14:02:43 <markwash> #startmeeting glance
14:02:44 <openstack> Meeting started Thu Oct 24 14:02:43 2013 UTC and is due to finish in 60 minutes.  The chair is markwash. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:02:45 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:02:48 <openstack> The meeting name has been set to 'glance'
14:03:26 <markwash> sorry for the delay, doing some last minute agenda editiing
14:03:41 <markwash> #link https://etherpad.openstack.org/p/glance-team-meeting-agenda
14:04:13 <markwash> Anybody here who doesn't see their item in the agenda, feel free to add it now
14:04:53 <markwash> let's get started
14:05:00 <markwash> #topic project status meeting updates
14:05:33 <markwash> well, this was short this week, all I think we did was talk about summit scheduling
14:06:27 <markwash> which is an item further on in the list, so yeah
14:06:33 <markwash> #topic broken tests in master
14:07:47 <markwash> I think most people here are familiar by now, the iso8601 datetime parsing changed a bit between 0.1.4 and 0.1.8 (the next okay release)
14:08:35 <markwash> nikhil|afk and zhiyan1 were getting after it, but we didn't end up getting much help from openstack/requirements
14:09:05 <markwash> I sent out this
14:09:06 <markwash> #link http://lists.openstack.org/pipermail/openstack-dev/2013-October/017302.html
14:09:19 <markwash> seems like consensus there was that we should just update to 0.1.8 across the board
14:09:37 <zhiyan1> markwash: global requirement has been merged
14:09:42 <zhiyan1> >=0.1.8
14:10:00 <markwash> oh, hurray
14:10:02 <markwash> late breaking news
14:10:23 <iccha> lol
14:11:00 <zhiyan1> https://review.openstack.org/#/c/53567
14:11:17 <markwash> we had so many different patches submitted, I didn't realize :-)
14:11:22 <markwash> I just abandoned mine
14:11:32 <markwash> okay great, so the glance patch needed is. . .
14:12:21 <flwang> zhiyan1 has submit a patch to remove the broken test case
14:12:48 <zhiyan1> flwang: sorry i have not remove them, but enhance them
14:13:09 <markwash> #link https://review.openstack.org/#/c/52894/8
14:13:15 <flwang> zhiyan1: I mean the '2011-09-05' case
14:13:56 <markwash> okay, let's see if we can get that fixed up and in today
14:14:09 <zhiyan1> flwang: my first case does that YYYY-MM-DD
14:14:17 <markwash> is there anything else to talk about for that issue here today?
14:14:52 <flwang> zhiyan1: ok, got
14:15:18 <markwash> okay, moving on
14:15:27 <markwash> #topic design summit sessions
14:15:41 <markwash> #link http://summit.openstack.org/
14:15:57 <markwash> I did a run through last night approving, rejecting, and adding comments about merging topics
14:16:42 <markwash> but I think maybe the comments are not readable by an anonymous user?
14:18:04 <markwash> If you have feedback about my selections, please share it, either here or privately later if you prefer
14:18:50 <markwash> questions about the summit?
14:19:06 <zhiyan1> markwash: btw, if i have clear bp in mind but not proposed to session, i ok?
14:19:50 <markwash> zhiyan1: can you clarify that? you have a bp, but no session for it?
14:20:14 <zhiyan1> i mean we don't need discuss all bp plan in the session .. just make sure, and i don discuss that with you directly later..
14:20:35 <markwash> right, we don't necessarily have to discuss every icehouse bp in the summit
14:20:37 <zhiyan1> s/don/will , sorry
14:21:16 <markwash> some sessions were (gently) refused because I didn't think we were likely to have any fight about the idea
14:21:17 <iccha> markwash: how many slots do we have?
14:21:21 <markwash> iccha: 5
14:21:40 <markwash> so there were several great proposals for stuff that needs to follow on the async and import work
14:22:10 <markwash> and most of the time I just said "I think everyone is mostly on the same page, let's just follow up on <issue x> in meetings and on the mailing list."
14:22:37 <zhiyan1> (thinking, can i have ~10 mins for "indexing" async work? )
14:23:15 <markwash> zhiyan1: at the summit? or here today?
