20:04:23 <markwash> #startmeeting glance
20:04:24 <openstack> Meeting started Thu Jun 27 20:04:23 2013 UTC.  The chair is markwash. Information about MeetBot at http://wiki.debian.org/MeetBot.
20:04:25 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
20:04:27 <openstack> The meeting name has been set to 'glance'
20:04:32 <markwash> o/
20:04:34 <flaper87> \o/
20:04:36 <jbresnah> o/
20:04:45 <iccha> \o
20:04:52 <ameade> /o/
20:04:53 <ameade> \ \
20:04:57 <nikhil> :D
20:04:59 <zhiyan> \o/
20:05:09 <rosmaita> \o
20:05:27 <markwash> glance wave complete
20:05:40 <markwash> #topic open discussion
20:05:41 <markwash> :-)
20:05:59 <nikhil> haha nice
20:06:07 <zhiyan> Base on our discussion yesterday (my tz), I know markwash and jbresnah consider it's too much implicit automation about the "store location proxy" change did (in https://review.openstack.org/#/c/34501/1 ), so if you/folks have alternative idea around it, please feel free to add comments to gerrit there. thanks. I'd like get your thoughs about it first and discussion with you, then to get a consistent disign.
20:06:20 <jbresnah> at some point I would like to revist the quota issue
20:06:41 <markwash> zhiyan: absolutely, that is high on my list, it might be the weekend before I can try to offer up a sensible alternative
20:06:59 <zhiyan> markwash: great! thanks :)
20:07:00 <jbresnah> zhiyan: it may be easier to judge that patch when it is hooked into code that is adding and removing locations from the API
20:07:05 <markwash> zhiyan: and I'm thrilled to have what you have now, it will work, its only a question of if we want to make a little change
20:07:20 <jbresnah> zhiyan: as it is now i think it is very clever but i am concerned that it would be too implicit when used
20:07:36 <jbresnah> zhiyan: if i saw it in use it would be an easier call to make
20:07:41 <zhiyan> yes, yes
20:08:00 <markwash> zhiyan: did you find the etherpad with how patch ought to work for locations to make sense?
20:08:10 * flaper87 will catch up with reviews tomorrow
20:08:34 <zhiyan> marwash: yes, i will take the details in this morning. after reveiw jbresnah new change on multiple-locations-metadata :)
20:08:58 <zhiyan> jbresnah: yes, so please add your comments to my change in review.
20:08:59 <jbresnah> speaking of reviews, i am on vacation again all next week and it would be great to get comments on this patch before then: https://review.openstack.org/#/c/34492/
20:09:19 <ameade> jbresnah: you live on vacation
20:09:21 <flaper87> jbresnah: I promisse I'll take a look at it
20:09:28 <zhiyan> jbresnah: sure, take a good reset
20:09:29 <jbresnah> zhiyan: will do
20:09:37 <markwash> jbresnah: cool, thanks, just bumped it up the priority list
20:09:49 <markwash> flaper87: how's registry db driver stuff going?
20:09:53 <zhiyan> markwash: thanks
20:10:13 <flaper87> markwash: so, it's doing well. I already submitted a patch for the json stuff and saw your +2
20:10:24 <flaper87> markwash: wrt the untested call, that is tested in the wsgi tests
20:10:39 <flaper87> that's why I didn't think it was worth it to run that again
20:10:43 <flaper87> does that make sense?
20:10:51 <markwash> flaper87: sure, cool
20:11:02 <markwash> I've been running tox -- --with-cover lately and loving it for reviewing tests
20:11:25 <flaper87> markwash: besides that, base db tests run perfectly
20:11:29 <flaper87> oh wait, just a thing
20:11:38 <flaper87> let me get a link for y'all
20:11:40 <markwash> flaper87: so you just need more reviews?
20:11:48 <jbresnah> markwash: that dovetails into another topic i have: unit test coverage
20:12:04 <ameade> +1
20:12:07 <jbresnah> markwash: can we make it policy that any change must have code coverage in unit tests?
20:12:08 <flaper87> markwash: I need another one for that json stuff and then I'll push the big one with the db driver implemented
20:12:14 <markwash> jbresnah: ++, maybe we should break away from #open for a sec to talk about tests
20:12:19 <jbresnah> it would be amazing if gerrit would do that for us...
20:12:38 <flaper87> jbresnah: I think we could request a no-voting gate for that
20:12:51 <markwash> I would love that as well, I've been giving it some thought
20:12:57 <flaper87> jbresnah: just drop an email in the -dev list
20:13:02 <jbresnah> flaper87: maybe that would be too harsh...
20:13:16 <markwash> I want to detect the *new* tests in a change, and the *new* code, and have a specific report that compares the new coverage with the new code
20:13:19 <jbresnah> perhaps just a apolicy of  'no coverage in unit tests is grounds for rejection?'
