16:01:19 <jgriffith> #startmeeting cinder
16:01:20 <openstack> Meeting started Wed Mar 27 16:01:19 2013 UTC.  The chair is jgriffith. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:01:21 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:01:23 <openstack> The meeting name has been set to 'cinder'
16:01:24 <jgriffith> hehe :)
16:01:30 <jgriffith> eharney: To the rescue!
16:01:45 <jgriffith> Ok... again I neglected to update the wiki page
16:01:52 <jgriffith> lemme make sure nobody else added anything :)
16:02:17 <jgriffith> Nope :)
16:02:23 <jgriffith> #topic RC3
16:02:36 <jgriffith> So you all likely saw we did a quick turn on an RC3
16:02:49 <jgriffith> hopefully that's it for Grizzly release candidates
16:03:02 <jgriffith> we found a few bugs this week that we added there, but nothing too serious
16:03:14 <jgriffith> mostly the quotas stuff (thanks guitarzan)
16:03:32 <jgriffith> I also threw in the fix for checking tenant in the context for quota updates
16:04:04 <jgriffith> I didn't test this as much as I would've liked but it seemed to merge cleanly with our other changes and a quick test everything looked good
16:04:25 <jgriffith> Anybody have anything regarding milestone-proposed/RC they want to mention?
16:05:47 <bswartz> jgriffith: the netapp bug, is not an easy fix
16:05:55 <bswartz> I will wait until havana to fix it
16:06:14 <jgriffith> bswartz: sounds good... go ahead and target it in launchpad if you would please (H1)
16:06:20 <DuncanT> I've just got more client bugs
16:06:21 <bswartz> oh ok
16:06:26 <jgriffith> DuncanT: :)
16:06:33 <jgriffith> So FYI the quota changes:
16:07:18 <jgriffith> https://review.openstack.org/#/c/25490/
16:07:28 <jgriffith> https://review.openstack.org/#/c/25326
16:07:39 <jgriffith> https://review.openstack.org/#/c/25251
16:07:43 <jgriffith> pheww
16:08:00 <j_king> jgriffith: ah, that is exactly the issue I've been working on. lol
16:08:05 <bswartz> jgriffith: I'm not sure I can set the target milestone -- it's not clear in the bug UI
16:08:10 <jgriffith> For my next trick, figure out how to cherry-pick all of those back into one commit for Folsom
16:08:24 <jgriffith> bswartz: No worries, I'll have a look at it
16:08:28 <jgriffith> bswartz: bug number?
16:09:16 <jgriffith> j_king: :)  Hopefully you haven't spent a ton of time on it
16:09:26 <jgriffith> j_king: sorry... about that
16:09:35 <jgriffith> initially suo said he wouldn't be able to do it
16:09:42 <jgriffith> then it came in last week and got burried
16:09:46 <bswartz> 1139129
16:10:14 <jgriffith> bswartz: done
16:10:26 <j_king> jgriffith: reviewing it now and noticing that we pretty much made the exact same changes. no worries.
16:10:35 <j_king> my patch has been sitting as a draft
16:10:38 <jgriffith> bswartz: I'm wondering if you have permissions to see that...  "Milestone" column?
16:10:46 <j_king> and I've been taking too long at it
16:10:52 <jgriffith> j_king: sorry bout that
16:10:57 <j_king> jgriffith: no worries. :)
16:11:15 <bswartz> yeah I can see it I just can't edit
16:11:28 <jgriffith> bswartz: ahh... ok
16:11:40 <jgriffith> Ok, so in summary... I think we have a pretty good RC
16:12:07 <jgriffith> The only core issue that is still an issue is that the secure delete on snapshots
16:12:08 * bswartz is bereft of power on launchpad
16:12:13 <jgriffith> ie we're not doing it
16:12:25 <jgriffith> but given the precise bug this may be for the best
16:12:51 <jgriffith> and I believe rushi may have had issues on Quantal as well
16:12:53 <jgriffith> :(
16:13:22 <jgriffith> So at this point if folks can test and document that would be fantastic
16:13:39 <jgriffith> I mostly want to look at installing via packages at this point
16:14:07 <jgriffith> make sure we've got things well documented in openstack-docs for things like how to configure multiple back-ends, and of course your respective drivers
16:14:12 <jgriffith> speaking of which...
