16:00:07 <jgriffith> #startmeeting cinder
16:00:08 <openstack> Meeting started Wed Apr 24 16:00:07 2013 UTC.  The chair is jgriffith. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:00:09 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:00:11 <openstack> The meeting name has been set to 'cinder'
16:00:16 <bswartz> hi
16:00:17 <jgriffith> Hey everyone!  Welcome back
16:00:20 <dachary> \o
16:00:22 <vincent_hou_> hi
16:00:22 <eharney> hi
16:00:37 <jgriffith> thingee is on a train so may be a few minutes late
16:00:37 <bswartz> yes I hope everyone had uneventful travel
16:00:41 <flaper87> o/
16:00:43 <kmartin> hello
16:00:44 <jgriffith> bswartz: indeed
16:00:49 <rushiagr> hi!
16:01:04 <vincent_hou_> yesterday i had a good sleep
16:01:11 <jgriffith> vincent_hou_: hehe
16:01:15 <jgriffith> jet lag is tough
16:01:23 <jgriffith> Ok...  let's get rolling
16:01:26 <jgriffith> #topic blueprints
16:01:37 <jgriffith> So we talked about a lot of things at the summit
16:02:02 <jgriffith> We need to get our thoughts/plans transferred in to blueprint form this week if we can
16:02:03 <thingee> hello
16:02:08 <jgriffith> thingee: hey... right on time
16:02:22 <jgriffith> thingee: and I said you were going to be late :(
16:02:48 <jgriffith> So I'm going to ask that everybody that had a topic they brought up to do a formal bp
16:02:58 <jgriffith> also, I'd like to do the BP's in chunks
16:03:10 <jgriffith> in other words if you have a large thing like volume migration
16:03:22 <jgriffith> break it down in to multiple dependent bp's
16:03:40 <dachary> ( avishay is on vacations for the next two weeks )
16:03:47 <jgriffith> dachary: I'm painfully aware
16:03:58 <bswartz> d'oh
16:03:59 <jgriffith> anybody know how to turn off his damn auto-reply :)
16:04:08 <dachary> ahah
16:04:09 <rushiagr> jgriffith: through his out of office replies?
16:04:12 <rushiagr> :P
16:04:17 <jgriffith> rushiagr: yes :)
16:04:29 <jgriffith> He's spamming launchpad as well as me :)
16:04:29 <bswartz> do any blueprints exist for any of the topics we discussed?
16:04:35 <jgriffith> bswartz: not really
16:04:46 * dachary write m. factor
16:04:47 <jgriffith> that's why I'm mentioning it
16:05:01 <jgriffith> I'd like to get formal BP's together and start targetting them
16:05:18 <jgriffith> and I don't want to do them all on my own :)
16:05:26 <kmartin> bswartz: yes, we have entered a few
16:05:35 <bswartz> I think the migration stuff will depend on the volume attach stuff, so migration doesn't feel like a H1 thing -- I'd suggest H2
16:05:41 <thingee> yes may 30th is due for h1, and we got a lot of work to do
16:05:50 <thingee> https://wiki.openstack.org/wiki/Havana_Release_Schedule
16:05:53 <kmartin> 38 in the current list
16:06:13 <kmartin> bswartz: https://blueprints.launchpad.net/cinder
16:06:40 <jgriffith> kmartin: thanks :)
16:07:00 <jgriffith> some of those are going to drop as hold overs from last release
16:07:12 <jgriffith> there's a number that I know folks have abandoned
16:07:42 <jgriffith> anyway....  my point is that if you have an item from the summit that you were going to work on
16:07:59 <jgriffith> make sure you update/create a bp and we'll get it targetted etc
16:08:21 <jgriffith> The plan this time around is to try and get as much of that work done up front
16:08:31 <jgriffith> Rather than just adding BP's every day as we go
16:08:34 <bswartz> jgriffith: +1
16:08:45 <thingee> jgriffith: cut off date
16:08:49 <jgriffith> and it will help with the folks that are coming along and saying "how can I help"
16:08:59 <jgriffith> thingee: ideally next Tuesday
16:09:08 <jgriffith> thingee: but that wouldn't be a *freeze* on bp's
16:09:30 <jgriffith> That's just when we need to have a reasonable outline for what we're doing in H
16:09:43 <thingee> excellent
16:09:50 <kmartin> some of us have to get legal approval before posting the BPs, but we are working on it. :(
16:10:08 <rushiagr> jgriffith: H, or H1?
16:10:10 <jgriffith> kmartin: you can't even post it?
16:10:19 <bswartz> kmartin: :-(
16:10:20 <kmartin> jgriffith: nope
16:10:25 <jgriffith> kmartin: what if you just let me know what you want to do and I'll post the BP for you
16:10:31 <jgriffith> Then when you get approval we can assign it
16:10:31 <bswartz> lol
16:10:38 <jgriffith> would that solve the legal issues?
