16:01:37 <jgriffith> #startmeeting cinder
16:01:38 <openstack> Meeting started Wed Oct 31 16:01:37 2012 UTC.  The chair is jgriffith. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:01:39 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:01:40 <openstack> The meeting name has been set to 'cinder'
16:01:52 <thingee> o/
16:02:05 <jgriffith> e'lo, e'lo
16:02:25 <vincent_hou> hello
16:02:27 <DuncanT> hey
16:02:27 <winston-d> hi~
16:02:35 <jgriffith> There's everybody :)
16:02:39 <jgriffith> cool..
16:02:45 <jgriffith> Agenda: http://wiki.openstack.org/NovaVolumeMeetings
16:03:05 <jgriffith> #topic removal of nova-volume code
16:03:07 <creiht> hehe
16:03:11 <jgriffith> :)
16:03:13 <dricco> hello
16:03:24 <creiht> I'll be here the first 30 mins
16:03:32 <rnirmal> made it
16:03:35 <jgriffith> So...  I think there was some confusion (on my part included) as to how the deprecation of nova-vol was going to happen
16:03:40 <jgriffith> rnirmal: excellent
16:04:00 <jgriffith> For those that haven't seen patches have been going into nova master to remove nova-vol
16:04:18 <jgriffith> A few folks were surprised by this, so thought I should make sure we brought it up today
16:04:31 <jgriffith> So it's deprecated in Folsom, out in Grizzly
16:04:42 <hemna> yah that makes sense
16:04:48 <hemna> it's confusing that both are in now
16:04:49 <jgriffith> I know this might cause some trouble for some folks (rnirmal)
16:04:54 <hemna> for those of us that want to run off of trunk
16:05:08 <jgriffith> any concerns/issues from other folks that might be using nova-vol/master in production?
16:05:22 <rnirmal> well didn't seem to be really deprecated in folsom right since all patches were backported
16:05:48 <jgriffith> rnirmal: Yep, agreed
16:05:50 <DuncanT> We were expecting nova-volumes of be bugfix only for G, not pulled
16:06:17 <jgriffith> So DuncanT were you all planning to *use* nova-vol in Grizzly?
16:06:24 <DuncanT> This change is definitely causing us to have to change plans for realignment with trunk
16:06:36 <DuncanT> jgriffith: Possibly for a short period, yes
16:06:37 <creiht> jgriffith: I don't think it was effectivly communicated to the community
16:06:58 <hemna> I thought n-vol was out in Grizzly
16:06:59 <rnirmal> the process as such.
16:07:01 <jgriffith> creiht: You are correct, it was not
16:07:06 <creiht> though I feared that might happen, and was one of the reasons we switched to cinder asap
16:07:16 <jgriffith> hemna: Yes, some thought it was, some thougt it wasn't
16:07:20 <hemna> heh ok
16:07:20 <winston-d> i thought n-vol will be there for 2,3 more release
16:07:22 <jgriffith> creiht: :)
16:07:31 <DuncanT> jgriffith: It isn't that we can't do things differently, just that I specifically asked, got an answer and the answer was wrong
16:07:33 <jgriffith> winston-d: No, but 1 maybe
16:07:51 <DuncanT> 1 more release, with no new features, was what we expected
16:07:58 <jgriffith> DuncanT: Yeah and I think that was more my fault than anybody elses
16:08:09 <jgriffith> DuncanT: TBH, that's what I expected as well
16:08:36 <jgriffith> So, here's my point
16:09:00 <jgriffith> If this is going to be a huge deal for folks we can always go to the Nova team and state a case for leaving it in as deprecated
16:09:17 <jgriffith> TBH I'm happy to see it gone, but I understand the *real* world works differently
16:09:47 <winston-d> jgriffith, TBH i'm happy too, but i'm not a operation guy.