20:13:28 <jbresnah> i can imagine cases where it would be ok
20:13:46 <jbresnah> markwash: yeah that would be amazing
20:13:50 <markwash> its definitely grounds for rejection, I think that's already true
20:13:56 <markwash> I guess at times we make exceptions
20:14:03 <ameade> yeah we just need to make sure reviewers aren't accepting something just because it has functional tests but still no unit tests
20:14:04 <jbresnah> markwash: even if it is in a functional test?
20:14:05 <flaper87> https://github.com/openstack/glance/blob/master/glance/db/sqlalchemy/api.py#L827
20:14:18 <flaper87> I've got an issue with this call (and related ones)
20:14:29 <jbresnah> ameade: right i did that today actually,  but in my defense i am about to submit a unit test for the change in question
20:14:45 <flaper87> thing is that it receives a property_ref wich is a property instance
20:15:09 <flaper87> that property instance is serialized in the registry driver and sent to the registry service
20:15:13 <ameade> jbresnah: i'm not calling you out :P
20:15:25 <markwash> I don't think we need to be extremely aggressive about requiring tests, but I think we should be evaluating coverage as part of review, and make sure that if anything doesn't have coverage, we think its *okay*
20:15:41 <iccha> +1 that
20:15:48 <flaper87> The registry service doesn't know how to de-serialize it and fails misserably.
20:15:56 <markwash> and as we switch more and more towards the approach in https://etherpad.openstack.org/glance-improving-test-cycle-times I hope we can find easy ways to require coverage almost all the time
20:15:57 <ameade> yeah it's all a judgement call (otherwise we should automate it)
20:16:22 <jbresnah> +1
20:17:01 <markwash> flaper87: yeah that looks like it might just be a bug. . does it make sense as a bug?
20:17:21 <flaper87> I could write a deserializer for models but that's not right since other calls just receive an id
20:17:30 <flaper87> markwash: it does make sense
20:17:45 <flaper87> markwash: I wanted to get everybody's aggrement on changing that
20:17:59 <flaper87> messing with the db driver is not fun, nor secure
20:18:36 <markwash> I don't think public functions in db api should depend on passed values being driver-specific entities (like sqlalchemy models)
20:18:49 <flaper87> markwash: +1
20:18:52 <markwash> sometimes that hurts performance, and we have to figure out a (usually kind of awkward) workaround
20:19:03 <flaper87> agreed, then, I'll change that
20:19:30 <flaper87> jbresnah: I don't think that's to harsh. Coverage should be enforced in every project, IMHO
20:20:03 <flaper87> and coverage doesn't mean it is well tested, it just means it is all tested.
20:20:13 <markwash> anybody had a chance to take a look at https://review.openstack.org/#/c/30512/ ?
20:20:15 <flaper87> don't get me wrong, I like the code to be covered
20:20:41 <zhiyan> yes, i saw it markwas
20:20:46 * markwash is really excited about moving away from forking functional tests, because those make coverage metrics very hard to generate
20:21:00 <jbresnah> markwash: and make debugging much harder
20:21:23 <jbresnah> markwash: i will give that another review today
20:21:28 <iccha> will put in on my list to checkout markwash , have not seen the latest patch
20:21:31 * markwash might resort to selling indulgences to get his patchs +2'd :-)
20:21:36 <jbresnah> heh
20:21:43 <jbresnah> quid pro quo?
20:22:00 <markwash> ;-)
20:22:08 <jbresnah> if i +2 all of your patches can we avoid that 4am glance meeting?
20:22:13 <markwash> haha
20:22:23 <zhiyan> flper87: and i have a question that seems i have no way to run a single test case from glance...
20:22:40 <flaper87> ?
20:22:56 <zhiyan> flaper87: sorry, just you:)
20:23:06 <markwash> hmm, I've had luck with that, using tox -e py27 -- --tests glance/tests/unit/path/to/test/file.py:TestClass.test_method
20:23:24 <flaper87> that works for me as well
20:23:48 <zhiyan> markwash: cool, i will use that in next test runing for dev
20:24:18 <markwash> I am a bit disorganized today (and maybe always)
20:24:26 <markwash> does anyone have status updates they want to share for ongoing blueprints?
20:24:38 <markwash> beyond what we've talked about already. . .