16:10:50 <kmartin> ok, add QoS support to the 3PAR drivers
16:11:05 <jgriffith> kmartin: hrmph
16:11:17 <jgriffith> ok, that one might be a little tough for me to wing it in the BP
16:11:23 <jgriffith> but at least maybe a place-holder
16:11:32 <kmartin> jgriffith: let's try it
16:11:40 <jgriffith> kmartin: sounds good
16:11:51 <jgriffith> I'm also not as concerned about driver updates/enhancements
16:11:55 * dachary proposes to act as a legal proxy too, if that helps ;-)
16:11:59 <jgriffith> I'm more interested in project wide changes
16:12:25 <jgriffith> Does that make sense to everyone?
16:12:31 <jgriffith> Or do folks disagree with that?
16:12:35 <bswartz> Avishay did make a blueprint for his volume-migration topic: https://blueprints.launchpad.net/cinder/+spec/volume-migration
16:12:40 <dachary> +1
16:13:00 <kmartin> agree..that's what I thought. We'll added the general one's already we'll ask for forgiveness later
16:13:01 <thingee> jgriffith: +1
16:13:14 <jgriffith> bswartz: yes, and that's a prime candidate for breaking into dependent chunks IMO
16:13:21 <jgriffith> kmartin: :)
16:13:39 <jgriffith> kmartin: lemme know how to help to keep things above board from the legal teams perspective
16:13:55 <thingee> jgriffith: are we committing that beyond h1 it would be unlikely accepting bps that involve project wide changes?
16:14:06 <thingee> or h2?
16:14:24 <jgriffith> thingee: I don't know for sure yet... I'd be curious what others thought
16:14:26 <kmartin> if we have feedback on some what's the process, reach out to the assignee or drafter?
16:14:32 <jgriffith> Personally H1 would be fantastic
16:14:41 <jgriffith> and of course there would be an exception process
16:14:51 <bswartz> I think it's reasonable to require a blueprint in H1 for project-wide changes, even if the implementation isn't done until H2 -- I'm thinking of new features that depend on other new features
16:14:54 <jgriffith> That wouldn't be "implemented" but at least planned
16:15:10 <jgriffith> bswartz: +1  that's kinda the direction I'm leaning
16:15:24 <jgriffith> thingee: does that sound good to you?
16:15:37 <jgriffith> kmartin: DuncanT everyone?
16:15:48 <DuncanT> Yup
16:15:56 <thingee> jgriffith: yea
16:15:57 <kmartin> sounds good we can always push to H2
16:16:11 <rushiagr> jgriffith: yes
16:16:17 <jgriffith> rushiagr: cool
16:16:18 <eharney> seems reasonable
16:16:33 <jgriffith> alright, let's plan on that.  Seems like folks are pretty good with the idea
16:16:41 <jgriffith> So we're saying:
16:17:03 <jgriffith> major functional changes/enhancments to the core project are expected to have BP's by H1
16:17:24 <jgriffith> After that major changes to the core project will need to have some discussion/exception process
16:17:46 <jgriffith> This does NOT mean the changes are implemented and in by H1, just that we have plans for them
16:18:13 <jgriffith> Ideally the bulk of these will be in later this week, but we're sure to think of "new" things in the coming weeks
16:18:45 <jgriffith> anything else on BluePrints/plans for H?
16:19:16 <jgriffith> Of course I'm speaking in general, not details of some of the items on our plate right now :)
16:19:50 <rushiagr> jgriffith: do we cover drivers also in this? e.g. a new vendor trying to get his driver in in H3?
16:20:11 <jgriffith> rushiagr: so drivers I'm not as worried about having a BP for by next week
16:20:28 <jgriffith> rushiagr: I would like us to have a seperate discussion on when we cut-off new drivers though
16:20:47 <jgriffith> rushiagr: and the difference between a *new* driver and a refactored or modified existing driver
16:20:48 <DuncanT> We were discussing having no driver interface changes after H2 as well... did that get agreed on in the end?
16:20:54 <bswartz> I think driver should be allowed to go in any time before H3, as long as it's not the last week of H3 -- then you get beaten with a nerf bat
16:21:19 <kmartin> on that topic, thingee mentioned that we should document https://wiki.openstack.org/wiki/Cinder#Minimum_Driver_Features in a dev doc or something that is checked in
16:21:26 <rushiagr> jgriffith: sounds reasonable
16:21:52 <jgriffith> bswartz: My concern is that like last time
16:22:02 <jgriffith> bswartz: the third milestone becomes all drivers all the time
16:22:12 <dachary> jgriffith: does it make sense to synchronize wih other components regarding this H1 deadline for project wide BP ? I'm thinking scheduling as an example because it involves at least nova & cinder, even for the case of local volumes. Or would this be handled as exceptions ?
