20:02:06 <markwash> #startmeeting glance
20:02:07 <zhiyan> and rosmaita!
20:02:08 <openstack> Meeting started Thu Jan  9 20:02:06 2014 UTC and is due to finish in 60 minutes.  The chair is markwash. Information about MeetBot at http://wiki.debian.org/MeetBot.
20:02:09 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
20:02:11 <openstack> The meeting name has been set to 'glance'
20:02:20 <arnaud> hi
20:02:25 <markwash> greetings everyone
20:02:28 <markwash> happy 2014
20:02:53 <markwash> agenda link
20:02:55 <markwash> #link https://etherpad.openstack.org/p/glance-team-meeting-agenda
20:03:50 <markwash> a lot of boiler plate to go through I think
20:03:56 <markwash> so I'll go ahead and get started
20:04:05 <markwash> #topic mini summit
20:04:20 <markwash> I hope folks saw my email announcing the details for the mini summit
20:04:34 <markwash> #link http://lists.openstack.org/pipermail/openstack-dev/2013-December/023153.html
20:04:47 <markwash> I've kicked off a draft agenda here
20:04:57 <markwash> #link https://etherpad.openstack.org/p/glance-mini-summit-agenda
20:05:01 <arnaud> do you have an idea approximately how many people will attend?
20:05:04 <markwash> it had some initial looks and feedback
20:05:14 <markwash> arnaud: I think the upper end is around 25 folks
20:05:22 <arnaud> ok sounds good!
20:05:23 <markwash> probably somewhere in the high teens
20:06:06 <markwash> so we need your help to make this meeting a success, have a look at the agenda, add to it, add your interest
20:06:15 <markwash> if there is anything that seems missing let me know
20:06:38 <markwash> I think by next week we should have enough to start actually scheduling so folks know what they need to prepare in terms of discussion notes / slides / etc
20:06:52 <iccha> sounds good
20:07:09 <markwash> this is the first time doing a mini summit for. . um. . all of us? so I guess we'll be playing most of the parts by ear :-)
20:07:25 <markwash> so we'll try to have fun :-)
20:07:26 <rosmaita> it's just like a regular summit, only mini
20:07:30 <markwash> any questions about the summit?
20:08:20 <iccha> so will we have separate time apart from these talks for discussions i hope ( i presume yes_
20:08:56 <markwash> yes, I think so
20:09:03 <markwash> there will be some downtime for sure
20:09:05 <markwash> and meals etc
20:09:36 <markwash> if you like, we can leave open some time for informal lightning talks etc
20:09:37 <arnaud> for now, there are ~10 talks, if all of them are 1h, 10h in 2 days, it seems there is time left for discussion
20:10:04 <markwash> I also thought there might be some room if folks want to do a little hackathon type thing
20:10:26 <markwash> but I'm not exactly sure what form that would take, and would like to have a good idea before we devote time to it
20:10:34 <markwash> anyway, food for thought in case anyone especially likes the idea
20:10:53 <markwash> moving on for today
20:10:58 <markwash> #topic Icehouse 2 is closing soon
20:11:01 <zhiyan> markwash: is there some plan to allow people "attend" the summit remotely who can't go there? like a remote video will be great
20:11:24 <markwash> zhiyan: hmm, no plan as yet but maybe we can figure something out? it will probably be a terrible timing for you :-)
20:11:27 <arnaud> +1 do we plan to at least record the talks?
20:11:28 <iccha> maybe set up etherpads for notes. or use the same one
20:11:48 <markwash> zhiyan: but since its a small group it is possible that it would be easier to share e.g. with a google hangout or skype or some other software
20:12:13 <zhiyan> humm...i'm ok for the time, if we have way to attend remotely, i will join on that time :)
20:12:19 <markwash> heh cool
20:12:53 <markwash> iccha: perhaps some rackspace video equipment could be commandeered
20:12:55 <zhiyan> arnaud: good idea
20:13:32 <iccha> markwash: fir recording or transmission? will ask around :)
20:13:44 <markwash> recording I guess is probably easiest and most important
20:13:51 <markwash> but I would love transmission if somebody knows how to do that
20:14:01 <zhiyan> markwash: that will be great, transmission. iccha
20:14:12 <markwash> we are only going to have 1 room / 1 track so transmission might be easier than with the regular summits
20:14:33 <zhiyan> markwash: indeed
20:15:00 <markwash> anyway, sometimes this stuff turns into a disaster, not much better than nothing, so let's not sink too much into it unless we feel confident
20:15:13 <markwash> this stuff -> trying to support full remote participation
20:15:30 <markwash> recording is a clear win, so if that's all we can get so be it
20:15:49 <markwash> any further mini summit thoughts?