16:10:00 <DuncanT> So my biggest concern is how to avoid this happening again with other nova <-> cinder integration issues in future. Seeing it back in and depricated according to communicated schedule would be nice, but is a secondary concern
16:10:05 <jgriffith> winston-d: yes, you and I are lucky
16:10:30 <DuncanT> Seeing it stay in until after G-1 would be a good compremise, maybe?
16:10:37 <jgriffith> DuncanT: Well, as far as Cinder... I'll learn from my mistakes :)
16:10:47 <jgriffith> DuncanT: I can't promise what other folks/projects will do
16:11:02 <rnirmal> yeah agree with DuncanT on atleast till G-1
16:11:15 <jgriffith> DuncanT: Although TBH as I mentioned I was just as surprised to see it removed
16:11:28 <jgriffith> DuncanT: rnirmal Ok, I'll talk to vish about it
16:12:04 <winston-d> so if n-vol stays in G, will new features be added to it? no?
16:12:13 <jgriffith> winston-d: no
16:12:20 <DuncanT> 'No' I believe is the right answer there
16:12:29 <jgriffith> winston-d: it would be deprecated, just there for compat issues
16:12:38 <creiht> yeah I don't think new features should be added
16:12:39 <winston-d> good then.
16:12:44 <jgriffith> Keep in mind this introduces a significant issue for cinder migrations though
16:12:59 <creiht> but "critical" bug fixes and security fixes are likely candidates to backport
16:13:14 <creiht> at least in my opinion
16:13:22 <winston-d> creiht, agree
16:13:25 <jgriffith> So one of the things that worked in our favor for migrating folsom-nova-vol -->cinder was everything was the *same*
16:13:36 <jgriffith> That was part of the logic behind the sync up between the two
16:14:03 <jgriffith> If we did the deprecated/leave it in Grizzly and move forward with cinder
16:14:17 <DuncanT> That is why I think post-G1 is a good compremise - they aren't carrying cruft for too long and certainly no where near a release, and those who were organised have already migrated
16:14:22 <jgriffith> migrating from folsom/nova-vol to grizzly/cinder could be problematic
16:14:36 <creiht> it already is problematic :)
16:14:58 <jgriffith> creiht: yes, it is... but I suspect it could get significantly worse
16:15:09 <creiht> oh certainly
16:15:15 <rnirmal> well if it's deprecated..  I don't think we need to support a grizzly migration path
16:15:24 <jgriffith> rnirmal: I wish that were true
16:15:28 <jgriffith> rnirmal: but I don't think it is
16:15:36 <rnirmal> hmm
16:15:52 <jgriffith> perhaps I'm wrong on that
16:16:05 <jgriffith> But regardless it doesn't seem like it would be the *right* thing to do
16:16:17 <jgriffith> and I'd REALLY like to put the issues associated with migrations behind us
16:16:28 <creiht> yeah you still have to have a migration path
16:16:29 <jgriffith> the reality is that if/when bugs came in for it we'd have to deal with them
16:16:43 <creiht> jgriffith: that's why I said only "critical" bugs
16:16:49 <jgriffith> So... it sounds like folks would like to see nova-vol in until G1 at least?
16:16:51 <creiht> like attach doesn't work any more
16:17:03 <jgriffith> creiht: I have a patch in to fix that :)
16:17:12 * jgriffith hints to folks to check reviews :)
16:17:19 <creiht> oh I was giving that as just an example :)
16:17:28 <creiht> didn't realize it was actually broken
16:17:32 <jgriffith> creiht: I know, but it was a perfect segway :)
16:17:39 <jgriffith> creiht: OHHH
16:17:45 <jgriffith> creiht: never mind then, nothing to see here
16:17:55 <creiht> jgriffith: perhaps if you could find someone to volunteer to be the steward of nova-volume?