20:24:53 <zhiyan> yes, folks, markwash and jbresnah, the BP about image multiple loations support on nova side (https://blueprints.launchpad.net/nova/+spec/image-multiple-location) had approved by Russell Bryant, and the change (https://review.openstack.org/#/c/33409 ) is beta available for you review. (thanks markwash given some valuable comments there and I have responsed :) ), so if you have time please give a review. I just want to parallel glance and nova
20:25:06 <jbresnah> with the review i mentioned i get pretty close to multiple-locations being complete
20:25:23 <rosmaita> here's a new try for protected props: https://wiki.openstack.org/wiki/Glance-property-protections
20:25:35 <rosmaita> don't think stuart has had time to look at it yet, though
20:25:50 <markwash> jbresnah: are you okay with us taking over a bit in your absence, if we're not quite ready to merge before your vacation?
20:26:02 <zhiyan> and iccha: i have updated https://etherpad.openstack.org/remove-sensitive-location-info-glance
20:26:30 <jbresnah> markwash: yeah
20:26:37 <jbresnah> markwash: definitely on the API side
20:26:47 <markwash> jbresnah: cool, thanks!
20:27:04 <jbresnah> markwash: i have a bit more interest in the metadata parts
20:27:34 <markwash> jbresnah: nod. . makes sense
20:27:54 <markwash> rosmaita: looks good to me
20:27:59 <jbresnah> but i think my current patch is pretty close
20:28:14 <markwash> jbresnah: it sounded like it was. . I just haven't had the time to look yet
20:28:28 <iccha> zhiyan: thanks for your comments, i would rather subtiture user and password with dummy values like 'hidden_user' and 'hidden_key', because even if it is encrypted it still is storing some form of the info
20:28:33 <zhiyan> jbresnah: sorry, i don't think markwash think so...https://etherpad.openstack.org/glance-manipulating-multiple-locations
20:29:31 <iccha> zhiyan: i am working on swift store for the same, and anyone else is welcome to pitch in for other stores :)
20:30:04 <markwash> jbresnah: zhiyan: sorry I just need to look again
20:30:38 <zhiyan> iccha: cool, i will take look your patch when it ready, and sync though with you...to make sure we are in the same position
20:31:03 <iccha> thanks zhiyan :)
20:31:06 <flaper87> iccha: I'll take care of GridFS
20:31:13 <zhiyan> markwash: sure, but cloud you pls update the doc when you ok, i mean here https://etherpad.openstack.org/glance-manipulating-multiple-locations
20:31:26 <zhiyan> iccha: wel, enjoy it
20:31:33 <markwash> zhiyan: absolutely
20:31:43 <markwash> zhiyan: is there a specific issue in contention?
20:32:40 <zhiyan> markwash: not yet, but seems your thoughts are not the same with jbresnah current did
20:33:26 <jbresnah> zhiyan: what i did in which patch?
20:33:39 <zhiyan> 'locations' content within 'GET /images' result will different...
20:34:28 <jbresnah> zhiyan: yeah that patch  document is not addressing the metadata
20:34:31 <zhiyan> https://review.openstack.org/#/c/31591/
20:34:36 <markwash> okay, I see so I just need to make sure that I'm being consistent, and make sure that the guidance in that etherpad and on the reviews is the same
20:34:49 <markwash> and that way I can stop being a blocker :-)
20:34:50 <zhiyan> jbresnah: yes, but not the key
20:35:45 <zhiyan> markwash: yes, the key i think is make a final design about PATCH interface for locations updating
20:35:50 <markwash> cool
20:35:56 * markwash will review and comment :-)
20:36:03 <jbresnah> sounds good
20:36:20 <zhiyan> i just don't think your thoughts is the same with jbresnah's
20:36:25 <markwash> right
20:36:35 <jbresnah> if it happens when i am gone note that i am very flexable about what the API looks like
20:36:45 <jbresnah> i just want metadata to flow back with a location
20:36:50 <jbresnah> and perhaps in the future flow in
20:36:50 <markwash> +1
20:36:52 <jbresnah> but that can wait
20:37:44 <jbresnah> so if we are looking for a topic, can i address quotas a bit?
20:37:48 <markwash> sure
20:38:13 <jbresnah> markwash: you recall at the summit that keystone people were going to add some functionality to support quotas?
20:38:24 <jbresnah> markwash: do you know if anything has happened there?
20:38:40 <markwash> yes; no I haven't heard anything about progress yet
20:39:25 <jbresnah> ok, at one point we thought it would be ok to add a basic config for max allowed storage in bytes to glance
20:39:31 <jbresnah> ...or at least i thought that was mentioned
20:39:34 <jbresnah> any thoughts on that?
20:40:31 <flaper87> I haven't put many thoughts in quotas (besides our early discussions about the implementation)
20:40:38 <markwash> I'm not sure its the best fit, but I think it could probably be okay
20:40:51 <flaper87> Have we thought a bit more about how it should be implemented?