20:16:23 <iccha> will ask about recording and let u know markwash
20:16:31 <markwash> cool
20:16:39 <markwash> On to Icehouse 2
20:16:50 <markwash> all my thoughts are there on the meeting agenda page
20:16:58 <markwash> #link https://launchpad.net/glance/+milestone/icehouse-2
20:17:12 <markwash> we've got less than 2 weeks, since we'll probably want to cut on the 20th
20:17:27 <markwash> the gate has been really rough lately, even without a crunch time
20:17:35 <markwash> so I'm expecting very bad weather, so to speak
20:17:56 <markwash> if you want to land something, you probably need to get it into the gate queue by the middle of next week
20:18:09 <markwash> and reviewers need to be on hand to help with that process
20:18:24 <markwash> I will note, we have some blueprints that aren't started yet
20:18:48 <markwash> should we consider deferring https://blueprints.launchpad.net/glance/+spec/new-download-workflow ?
20:18:54 <markwash> or is the work just not up to date?
20:19:15 <markwash> also, flaper87 was working on https://blueprints.launchpad.net/glance/+spec/split-image-size but I haven't seen a patch yet, am I just missing it?
20:19:29 <iccha> we do not have import script merged yet. so maybe we will have to jump similar hurdles for export
20:19:43 <iccha> i know nikhil started work on it but not sure where he is at
20:21:19 <markwash> hmm okay
20:21:32 <markwash> well, if I don't hear back by tuesday I'm going to defer those two
20:21:50 <markwash> any other thoughts on icehouse-2? questions?
20:22:26 <arnaud> I think it would be good at the minisumit to have some time allocated to look at what is done, what was expected, what is left
20:22:29 <arnaud> for icehouse
20:22:42 <rosmaita> +1
20:22:47 <markwash> arnaud: good idea
20:23:24 <markwash> that should be a good motivating discussion
20:23:36 <iccha> yeah could tie into the blueprint discussion
20:23:38 <markwash> and help constrain our wilder plans on some of the other topics
20:23:55 <iccha> ashwini: says videoeqipment is a yes!
20:24:07 <markwash> huzzah, thanks ashwini
20:24:08 <arnaud> nice :)
20:24:35 <ashwini> sorry joining let but yes i will take care of that
20:24:44 <ashwini> s/let/late
20:24:46 <zhiyan> ashwini: iccha thanks
20:24:50 <markwash> okay cool, let's take a look at the review queue, my favorite topic :-P
20:24:57 <markwash> #topic review backlog
20:25:13 <markwash> I posted a bunch of links in the agenda, not going to repost here
20:25:24 <markwash> but I think something ameade brought up in the past is still very relevant
20:25:43 <markwash> I think we're still just trimming our queue with "abandoned due to inactivity"
20:25:46 <markwash> sad times
20:26:10 <markwash> we've had queue growth over the past 3 months
20:27:01 <markwash> I'm a bit afraid we're only going to recover from this when we figure out how to block out some of the noise
20:27:10 <markwash> in terms of patches that are more churn than value
20:27:26 <markwash> hopefully some automation around bugs, blueprints, and patches can help that
20:27:41 <markwash> Anyone have thoughts on how we can fix our review backlog?
20:28:12 <iccha> is the problem not enough reviews or not enough reviewers or neither or both?
20:29:14 <iccha> Glance Core team size: 10 (avg 1.0 reviews/day)
20:29:23 <iccha> Nova Core team size: 19 (avg 2.5 reviews/day)
20:29:26 <iccha> for 90 days
20:29:46 <markwash> I suppose the glance core team is a bit inflated, we probably have more like 5 active core members
20:29:48 <arnaud> some of the core glance are not reviewing code right?