16:17:56 <jgriffith> creiht: It was actually non-cinder related
16:18:10 <jgriffith> creiht: interesting idea
16:18:21 <jgriffith> creiht: So there would be two sides to it
16:18:27 <jgriffith> 1. finding a volunteer
16:18:36 <jgriffith> 2. getting together with the nova team to agree
16:18:40 <creiht> jgriffith: because I do agree that it would be a lot on your plate to try to manage both cinder and nova-volume
16:18:54 <hemna> Is the Nova team wanting to nuke it now?
16:18:59 <jgriffith> It's not even that so much as it just *sucks* :)
16:19:09 <jgriffith> hemna: They've already started
16:19:23 <creiht> jgriffith: well my thought is that if someone really needs it still, they will want to make sure it still works :)
16:19:37 <jgriffith> creiht: agreed
16:19:38 <creiht> if nobody stands up, then maybe that means that they can remove it
16:19:50 <rnirmal> creiht: true.. whether that needs to happen in the open or just have them manage it internally
16:19:59 <jgriffith> Wondering if we should take a vote?
16:20:13 <winston-d> ok
16:20:14 <DuncanT> I need to have a wider discussion here to find out how concerned we are... can I get 24 hours or so to do that please?
16:20:22 <jgriffith> DuncanT: absolutely
16:20:34 <rnirmal> same here
16:20:39 <jgriffith> So we can sit on this for the moment and wait for some feedback?
16:20:43 <jgriffith> That's cool by me
16:20:50 <hemna> +1
16:20:56 <jgriffith> Do what you need to and hit me up with your cases
16:21:07 <DuncanT> Cool, cheers
16:21:10 <jgriffith> Ok... shall we move on?
16:21:20 <winston-d> i can collect some feedback from local cloud vendors
16:21:38 <jgriffith> I'm not making any promises by the way
16:21:53 <jgriffith> Just saying if it's a major issue we can look at fixing it
16:21:54 <rnirmal> jgriffith: :)
16:22:52 <jgriffith> Ok.. next topic
16:23:07 <jgriffith> #topic blueprint updates
16:23:23 <creiht> jgriffith: yeah I still need to get those in... :)
16:23:36 <creiht> for the api stuff
16:23:52 <jgriffith> creiht: :)
16:24:19 <jgriffith> I was hoping to have the G1 stuff at minimum all set up by end of week or Monday at latest
16:24:24 <DuncanT> No progress from us either
16:24:35 <jgriffith> DuncanT: creiht Does end of this week seem doable?
16:25:03 <creiht> jgriffith: yeah I think I can try to focus on that Thurs/Fri
16:25:14 <jgriffith> I don't think it's the end of the world, but it would just be nice to have a reasonable road-map going sooner rather than later
16:25:27 <creiht> yeah I agre
16:25:29 <creiht> agree
16:25:32 <kmartin> jgriffith, we have entered the FC Channel support blueprint and meeting again tomorrow with the list of interested parties(Broacade, IBM, ...)
16:25:36 <hemna> stupid question, is G1 the first sprint of Grizzly dev (month from now) ?
16:25:38 <DuncanT> I'll see what we can do. Seems like a reasonable aim
16:25:41 <creiht> gotta run, sorry... will check the backlog later
16:25:46 <jgriffith> hemna: yes
16:25:49 <hemna> thnx
16:25:51 <jgriffith> creiht: No worries... catch ya later
16:26:21 <jgriffith> grizzly release schedule: http://wiki.openstack.org/GrizzlyReleaseSchedule
16:26:32 <hemna> thank you
16:27:07 <jgriffith> Ok, so just wanted to throw that out as a reminder.  If you have blueprints you know you're going to want try to get them in and we should try and get them targetted by Monday
16:27:21 <rnirmal> jgriffith: how are we prioritizing items for G1 ?