16:23:03 <jgriffith> dachary: the other projects are going to be shooting for the same time-lines regarding bp's and features
16:23:25 <dachary> #link https://etherpad.openstack.org/havana-cinder-local-storage-library
16:24:12 <jgriffith> dachary: yes, and I've already started bp's for oslo and nova for this
16:24:24 <dachary> ok :-)
16:24:33 <jgriffith> still need the cinder version and prototyping some things to see how I really want to do this
16:24:46 <jgriffith> ^^ cinder-version == bp
16:24:48 <uvirtbot> jgriffith: Error: "^" is not a valid command.
16:24:56 <jgriffith> whatevs uvirtbot
16:24:58 <rushiagr> uvirtbot: haha!
16:25:00 <uvirtbot> rushiagr: Error: "haha!" is not a valid command.
16:25:06 <rushiagr> oops
16:25:08 <rushiagr> :/
16:25:11 <hemna> heh
16:25:12 <jgriffith> ha!
16:25:26 <jgriffith> guitarzan: any opinions on this ?
16:25:36 <jgriffith> not the virtbot discussion :)
16:25:43 <guitarzan> I was curious what cinder version was referring to
16:25:55 <jgriffith> a bp for cinder, :)
16:26:16 <guitarzan> I think getting blueprints in early for big stuff makes a lot of sense
16:26:21 <jgriffith> cool
16:26:27 <jgriffith> alright, we have a plan
16:26:29 <hemna> yup
16:26:33 <hemna> like the state machine :P
16:26:42 <jgriffith> hemna: +100000
16:26:57 <guitarzan> indeed
16:26:59 <DuncanT> Am I writing a bp for that or is somebody else?
16:27:04 <jgriffith> #topic open discussion
16:27:10 <hemna> I'd like to talk about that today with you at some point
16:27:15 <jgriffith> DuncanT: I was going to, we can just see who gets to it first
16:27:33 <jgriffith> DuncanT: I was just going to have a simle BP like:  "implement a state machine for Cinder"
16:27:34 <hemna> I have a BP I wrote as a place holder for the state machine
16:27:40 <jgriffith> hemna: right on
16:27:49 <jgriffith> hemna: and yes, we can chat about that today
16:28:04 <jgriffith> Unless you want to do it now while everybody is here
16:28:10 <DuncanT> jgriffith: Fair enough. We've a meeting tomorrow to work out what we need to write up and who is going to do it... If you've already started I can always add to it
16:28:31 <hemna> https://blueprints.launchpad.net/cinder/+spec/cinder-state-machine
16:28:32 <hemna> fyi
16:28:34 <jgriffith> DuncanT: sounds good
16:28:44 <jgriffith> hemna: ha!
16:28:47 <jgriffith> hemna: wins!!
16:28:51 <hemna> :P
16:28:57 <guitarzan> yeah, I noticed he'd already gotten that one in :)
16:29:06 * jgriffith has yet to go through the bp list :(
16:29:38 <hemna> I spent 10 minutes yesterday googleing for python FSM's yesterday
16:29:38 <jgriffith> alright... anybody have anything else?
16:29:47 <kmartin> thingee mentioned at the summit that we should document https://wiki.openstack.org/wiki/Cinder#Minimum_Driver_Features in a dev doc or something that is checked in instead of the wiki
16:29:50 <jgriffith> hemna: haha!  There's no shortage
16:29:56 <hemna> yah
16:30:05 <jgriffith> kmartin: yes sir!!
16:30:13 <hemna> we need to just decide what we want out of it.
16:30:14 <jgriffith> kmartin: so since I'm very unpopular on this topic :)
16:30:25 <vincent_hou_> How does cinder manage the event so far?
16:30:27 <hemna> kmartin, +1
16:30:32 <jgriffith> kmartin: my thought here was that I would bring it up as a TC disucssion
16:30:36 <jgriffith> discussion
16:30:45 <kmartin> ok...anyone know where the doc is, I'll update it
16:31:19 <kmartin> jgriffith: ok check with the TC
16:31:23 <thingee> jgriffith: +1
16:31:33 <thingee> kmartin: cinder/doc in the repo
16:31:37 <jgriffith> You all know my thoughts :)
16:32:30 <thingee> kmartin: I've already started updating the docs with volume_stats...I was mainly concerned with that being consistent :)
16:32:35 <jgriffith> vincent_hou_: any thoughts on the comments regarding your download image patch?
16:33:17 <thingee> kmartin: I want eyes once that review is posted to make sure people are fine with that before we start enforcing that. Mainly so I have somewhere to point people in reviews :)
16:33:18 <jgriffith> https://review.openstack.org/#/c/25365/
16:33:25 <vincent_hou_> jgriffith: I would like to add a condition check to make sure the volume is a proper place.
