-target-api from the http://github.com/ceph/ceph-iscsi project
16:27:02 <hemna> the rbd-target-api python REST flask server is meant to coordinate the iscsi gw configs between ceph iscsi gateways
16:27:11 <hemna> so it's not really meant to do what I'm using it for
16:27:16 <hemna> but it happens to work at this point
16:27:31 <hemna> I'm getting close to having the initialize_connection working
16:27:38 <hemna> I was able to get a single attach and detach work
16:27:40 <hemna> once
16:27:56 <hemna> but then discovered issues, which is leading to more work for both initialize_connection and terminate_connection
16:28:18 <hemna> there is zero documentation on how to properly use the rbd-target-api to do an export
16:28:42 <hemna> but I'm kinda figuring it out as a go from a standalone script using my ceph_iscsi_client.py REST client
16:28:58 <smcginnis> Any conversations in the ceph community to just get some assurance that this is an acceptable use of it?
16:29:10 <hemna> nah not yet
16:29:22 <hemna> my hope is to verify that this works
16:29:39 <hemna> then the thought was to create a separate client pypi project
16:29:49 <hemna> that is the REST client that can talk to the rbd-target-api
16:30:03 <hemna> which is my ceph_iscsi_client.py
16:30:35 * jungleboyj is thinking it is good that you are the one working on this.
16:30:39 <hemna> the rbd-iscsi.py driver is a derivation of the existing rbd.py cinder driver
16:30:51 <hemna> so I got most of the functionality for free so far
16:30:55 <hemna> other than attach/detach
16:31:03 <jungleboyj> hemna:  Awesome.  I was hoping that that would be the case.
16:31:11 <hemna> but I've done zero testing with it other than create a few volumes and work on attach/detach
16:31:35 <hemna> this is a research project still at this point
16:31:44 <jungleboyj> Understood.
16:31:52 <hemna> with the hopes that it can get stabilized and then work on cleaning it up and making it a viable driver
16:32:11 <geguileo> great progress!!
16:32:20 <jungleboyj> ++
16:32:23 <hemna> there is a lot of setup on the ceph cluster side to make sure the iscsi gateway works properly
16:32:37 <hemna> so I'm not sure how much effort it'd take to do the CI side of things
16:32:58 <hemna> if someone is familiar with the existing ceph CI, I could help them on the iscsi gw setup
16:32:59 <hemna> for CI
16:33:01 <jungleboyj> We have automation in Lenovo that I think does some of that.
16:33:11 <jungleboyj> I wonder if we can help with that.
16:33:24 <hemna> I had to manually update my 7 ubuntu VMs to linux kernel 4.18
16:33:37 <smcginnis> That's going to be a big issue.
16:33:39 <jungleboyj> :-(
16:33:45 <hemna> 4.15 wasn't happy
16:34:05 <hemna> luckily it's an apt-get install
16:34:11 <hemna> instead of a compile
16:34:50 <hemna> so I'm hoping in the next day or so, to stabilize initialize_connection
16:34:54 <hemna> it's still giving me fits
16:35:19 <hemna> the rbd-target-api is a bit compilicated in what it needs done at certain times.
16:35:48 <hemna> anyone wants to help with the CI side of things I'd appreciate it
16:36:30 <hemna> but at least the driver is possible at this point still.  so I'm hopeful
16:36:31 <jungleboyj> hemna:  Ok.  Let me see if there is anything we have from the iSCSI work we did previously in Lenovo.
16:36:44 <jungleboyj> Will get back to you.
16:36:50 <jungleboyj> hemna:  Hopeful is good!
16:37:20 <hemna> I'll keep updating the driver review with code as I go along
16:37:25 <jungleboyj> hemna:  Anything else?
16:37:32 <hemna> that's it.
16:37:40 <jungleboyj> Cool.  Thank you for the update.
16:37:47 <eharney> i assume the CI work would involve adding a new mode to devstack-plugin-ceph?
16:37:49 <geguileo> hemna: thanks for the update and the work!
16:38:00 <jungleboyj> geguileo: ++
16:38:04 <hemna> eharney: yah I'm assuming so
16:38:13 <hemna> installing all the required ceph-iscsi packages
16:38:27 <hemna> and setting up the rbd-target-gw
16:38:34 <hemna> and installing the rbd-target-api
16:39:08 <hemna> I documented as much as I could in the etherpad
16:40:10 <hemna> geguileo: np.  I'd really love to get this working and in cinder.
16:40:26 <jungleboyj> hemna: ++
16:40:38 <jungleboyj> Ok.  That was all we had for topics this week.
16:40:43 <jungleboyj> #topic Open Discussion
16:40:56 <whoami-rajat> I've something that was brought up yesterday
16:41:13 <eharney> i'd love to get lvm multiattach fixed in stable: https://review.opendev.org/#/q/Icabc4a67d3f5462fe24e4027e84e56a001e1b2b8
16:42:46 <hemna> eharney: ++
16:42:50 <whoami-rajat> tempest is capping its api-extensions in devstack [1] that is blocked by cinder as we "mistakely" removed an extension in a MV bump which wasn't backward compatible. i've a patch up for it, will appreciate some reviews on it. https://review.opendev.org/#/c/663049/
16:42:50 <whoami-rajat> [1] https://review.opendev.org/#/q/topic:cap-extensions+(status:open+OR+status:merged)
16:43:09 <jungleboyj> eharney:  Ok, will watch those and get them through.
16:43:43 <smcginnis> geguileo: Do we still need to sort out this whole mess? https://review.opendev.org/574681
16:44:05 <eharney> i looked at the os-sch-hnt change and was fairly unclear on the details involved there
16:44:21 <geguileo> smcginnis: not sure...
16:44:44 <smcginnis> eharney: Which part?
16:45:21 <eharney> smcginnis: i think e0ne was going to follow up on syncing scheduler stats from the ptg, which i think is the same area that spec is about
16:45:41 <geguileo> smcginnis: I mean, the issue is still there, that's for sure
16:45:41 <smcginnis> Oh right, forgot about that from the PTG.
16:45:55 <jungleboyj> eharney: ++
16:46:00 <smcginnis> Wasn't sure if that spec should be abandoned or not. Sounds like it should be.
16:46:20 <whoami-rajat> eharney:  i replied to your comment there, hope it helps.
16:46:49 <geguileo> smcginnis: probably a good idea to abandon it for now...
16:46:59 <e0ne> I need to  read Tommy's spec
16:47:54 <smcginnis> e0ne: I think he had three different ones out there at one point trying to get opinions on different approaches.
16:48:44 <jungleboyj> :-)  That is one way to do it.
16:50:09 <jungleboyj> Other topics for today?
16:51:44 <jungleboyj> Doesn't sound like it ...
16:52:14 <jungleboyj> Ok.  So lets wrap this up.
16:52:39 <jungleboyj> Thanks everyone for joining and for continuing to work on Cinder.
16:52:52 <smcginnis> Thanks!
16:52:54 <jungleboyj> #endmeeting