16:27:27 <DuncanT> End of the week for blueprints looks far more reasonable with that on my screen :-)
16:27:32 <rnirmal> just based on if someone is going to get it done by g1
16:27:41 <jgriffith> DuncanT: :)
16:28:04 <jgriffith> rnirmal: So far it's been completely subjective... I look at the BP and mark it
16:28:08 <jgriffith> :)
16:28:13 <rnirmal> ah ok
16:28:20 <jgriffith> rnirmal: Right now that's fine
16:28:33 <jgriffith> rnirmal: If folks start getting stuff posted that will change
16:29:07 <jgriffith> hemna: I have a meeting this afternoon to talk about your FC blueprint
16:29:13 <hemna> cool
16:29:21 <jgriffith> hemna: That's going to need some work/detail added
16:29:23 <rnirmal> cool... would be better to get a clearer picture of what's being worked on looking at the blueprints
16:29:29 <hemna> the 3Par driver blueprint should be moved to G2
16:29:31 <jgriffith> rnirmal: agreed
16:29:46 <jgriffith> hemna: ok
16:30:04 <hemna> It's basically done, but there are external deps I'm still working on ironing out.
16:30:13 <jgriffith> anybody else have bp's they're active on that they'd like to see targetted to a specific release?
16:31:17 <DuncanT> Will let you know by the end of the week
16:31:20 <jgriffith> hemna: I've updated the BP and assigned you.  Make sure you keep things updated
16:31:22 <kmartin> jgriffith: we can talk about schedule for the FC this afternoon thinking G2 or G3
16:31:27 <hemna> thanks, will do
16:31:28 <jgriffith> DuncanT: thanks
16:31:33 <hemna> I'll add more details to the BP
16:31:41 <jgriffith> kmartin: Sure
16:31:57 <jgriffith> speaking of BP's... no bswartz?
16:32:03 <jgriffith> anybody else from NetApp around?
16:32:05 <winston-d> i'd like to see if vol-type-scheduler can make it in G1. and volume RPC API in G1 too.
16:32:18 <jgriffith> winston-d: Yes!  I agree on both
16:32:32 <jgriffith> winston-d: Just need to get the reviews in
16:32:44 <winston-d> jgriffith, :)
16:32:49 <jgriffith> Ok, so we don't need to talk NFS today?
16:33:07 <jgriffith> Or do folks have some input on it?
16:33:19 <jgriffith> I still haven't seen anything on the ML either
16:33:25 <hemna> NFS part of cinder???
16:33:38 <jgriffith> hemna: Yes, that's the proposal
16:34:08 <jgriffith> alright, I'm going to move off of that subject :)
16:34:28 <jgriffith> I didn't get the volume-type wiki done but should have it in the next day or two
16:34:33 <arosen> .hiub #openstack-ci
16:34:40 <jgriffith> I'll let folks know when it's up
16:34:57 <jgriffith> One other thing I wanted to ask....
16:34:58 <hemna> ok thanks, I'm interested in that
16:35:24 <jgriffith> jdurgin: or thingee or anybody that's worked on the write_image to volume code
16:35:45 <jgriffith> I've worked with a couple of folks trying to get this working and there was a lot of confusion
16:36:10 <jgriffith> I'm wondering if it would be possible for somebody that has gotten it to work to write up a short wiki on the steps on how to do it
16:36:13 <jgriffith> ?
16:36:53 * jgriffith hears crickets chirping...
16:36:57 <DuncanT> I think we've probably somebody who can make some input on that, sure
16:37:23 <jgriffith> DuncanT: that would be great, anything would be useful
16:37:24 <DuncanT> I think we've had to add patches to cope with QCOW images too, which we should push out some time soon
16:37:29 <thingee> sorry write_image? clone image or copy image to volume?