16:14:30 <jgriffith> https://review.openstack.org/#/c/25490/
16:14:46 <jgriffith> I seperated the cinder admin docs and driver files
16:15:02 <jgriffith> please take a look and make sure I didn't break any of your stuff in the process :)
16:15:51 <jgriffith> One other thing I noticed last night...
16:16:10 <jgriffith> It seems the sync of oslo-requires nuked eventlet from test-requires
16:16:31 <jgriffith> Not an issue regarding release, but if you're trying to do testing and notice something funky there, that's why
16:16:39 <jgriffith> I'll push a patch to get that put back in later
16:16:53 <thingee> jgriffith: I can put in a update to openstack-infra to build the admin docs.
16:16:59 <jgriffith> it's still in pip-requires, but it's needed for tests as well
16:17:10 <annegentle> thingee: I have one in review I think, let me dig up the url
16:17:12 <xyang_> jgriffith: where is the new doc located?  so it is not in http://docs.openstack.org/trunk/openstack-compute/admin/content/ch_volumes.html any more?
16:17:13 <jgriffith> thingee: that would be great (since I don't know how to do that anyway) :)
16:17:24 <thingee> annegentle: perfect :)
16:17:33 <jgriffith> xyang_: that's what thingee is referring to is making that happen
16:17:35 <jgriffith> :)
16:17:55 <annegentle> thingee: https://review.openstack.org/#/c/25384/ yep it merged
16:18:12 <annegentle> thingee: next step, edit the index.html in openstack-manuals/www where you want it to be linked
16:18:20 <thingee> xyang_: it'll likely be moved to http://docs.openstack.org/trunk/openstack-block-storage/admin/content
16:18:29 <jgriffith> annegentle: and thingee are my heros!
16:18:34 <xyang_> thingee: ok. thanks
16:18:48 <annegentle> jgriffith: aw shucks
16:18:53 * jgriffith is very excited to finally have Cinder docs!
16:19:35 <jgriffith> Ok... so any questions/concerns on RC and docs?
16:19:51 <jgriffith> #topic encryption
16:20:02 <jgriffith> Not sure how many folks have been following discusion on the ML
16:20:17 <jgriffith> I have some concerns about how this is planned to be implemented in Nova
16:20:37 <jgriffith> particularly for those of us that have or will have back-ends that implement encryption themeselves
16:21:09 <jgriffith> I talked with Laura the other day and it looks like the design will allow us to specify encryption on volume create
16:21:30 <jgriffith> and we can put hooks in to modify which version/implementation is used so that should be ok
16:21:49 <DuncanT> I still don't like it being in nova full stop
16:21:52 <jgriffith> I still have questions about snapshots and clones, I'm labbing some stuff out with LVM to see how this may or may not work
16:21:56 <jgriffith> DuncanT: agreed!
16:22:12 <jgriffith> DuncanT: I've objected on the ML as well as personally to Laura
16:22:21 <jgriffith> DuncanT: The plan at this point....
16:22:35 <DuncanT> I'm just looking for the thread... I'm way behind on the list
16:22:35 <jgriffith> I'm going to accept the proposed encryption session to the cinder track
16:22:45 <jgriffith> there's also going to be one on the Nova track
16:23:06 <jgriffith> I'm going to work with russellb and hopefully have it so the nova session isn't on a cinder day
16:23:15 <jgriffith> so we can participate in both sessions and vice versa
16:23:33 <bswartz> jgriffith: +1
16:23:44 <jgriffith> DuncanT: the thread has digressed considerably into a discussion on key management
16:24:02 <jgriffith> DuncanT: but the proposed patch is up, everyone should take a look
16:24:02 <DuncanT> I'll catch up and wade in :-)
16:24:29 <DuncanT> I -1'd one patch ages ago, not been keeping up since, sorry
16:24:39 <jgriffith> DuncanT: no worries
16:24:41 <russellb> jgriffith: what day is cinder day
16:24:52 <russellb> jgriffith: and do you want the nova one after the cinder one?
16:24:58 <jgriffith> russellb: As it stands now I beleive it's still Thursday
16:25:05 <russellb> ah ok
16:25:13 <russellb> nova first it is
16:25:19 <jgriffith> russellb: I don't have a strong preference and I think Cinder being thurs limits the options a bit
16:25:22 <jgriffith> :)
16:25:24 <jgriffith> russellb: thanks :)
16:25:28 <russellb> np
16:25:51 <xyang_> jgriffith: do you have a link to the proposed patch?