14:23:22 <zhiyan1> summit
14:23:29 <zhiyan1> ..within import work
14:24:05 <markwash> zhiyan1: I'm not sure, I don't see a great fit with the existing approved talks
14:24:13 <markwash> zhiyan1: but I think we'd all like to hear about it
14:24:28 <markwash> keep in mind, a lot of glance core actually won't be at the summit
14:24:36 <zhiyan1> yes, we can talk off line after session if you and nikhil|afk like
14:24:38 <markwash> so it might be more useful overall to try to bring it up in a different forum
14:24:42 <esheffield> l
14:24:45 <iccha> we have 5 in preapproved state, so looks like we re in good shape.
14:25:02 <markwash> okay, I think we can move on to the next topic
14:25:22 <markwash> #topic taskflow
14:25:48 <markwash> harlowja had the time to sit down and school me about taskflow this past week
14:26:12 <markwash> I think, given its traction with other projects and its similarity with what we're trying to accomplish, it makes a lot of sense to use
14:26:16 <markwash> I was a bit hesitant before
14:26:30 <markwash> I think the question for us is, to what extent can we integrate in Icehouse
14:26:39 <iccha> what were the main take aways from the conversation?
14:27:02 <markwash> the main takeaway was that there are rather 3 levels of integration that might make sense
14:27:20 <markwash> 1) just breaking up and defining the work to be done as tasks and flows, similar to what cinder has right now
14:27:28 <markwash> 2) actually using an engine to construct the flows
14:27:48 <markwash> 3) something complicated that i don't remember… maybe having the flows be more portable to deal with machine failures?
14:27:59 <markwash> it seemed like the question for us for Icehouse was, #1 or #2?
14:28:24 <markwash> I think at #2, we might expect that the engines to more or less replace our executor code
14:28:26 <zhiyan1> markwash: cool. saw your message with harlowja, so seems you(s) plan not implement worker in glance internal, at stage 2, just waiting taskflow implement 'RPC worker/engine' part in I, and we focus on stage 1, right?
14:28:55 <rosmaita> markwash: what are the deployment expectations with taskflow?
14:29:05 <rosmaita> would it live on glance node?
14:29:23 <markwash> zhiyan1: I think its still an important question, whether we implement a worker or just use an RPC engine
14:29:30 <zhiyan1> iirc 3) is "job posting" mechanism
14:29:32 <markwash> zhiyan1: and probably depends as much on the timing as anything
14:29:58 <markwash> rosmaita: I think the deployment expectations are the same
14:30:31 <markwash> rosmaita: i.e. the deployer has the option to have the work done either on the api node (like our eventlet model before) or remotely on a pool of worker nodes
14:30:45 <rosmaita> ok
14:30:48 <markwash> rosmaita: does that agree with your understanding of it / concerns?
14:31:15 <zhiyan1> markwash: ok (i have a topic for that, we can discuss glance-worker on session)
14:31:18 <rosmaita> my concern is mainly not requiring something too heavy-duty as part of glance deployment
14:31:35 <iccha> +1
14:32:07 <markwash> I don't want this taskflow thing to be a disruptor in our current work, so let me know if that's seeming to be the case
14:32:13 <rosmaita> i am worried about building too long a dependency chain on incubating projects
14:32:23 <iccha> if we decide to use it , it should be pluggable
14:32:37 <markwash> re: heavy duty, I think we're going to be okay there as long as we're just using it to define the task scripts
14:32:43 <iccha> not a required dependency
14:33:24 <markwash> iccha: that makes sense to me for the engines, but not as much for the simpler case of script definition
14:33:51 <markwash> if you're just doing step #1 above, its very lightweight in a deployment and dependency sense
14:33:57 <iccha> ok gotcha.. i need to do more homework on taskflow
14:34:25 <markwash> #link https://github.com/openstack/cinder/blob/master/cinder/volume/flows/create_volume/__init__.py
14:34:27 <markwash> ^^ might help
14:34:36 <markwash> okay, other thoughts on taskflow
14:34:47 <markwash> again, I don't want this to be seen as moving the target for our current work
14:34:49 <iccha> thanks markwash
14:35:09 <markwash> so if it is, let's pull back and make sure taskflow isn't disrupting our plans
14:35:47 <rosmaita> i think it's important to get import task done, then revisit taskflow
14:36:17 <rosmaita> once we see an actual workflow in actino
14:37:00 <markwash> rosmaita: hmm, okay. . but I think we should keep it in mind becuase it might actually make the import script coding part easier and faster to do from where we are now
14:37:28 <markwash> at least when we start to look at doing the more complex parts (i.e. plugins for conversion / validation / etc)
14:38:02 <markwash> but yeah, if we need to move ahead without taskflow and revisit it later in the cycle I'm fine with taht
14:38:07 <zhiyan1> markwash: actually rosmaita's message ask me remember my question 1 and 2 within http://lists.openstack.org/pipermail/openstack-dev/2013-October/016917.html
14:39:10 <zhiyan1> markwash: ... they are pending still after our last discussing.. humm
14:39:46 <markwash> zhiyan1: yes, I'm not sure I'm ready to follow up on that. . I still feel generally that we should not allow plugins to define new task types
14:40:20 <markwash> because it seems too diffusive, I want different deployments to feel the same
14:40:21 <zhiyan1> let's talk this off line... seems it need more time.