16:37:40 <jgriffith> DuncanT: ahhh.. that may be why it wouldn't work for me
16:37:54 <jgriffith> thingee: my bad.. .copy image to volume
16:37:59 <thingee> gotcha
16:38:01 <jgriffith> thingee: and how to boot it
16:38:02 <DuncanT> jgriffith: Yeah, with the default code you need a 'raw' image
16:38:16 <jgriffith> DuncanT: Well that's why mine kept bombing out most likely
16:38:21 <thingee> :)
16:38:26 <DuncanT> Got some instructions for converting a standard qcow image to raw somewhere, will send them on
16:38:33 <jgriffith> Ok, if we could get a wiki on this it would be immensely helpful for folks
16:38:41 <thingee> jgriffith: ok
16:39:02 <thingee> marking it
16:39:43 <jgriffith> I propose adding a section here: http://wiki.openstack.org/Cinder
16:39:59 <jgriffith> Mabye a Features/How-To section
16:40:13 <jgriffith> Then we can port them to docs.openstack.org at some point
16:40:19 <jgriffith> s/we/I/
16:40:21 <jgriffith> :)
16:40:26 <thingee> sounds good
16:40:32 <jgriffith> awesome!
16:40:53 <winston-d> good point, definitely helpful for docs.
16:41:05 <jgriffith> yeah, we really need to fix that problem at some poing :)
16:41:22 <jgriffith> point!
16:41:32 <jgriffith> #topic bug management
16:41:56 <jgriffith> Just a reminder, especially for folks that want to get involved but aren't sure how/where
16:42:05 <jgriffith> check out the bugs list and grab one :)
16:42:29 <jgriffith> Most of the valid ones are assigned/in-progress so that's good
16:42:48 <jgriffith> Let's just make sure we don't get behind the curve
16:43:05 <jgriffith> rnirmal: I'm waiting on a couple for the driver layout refactor
16:43:26 <jgriffith> rnirmal: Not sure if you saw my comments on that last patch or not
16:43:37 <rnirmal> jgriffith: yeah did see that
16:43:57 <jgriffith> everybody else: https://review.openstack.org/#/c/15038/
16:44:05 <jgriffith> rnirmal: Does my proposal suck?
16:44:06 <rnirmal> I'll update that today... so good with cinder/volume/driver/*
16:44:14 <jgriffith> rnirmal: cool!
16:44:29 <rnirmal> yeah that just breaks backward compatibility in config for the san drivers
16:44:42 <rnirmal> actually for all the other ones too
16:44:45 <jgriffith> rnirmal: Unless we go volume/driver/san right?
16:44:46 <rnirmal> if that's ok
16:44:59 <rnirmal> except for the ones in driver.py already
16:44:59 <jgriffith> Hmmm ?
16:45:24 <rnirmal> jgriffith: talking about moving all the other drivers under cinder/volume/driver/ too right?
16:45:44 <rnirmal> this leads to something jdurgin mentioned glance was doing
16:45:57 <jgriffith> rnirmal: yes
16:46:05 <rnirmal> not depending on the python modules structure to configure the drivers
16:46:14 <jgriffith> rnirmal: they all inherrit from the base VolumeDriver don't they?
16:46:21 <rnirmal> jgriffith: yes
16:46:51 <jgriffith> So I was thinking this layout would work by replacing driver.py with and  __init__ in volume/driver/
16:46:56 <jgriffith> but maybe I'm wrong
16:47:05 <jgriffith> Also interested in what you and jdurgin talked about
16:47:06 <rnirmal> jgriffith: yes it works for the ones already in driver.py
16:47:27 <jgriffith> ahh.. ok, just the ones that aren't there.  Hmm
16:47:48 <jgriffith> Well, IMO it's important enough to require a change at some point
16:47:59 <rnirmal> take for example solidfire... currently it is volume_driver=cinder.volume.solidfire.SolidFire
16:48:16 <rnirmal> it will become volume_driver=cinder.volume.driver.solidfire.SolidFire
16:48:40 <jgriffith> Yeah, and I have NO issue with that changing.  Unless the Glance model is something we could adopt
16:48:50 <rnirmal> yeah that's what I was thinking
16:49:02 <jgriffith> rnirmal: alright, I'll stop bugging you and let you sort it out :)
16:49:03 <winston-d> i guess that's the pain we have to endure sooner or later?