16:26:02 <jgriffith> xyang_: looking....
16:26:41 <kmartin> jgriffith: annegentle Sorry I'm late was working away on the docs and lost track of time :) I'll have new patches up today for the feedback I got back from Anne.
16:28:33 <jgriffith> xyang_: hmm... I've lost the patch :(
16:28:43 <jgriffith> xyang_: I'll have to dig through the ML to find it again
16:28:56 <xyang_> jgriffith: I see this one https://review.openstack.org/#/c/21262/.  It is abandoned.
16:29:02 <winston-d> jgriffith: what do you guys think about doing volume rate-limiting on nova?
16:29:26 <jgriffith> winston-d: how do you mean?  IO?
16:29:34 <winston-d> jgriffith: yup, IO
16:29:43 <jgriffith> winston-d: I'm violently opposed
16:29:46 <jgriffith> :)
16:29:53 <winston-d> :)
16:30:17 <DuncanT> winston-d: We want to do rate limiting somewhere. Got any proposals?
16:30:56 <DuncanT> winston-d: We were certainly looking at doing it in the hypervisor, but the hypervisor support for it looks rather flakey
16:31:00 <jgriffith> DuncanT: winston-d so I'd be more concerned about the implementation here
16:31:07 <winston-d> DuncanT: ultimately i think it's best/easy to do that on Nova/hypervisor
16:31:37 <DuncanT> jgriffith: It would certainly be easy to come up with something ugly :-)
16:31:50 <jgriffith> DuncanT: haha :)
16:32:10 <jgriffith> So we can definitely investigate...
16:32:32 <jgriffith> but from my experience it's a pretty complex problem and I'd be curious about the demand
16:32:55 <jgriffith> It's also lower on my priority list compared to things like encryption, migration of volumes, multi-attach etc
16:33:06 <DuncanT> We have what we think is a requirement for it, but no serious attempts at implementation or testing
16:34:05 <jgriffith> alright, we should talk about it then
16:34:20 <jgriffith> winston-d: DuncanT maybe the two of you could outline exactly what you have in mind
16:34:24 <jgriffith> the use case etc
16:34:35 <winston-d> jgriffith: sure. will talk to DuncanT
16:34:35 <DuncanT> Sure.
16:34:59 <jgriffith> I do think there are some other things that are higher priority, but maybe that's not accurate
16:35:18 <jgriffith> and of course there's the obvious solution that already exists :)
16:35:40 <DuncanT> Oh, we want all of the above too, and the moon on a stick please, and two ponies
16:35:52 <jgriffith> three ponies!!
16:35:53 <DuncanT> (The obvious solution? Source rate limiting?)
16:35:54 <jgriffith> and a unicorn
16:36:00 <jgriffith> DuncanT: SolidFire :)
16:36:11 <jgriffith> sorry... couldn't resist
16:36:13 <jbr_1> volume_type
16:36:22 <winston-d> :)
16:36:33 <jgriffith> Ok... since we're talking sessions
16:36:38 <jgriffith> #topic summit sessions
16:36:55 <jgriffith> http://summit.openstack.org/
16:37:12 <thingee> DuncanT: I saw there was a proposal for testing. I was think making this a bit more low key unconference.
16:37:18 <jgriffith> We have 17 proposals and 11 slots
16:37:27 <jgriffith> thingee: +1
16:37:32 <jgriffith> DuncanT: you good with that?
16:38:06 <jgriffith> Also I'd like to look at consolidating a couple that may be related
16:38:22 <DuncanT> Fine by me
16:38:34 <DuncanT> I'm fine with having it in a bar somewhere if necessary...
16:38:41 <thingee> DuncanT: +1
16:38:42 <jgriffith> bar :)
16:38:50 <thingee> best unconference evar
16:38:59 <kmartin> with a Pliny in hand
16:39:05 <jgriffith> DuncanT: I'd also like to combine your give/take proposal with Avishays migration proposal
16:39:18 <jgriffith> I think they'll tie together if that's ok with you?
16:39:40 <DuncanT> Ok. They're not very related, but I've no problem with sharing a slot... it worked out mostly fine last year
16:40:04 <jgriffith> DuncanT: hmmm.... you don't think we could leverage the work for the two?