20:29:56 <markwash> which would bring us up to 2.0 reviews/day
20:30:33 <markwash> the whole thing needs to be considered against our patch volume as well
20:31:09 <markwash> we get about 15 patchsets a day
20:31:21 <markwash> and we get about 4 patches per change
20:31:27 <markwash> so that's really only about 4 changes per day
20:31:36 <markwash> oh, 3.7 per day over the past 90 days
20:32:10 <markwash> of course, the number of reviews needed is something like patchsets / day + changes / day
20:32:29 <iccha> Changes abandoned in the last 90 days: 97 (1.1/day)
20:32:30 <markwash> so that has us needing 18.5 reviews a day
20:33:10 <markwash> I can see a few ways to affect this
20:33:24 <markwash> more reviews (more reviewers * more reviews per reviewer)
20:33:27 <markwash> fewer changes
20:33:33 <markwash> fewer patchsets per change
20:34:00 <arnaud> how can you control fewer changes?
20:34:06 <nikhil___> hi
20:34:10 <iccha> more reviews on a given patchset
20:34:16 <iccha> or more complete reviews
20:34:40 <markwash> arnaud: I think we can potentially have some automatic -2s that essentially remove changes from consideration
20:34:50 <arnaud> I see, interesting
20:34:56 <markwash> arnaud: its essentially figuring out a way to raise the barrier to entry
20:35:05 <markwash> which isn't considered a good thing by many
20:35:20 <arnaud> yes I see, a queue before the queue :)
20:35:24 <markwash> however, I just don't see how we can have much in the way of sanity if our review queue isn't averaging something close to zero
20:35:28 <iccha> for example jenkins job fialing
20:35:33 <iccha> on py27 tests or soemthing
20:36:45 <markwash> iccha: I do feel like maybe something along the lines of better blueprint triage, like you've been looking into, could allow us to say "auto -2" to things that didn't have approved blueprints
20:36:58 <markwash> I think better bug triage is another part of that picture
20:37:39 <markwash> since if we tightly control only one of blueprints or bugs, then the uncontrolled one becomes a shunt and everyone with an untagged change will just tag it as a bugfix or blueprint, whichever is easiest
20:37:58 <markwash> without really adding any value in terms of up-front consideration
20:38:38 <markwash> oh well, again more food for thought. . if the 5 active core folks we had just all tried to get 4 reviews per day, I think the queue would gradually go down
20:38:39 <iccha> is there a repo for gerrit or all this automation?
20:38:59 <markwash> I'm trying to do 10-20 a day to get down to a lower queue level but I burn out kinda fast
20:39:13 <zhiyan> markwash: cool
20:39:22 <markwash> zhiyan: thanks so much for your help in that, btw!
20:39:33 <zhiyan> markwash: np at all
20:39:36 <markwash> I love how we seem to trade off timezones and the "Needs One More +2" queue :-)
20:40:04 <zhiyan> markwash: yes, actually that's my first page on the morning
20:40:09 <markwash> mine too :-)
20:40:14 <zhiyan> :)
20:40:22 <iccha> awwww :)
20:40:27 <iccha> sorry could not help it :p
20:40:30 <markwash> lol
20:40:35 <nikhil___> do you guys use some kinda of special query for that?
20:40:46 <nikhil___> sorry jenkins filters dont seem intuitive to me
20:40:50 <markwash> https://review.openstack.org/#/q/status:open+project:%255Eopenstack.*glance.*+branch:master+label:CodeReview%253D2+-label:CodeReview%253D-1+-+label:CodeReview%253D-2+-label:Approved%253D1,n,z
20:41:22 <nikhil___> ty markwash
20:41:37 <zhiyan> markwash: btw, could you pls share another worth quick-link with me/us? (maybe you can help do that on each meeting)
20:42:04 <markwash> zhiyan: I don't exactly follow, can you elaborate?
20:42:28 <zhiyan> markwash: oh, another link, just like this "Needs One More +2"
20:42:55 <zhiyan> markwash: which can help me/us accelerate review/landing change
20:43:05 <markwash> sure, what do you want this other link to point to?
20:43:52 <arnaud> is there anything reviewers (who are not core), can do to accelerate the review process?
20:44:11 <iccha> reviews from everyone are welcome
20:44:11 <markwash> well, we could promote some active non-core folks to core
20:44:13 <iccha> core or not
20:44:14 <arnaud> (maybe focus of reviews related to approved bp?)
20:44:16 <markwash> to help move things along?