16:49:04 <rnirmal> if we are changing it.. might as well move to a better model than module names
16:49:15 <jgriffith> rnirmal: +1
16:49:27 * jgriffith gets out of the way
16:49:36 <rnirmal> winston-d: yes.
16:50:07 <jgriffith> winston-d: Yup, so we should do it *right* when we do it for sure
16:50:27 <winston-d> totally agree.
16:50:31 <jgriffith> cool
16:50:49 <jgriffith> alright, I have a hard stop coming up here so I'm going to move to the next topic
16:50:56 <jgriffith> #topic status updates
16:51:05 <rnirmal> docs?
16:51:15 <jgriffith> rnirmal: yes, docs :)
16:51:37 <rnirmal> jgriffith: I suppose that's a status update :P
16:51:45 <jgriffith> haha
16:51:56 <jgriffith> sadly the status is we haven't really made any progress there and we need to
16:52:37 <jgriffith> any volunteers?
16:52:43 <jgriffith> chirp, chirp, chirp....
16:53:09 <jgriffith> I'll make it easy, even if you just want to write up a short google doc on your features etc
16:53:19 <jgriffith> I'll put it into the official docs repo
16:53:44 <winston-d> i'll do that for scheduler
16:53:53 <jgriffith> winston-d FTW!!!!!
16:54:03 <winston-d> volume types, extra specs
16:54:11 <DuncanT> I have already foolishly volenteered to do a howto for BfV
16:54:29 <jgriffith> winston-d: I'll get the volume_types/extra-specs via the wiki I promised last week
16:54:37 <jgriffith> winston-d: You're plate is getting full I noticed :)
16:54:51 <jgriffith> DuncanT: Yes, you're now officially on the hook :)
16:55:02 <hemna> What is left to doc?
16:55:05 * jgriffith is happier now
16:55:09 <jgriffith> hemna: EVERYTHING
16:55:12 <jgriffith> :(
16:55:12 <hemna> lol
16:55:40 <jgriffith> hemna: If we can get the items we mentioned here though it will go a long way
16:55:47 <rnirmal> jgriffith: should this be coordinated with annegentle to get an official cinder doc
16:55:48 <jgriffith> hemna: One other thing would be install/config
16:55:51 <winston-d> so do we have a doc structure, like how many chapters, and what are they?
16:56:02 <rnirmal> winston-d: that's what I was getting at
16:56:11 <jgriffith> rnirmal: yes, but I'd like to just get something rough and then get her help with making it official
16:56:23 <jgriffith> she's already been very helpful and offered to work with us multiple times
16:56:30 <jgriffith> I just haven't had any content to provide :(
16:56:33 <hemna> I'd volunteer, but I haven't been plugged in long enough yet
16:56:42 * hemna is still getting up to speed
16:56:42 <rnirmal> jgriffith: understood cool.
16:56:56 <CaptTofu> Greetings all!
16:57:14 <jgriffith> So everybody take a look at docs.openstack.org
16:57:27 <jgriffith> That will give you an idea of what we *should* have, versus what we do have
16:57:56 <CaptTofu> specific link on that site?
16:58:05 <jgriffith> The Developer docs are in the cinder source... cinder/doc
16:58:20 <jgriffith> CaptTofu: That is the link: http://docs.openstack.org/
16:58:32 <Kiall> CaptTofu, the cinder meeting is still going on
16:58:39 <winston-d> i can provide some installation docs based on Folsom.
16:58:41 <CaptTofu> oh, so very sorry.
16:58:45 <jgriffith> haha
16:59:14 <jgriffith> winston-d: great, the biggest thing folks miss right now is /etc/tgt/conf.d/
16:59:30 <jgriffith> winston-d: and set_path was broken in ubuntu packaging
16:59:39 <jgriffith> and on that note....
16:59:44 <jgriffith> we should give up the room
16:59:48 <jgriffith> Thanks everybody!
16:59:56 <jgriffith> #endmeeting cinder