14:40:34 <markwash> okay, moving on
14:40:54 <markwash> #topic async workers
14:40:57 <zhiyan1> markwash: make a action between us?
14:41:09 <markwash> I keep looking for stuff to review, but its always based on an outdated patch
14:41:13 <markwash> am I just bad at finding the right review?
14:42:02 <markwash> zhiyan1: I'm not sure exactly what I would put in the action item
14:42:11 <flwang> markwash: I think you can start from https://review.openstack.org/#/c/43842/ and  https://review.openstack.org/#/c/46224/ :)
14:43:02 <markwash> flwang thanks!
14:43:42 <markwash> any other quick things to note about async workers progress? I think we've mostly been blocked at the gate this week :-(
14:44:31 <markwash> okay
14:45:06 <markwash> #topic doc review
14:45:16 <markwash> rosmaita is this your item?
14:45:24 <markwash> I get a 404 on the link :-(
14:45:48 <markwash> I think maybe the last number was clipped somehow
14:45:53 <rosmaita> https://review.openstack.org/#/c/51704
14:46:08 <rosmaita> sorry about that, chief!
14:46:09 <markwash> good ol gate-noop
14:46:11 <markwash> no worries
14:46:18 <markwash> folks: have a look at that review, I will too
14:46:31 <markwash> rosmaita: anything else to cover other than a shout out for review?
14:47:01 <rosmaita> not from me
14:47:32 <markwash> #topic tasks authorization layer
14:47:47 <markwash> hmm, I think nikhil|afk suggested this? but is still afk
14:48:44 <iccha> I think that is soemthing we can consider for glance as a whole, not necessarily just tasks
14:49:25 <markwash> ah, okay
14:49:26 <iccha> but I am not sure how that review comes into picture.
14:49:27 <zhiyan1> iirc iccha yesterday give some great response to explain why we need this layer ...
14:49:40 <markwash> I thought so too
14:49:58 <iccha> maybe nikhil|afk had something else in mind wrt review, I am not sure
14:50:06 <markwash> let's table it for now
14:50:42 <markwash> #topic open discussion
14:51:10 <markwash> I finallly proposed a session about v1
14:51:24 <markwash> please let me know if you think its a dud and I should let some other talk take the space
14:52:13 <iccha> if some cross project talks come up we can consider, otherwise I think its ok because most ppl still use v1
14:53:02 <zhiyan1> markwash: after fast glance, IMO v1 is useful, since some function only be supported by v1 but v2..
14:53:58 <iccha> I agree till we have import and export we may wanna keep it around
14:53:58 <zhiyan1> i mean this topic is valuable probably, we can talk details on the session
14:54:04 <markwash> zhiyan1: well a main thing I would like to stop doing is implementing new functionality both the right way (in a code library like the domain) and the wrong way (in the v1 controller)
14:54:09 <iccha> +1
14:54:16 <iccha> just makes v1 clunkier
14:54:26 <iccha> if clunkier is a word :p
14:54:29 <markwash> +1
14:54:33 <zhiyan1> freeze v1, +1
14:54:48 <markwash> zhiyan1: yeah, freeze is a good option to consider, good point
14:55:01 <flwang> +1 to avoid dup work anymore
14:55:37 <flwang> btw, when the glance client will use v2 as the default?
14:56:21 <markwash> hmm good question
14:56:36 <markwash> oh I forgot to talk about glanceclient (which I of course neglected to release again :-()
14:56:44 <zhiyan1> iccha: btw "clunkier" ?
14:57:17 <flwang> I think before we let other components using v2, we should eat the dog food by ourself :)
14:57:18 <iccha> zhiyan1: http://www.thefreedictionary.com/clunkier
14:58:11 <markwash> okay, time for me to get some breakfast
14:58:15 <markwash> thanks folks
14:58:17 <markwash> #endmeeting