16:40:20 <winston-d> these two are almost identical http://summit.openstack.org/cfp/details/146 + http://summit.openstack.org/cfp/details/152
16:40:25 <DuncanT> give/take is purely ownership, it doesn't affect anything except that database
16:41:01 <DuncanT> jgriffith: It doesn't need a full session though
16:41:06 <jgriffith> DuncanT: agreed, but I think it *sort of* falls in the category of dynamic volumes so to speak
16:41:20 <winston-d> and Avishays agree to combine those two.
16:41:21 <jgriffith> DuncanT: Ok, I'll look at combining or seeing if something more logical falls out
16:41:25 <jgriffith> winston-d: looking
16:41:45 <jgriffith> winston-d: yes, for sure, those will fold in to one
16:42:23 <jgriffith> I think that will put us in pretty good shape
16:42:28 <DuncanT> The plugability one requires no code changes at all but might turn into a political bunfight
16:42:31 <jgriffith> the only other one is the plugins from Chuck
16:42:38 <jgriffith> DuncanT: :)
16:42:46 <winston-d> and the independent scheduler one http://summit.openstack.org/cfp/details/97 here isn't valid since scheduler is already independent.
16:42:50 <jgriffith> I haven't gotten a good feel for the level of interest there
16:43:04 <jgriffith> I think I'll drop it if we're out of sessions
16:43:12 <jgriffith> but I'd like to talk to folks about it in Portland
16:43:14 <DuncanT> Seems reasonable
16:43:25 <jgriffith> explain a bit more about what I have in mind and what Chuck is thinking there
16:43:59 <jgriffith> Ok... anything else on sessions?
16:44:06 <jgriffith> Anything folks thing we need to add?
16:44:48 <winston-d> maybe we can also talk about volume/task state machine in unconference?
16:44:57 <jgriffith> winston-d: OHHHH yes!
16:45:13 <j_king> winston-d: does this have to do with co-ordination?
16:45:15 <jgriffith> winston-d: I'd like to see if we can have a true session for that
16:45:29 * j_king interested in distributed co-ordination problems in general
16:45:32 <jgriffith> I'll see if I can futz the schedule around to make that happen
16:45:40 <DuncanT> +1 million
16:45:43 <winston-d> cool
16:45:53 <DuncanT> Clean forgot about that, despite having pages of notes on the subject
16:46:19 <kmartin> jgriffith: I'll get you some more details on the mutli-attach next week and we can decide if we need to have a unconference session or piggy back your read-only multi-attach
16:46:44 <jgriffith> kmartin: I was thinking we keep that as a formal session and collapse read-only in to it
16:47:03 <kmartin> jgriffith: ok
16:47:06 <jgriffith> kmartin: i think R/O would just be an option to a volume and should be pretty straight forward
16:47:16 <jgriffith> the bulk of the discussion should be multi-attach
16:47:49 <kmartin> I'll write something up next week and forward it on to you
16:48:34 <j_king> i brought up the possibility of a python-paxos implementation at the zookeeper talk at PyCon for this sort of thing. not sure if it's applicable to what you're talking about but if there's a ML thread about it, I'll hop in. (can't be at summit)
16:49:09 <jgriffith> j_king: I'd be interested
16:49:20 <jgriffith> j_king: say whaaaa?
16:49:23 <jgriffith> j_king: why not?
16:50:09 <j_king> jgriffith: 4mo old daughter to take care of and I already went to pycon. ;)
16:50:24 <j_king> plus not currently employed
16:50:38 <jgriffith> j_king: ahh.. those are good reasons
16:51:26 <jgriffith> alright, anybody have anything else?
16:51:38 <jgriffith> #topic cinderclient
16:51:59 <jgriffith> Just a heads up, prioritize testing and reviews on cinderclient if you could the next couple days
16:52:12 <jgriffith> I'd like to release to PyPi on Friday at the latest
16:52:21 <jgriffith> ok... if there's nothing else
16:52:35 <jgriffith> we'll finish at least a *little* bit early
16:52:51 <jgriffith> thanks everybody... as always grab me on IRC if you want to talk about something
16:52:59 <jgriffith> #end meeting
16:53:06 <jgriffith> #endmeeting