20:44:29 <zhiyan> markwash: no particular idea, just want to know your experience
20:44:30 <markwash> I think -1s are always helpful when needed, also
20:44:36 <markwash> zhiyan: ah sure thing
20:46:01 <markwash> zhiyan: I can post all of the ones I use in an email if you like
20:46:12 <zhiyan> markwash: lzy.dev@gmail.com
20:46:13 <arnaud> there is something I noticed, is that sometime, reviewers give one review on a patch, and then when the code is updated or whatever, it is harder to get another review.. do you guys have the same feeling?
20:46:22 <markwash> yeah absolutely
20:46:26 <iccha> markwash: maybe share it in meeting etherpad?
20:46:30 <zhiyan> markwash: thanks!
20:46:31 <iccha> so all of us can have access to it
20:46:34 <markwash> that's part of the insanity of having such a long queue
20:47:00 <markwash> arnaud: if you tend to feed off the bottom of the queue, then the second review after a fix takes forever to loop back around
20:47:16 <markwash> I wonder if to respond to that we should try to top-feed off of the queue of reviewable items
20:47:32 <markwash> btw, link to reviewable changes:
20:47:33 <markwash> https://review.openstack.org/#/q/-label:CodeReview%253C%253D-1+-label:Verified%253C%253D-1+-label:Approved%253E%253D1++-status:workinprogress+-status:draft+-is:starred+-owner:mark.washenberger%2540markwash.net+project:%255Eopenstack.*glance.*,n,z
20:47:46 <markwash> probably want to put in your own name to that filter
20:49:01 <zhiyan> markwash: cool
20:49:02 <markwash> the problem with that link I just shared is that its hard to find out when someone simply disagrees with a -1 and doesn't push a new patchset
20:49:13 <markwash> because any -1 removes you from the queue
20:49:41 <markwash> okay, I wanna open up for discussion, but we can keep talking about reviews
20:49:53 <markwash> I don't have any action items for this yet, I just want people to know the scale of the problem
20:49:57 <markwash> #topic open discussion
20:50:08 <nikhil___> about the export script
20:50:37 <nikhil___> Have started the work on swift store side of things, Fei had mentioned about working on the filesystem one
20:50:55 <nikhil___> he was waiting on the executor to be pushed up
20:51:07 <arnaud> reviews out?
20:51:14 <iccha> This is an important change and more diverse reviews the better on this one: https://review.openstack.org/#/c/34801/
20:51:14 <nikhil___> I'm planning to work for next 3-5 days on export and then work on executor
20:51:26 <nikhil___> markwash: ^^
20:51:55 <markwash> nikhil___: okay great, daily or semi-daily report emails appreciated, they're great reminders for me
20:52:05 <markwash> don't have to be long or formal at all :-)
20:52:32 <markwash> iccha: does that patch do what you guys want it to at RS?
20:52:45 <markwash> I've been a bit nervous though I look at it in my queue every day
20:53:15 <nikhil___> markwash: sure will start them back on starting tomorrow. (Have just been distracted by either being sick or having my apartment getting flooded), Focus on reviews and MP would be back on next week onwards.
20:53:22 <markwash> oh my
20:53:36 <markwash> well I hope your health and apartment are returning to working order!
20:53:46 <iccha> markwash: it looked like the best temporary solution. it makes me nervous too. thats why not rushing it and would like more eyes on it
20:54:03 <nikhil___> thank you!
20:54:54 <markwash> iccha: okay cool I'll stop avoiding it then :-)
20:55:06 <arnaud> iccha: will look at it too
20:55:11 <markwash> Any other thoughts? or shall we close out? I can start my lunch ;-)
20:55:33 <nikhil___> markwash: just one comment
20:56:02 <nikhil___> I will try to have the tasks demo in the mini summit on RS cloud + swift
20:56:08 <markwash> oh awesome!
20:56:12 <markwash> thanks nikhil___
20:56:16 <nikhil___> not sure if I can have it done on filesystem +devstack though
20:56:19 <nikhil___> great
20:56:35 <markwash> that sounds really good
20:56:51 <markwash> okay, thanks everyone, have a nice rest of your week and weekend
20:56:55 <rosmaita> bye!
20:56:57 <arnaud> bye
20:56:57 <iccha> thanks markwash
20:57:00 <markwash> think about what you want out of the mini summit!
20:57:03 <zhiyan> bye!
20:57:03 <markwash> #endmeeting