16:33:47 <jgriffith> vincent_hou_: what do you mean by that?
16:33:50 <vincent_hou_> For example, it is brand new without any modifications.
16:34:04 <jgriffith> vincent_hou_: but then I would ask even more... why?
16:34:08 <kmartin> thingee: great...DuncanT said  he would to pull the drivers out that didn't meet the requirement, but that was a few beers in at one of the parties
16:34:16 <jgriffith> vincent_hou_: why not just create a new volume?
16:34:18 <vincent_hou_> It the volume has data on it, it cannot do the download.
16:34:27 <jgriffith> vincent_hou_: understand
16:34:37 <thingee> kmartin: I don't mind being "that guy".
16:34:52 <jgriffith> thingee: is hereby elected as "that guy"
16:35:02 <hemna> lol
16:35:16 <kmartin> thingee: +1
16:35:31 <dachary> ahaha
16:35:45 <vincent_hou_> if a volume is already created, just brand new without doing anything on it. Then it can be a destination where an image is downloaded into.
16:36:07 <DuncanT> kmartin: I'm still quite happy to do it
16:36:31 <that_guy> DuncanT: we'll flip a coin.
16:36:40 <guitarzan> you already got the nick
16:36:41 <DuncanT> vincent_hou_: We don't track whether a volume has stuff on it... once it has been attached, how could we?
16:36:49 <DuncanT> that_guy: Swords at dawn
16:37:01 <jgriffith> vincent_hou_: DuncanT I guess I'm just not sure of the value
16:37:12 <DuncanT> vincent_hou_: And if it hasn't got anything on it, ignore it and create a new volume
16:37:19 <DuncanT> jgriffith: I agree
16:37:19 <jgriffith> vincent_hou_: DuncanT sure, we could look at last attach entry but why?
16:37:39 <hemna> If it's not attached, then I'd assume it's available for whatever.
16:37:41 <vincent_hou_> OK. i see the point.
16:37:56 <jgriffith> so this goes back to my rant about being "cloudy"
16:37:58 <guitarzan> new volume +1
16:38:03 <jgriffith> if that makes sense to anyone
16:38:15 <jgriffith> The idea is abstraction and elasticity
16:38:18 <hemna> you aren't really supposed to care what's on it, or where it came from (from a user's perspective)
16:38:41 <DuncanT> jgriffith: I totally agree
16:38:45 <jgriffith> kk
16:38:51 <jgriffith> vincent_hou_: you ok with that?
16:39:03 <jgriffith> vincent_hou_: If you have a real compelling use case we can take another look
16:39:26 <jgriffith> vincent_hou_: I'm just not seeing the value/use case for it other than "because we can"
16:39:28 <vincent_hou_> yes.
16:39:34 <jgriffith> Ok... cool
16:39:55 <jgriffith> vincent_hou_: this is likely to come up again
16:40:06 <jgriffith> vincent_hou_: so keep the code around :)
16:40:17 <vincent_hou_> sure.
16:40:35 <jgriffith> that_guy: I'm going to change all of my filters/highlights now
16:40:45 <jgriffith> s/thingee/that_guy/g
16:41:14 <jgriffith> kk... anything else from anyone?
16:41:27 <jgriffith> Doh!!!
16:41:34 * jgriffith goes back and changes everything again
16:41:35 <jgallard> FYI, I'm working on multi backend tempest tests : https://review.openstack.org/#/c/23923/
16:41:46 <jgriffith> jgallard: yes, I've been following that
16:41:49 <thingee> jgallard: yes thanks for that :)
16:41:52 <jgriffith> jgallard: thanks a ton for working on those
16:41:57 <jgallard> you'r welcome!
16:42:00 <bswartz> jgriffith: who own the "brick" project / refactor attach stuff?
16:42:14 <jgriffith> bswartz: I'm actively working on it
16:42:24 <bswartz> jgriffith: is there a BP yet?
16:42:37 <jgriffith> bswartz: that's what we were saying earlier
16:42:47 <jgriffith> bswartz: I have one up for oslo, need to put one up for cinder and nova
16:42:53 <thingee> https://github.com/j-griffith/brick
16:43:16 <hemna> ooh we have brick already?
16:43:18 <hemna> ship it!
16:43:19 <bswartz> k
16:43:25 <jgriffith> that's subject to look rather different by the end of the day :)
16:43:28 <jgriffith> I'll get a bp up
16:43:39 <jgriffith> alright.. I gotta run unfortunately
16:43:43 <bswartz> sounds good
16:43:50 <jgriffith> anything else real quick?
16:44:18 <jgriffith> Ok.. I'll be back in about 15 minutes on cinder channel
16:44:20 <jgriffith> thanks everyone
16:44:23 <jgriffith> #endmeeting