20:40:57 <jbresnah> ok, i may put some thoughts in code then
20:41:05 <markwash> it seems like storage quotas could be both broader and narrower than a glance store
20:41:05 <flaper87> sounds good
20:41:13 <jbresnah> nod markwash
20:41:24 <markwash> broader, becuase you might view your e.g. fs store as a shared resource across glance, nova, etc
20:41:35 <jbresnah> i am thinking this would be just 1 way to do it that could make sense for simple (which is probably most) deployments
20:41:45 <markwash> narrower, because you might view your fs store and your swift store as two completely independent storage systems, with different quotas
20:42:16 <jbresnah> nod, i agree
20:42:24 <markwash> cool, well, I'm interested to see some code
20:42:29 <flaper87> +1
20:42:39 <jbresnah> ultimately i think it is clear that quotas need to be cross cutting
20:42:50 <markwash> I'm sure lots of folks would like a 70% solution
20:43:00 <jbresnah> right ok cool
20:43:11 <markwash> anybody want to do another shared review day? maybe sometime next week?
20:43:29 <markwash> I bring this up and then never schedule it. . .
20:43:29 <jbresnah> i would love to do one, but i am out next week
20:43:31 <flaper87> markwash: I won't be around next week. What about the one after next ?
20:43:47 <ameade> markwash: I'm down for getting my head into some reviews
20:43:56 <markwash> hmm, that could be better than. . can other folks confirm availability week after next?
20:44:10 <ameade> +1
20:44:11 <markwash> . . . / would monday work?
20:44:14 <flaper87> +1
20:44:23 <flaper87> vote ?
20:44:25 <flaper87> :D
20:44:30 <rosmaita> you mean july 1?
20:44:37 <flaper87> we never use thouse great features
20:44:40 <flaper87> rosmaita: 8th
20:44:42 <markwash> I think I mean the 8th
20:44:47 <rosmaita> ok
20:44:47 <jbresnah> 7/8 works for me
20:45:02 <esheffield> I'm a bit late to the party, but I'm out July 4 - 14
20:45:05 <iccha> works for me
20:45:07 <esheffield> don't let that stop you tho
20:45:28 <ameade> +1 for 8th
20:45:38 <markwash> I'm hearing mostly okay, so I'll add that date to my list
20:45:40 <markwash> and send out reminder
20:45:41 <markwash> s
20:46:02 <markwash> I'd love to close it off 15 minutes early today, if that's okay with everyone. . .
20:46:14 <rosmaita> got a quick question
20:46:17 <markwash> go for it
20:46:23 <flaper87> +1 not much to say. I'm catching up with many stuff
20:46:35 <zhiyan> kindly reminder again, pls review https://review.openstack.org/#/c/33409 if you ok, thanks :)
20:46:40 <rosmaita> was updating some docs, wrote up a page about where the glance docs are source and posted:
20:46:41 * ameade thinks we should sit here in silence for 15 min
20:46:47 <rosmaita> https://wiki.openstack.org/wiki/Glance-where-are-the-docs
20:46:52 <markwash> rosmaita: timely!
20:46:58 <markwash> ameade: lol
20:47:00 <rosmaita> anyway, i was adding stuff about the notifications
20:47:10 <rosmaita> someone put a comment that i should include examples
20:47:16 <rosmaita> but i am lazy
20:47:28 <ameade> +1
20:47:28 <rosmaita> and also don't think they belong in the "developer" docs anyway
20:47:34 <jbresnah> if we end with time an update on async workers would be nice
20:47:35 <iccha> this is awesome rosmaita
20:47:38 <nikhil> so much xml support
20:47:46 <rosmaita> and markdown
20:47:58 <zhiyan> rosmaita: really coool
20:48:00 <nikhil> jbresnah: m working on a use case for it
20:48:14 <rosmaita> anyway, was wondering what you think about examples in the dev docs
20:48:30 <rosmaita> was thinking they'd be better in the operator docs
20:48:38 <rosmaita> since they'd be consuming the notifications anyway
20:48:39 <nikhil> markwash: jbresnah stuck on figuring the design logic for transfer of image_data (in the import case)
20:48:45 <markwash> rosmaita: for notifications, yeah that makes sense
20:49:05 <markwash> *putting notification examples in operator docs
20:49:07 <jbresnah> nikhil: I would be happy to talk that out with you sometime if you think it would help
20:49:15 <rosmaita> except, of course there's not an obvious place for notifications in theoperator docs right now!
20:49:26 <nikhil> jbresnah: sure, lemme try to put up a patch tomorrow
20:49:33 <flaper87> nikhil: +1
20:49:36 <nikhil> and we can sync up on glance channel sometime?
20:49:40 <markwash> I'd like to use your "where are the docs" page as a launching point for re-evaluating the general flow of glance docs, and hopefully directing an effort towards improvement later in havana
20:49:40 <nikhil> flaper87: cool
20:49:50 <rosmaita> works for me
20:49:50 <jbresnah> nikhil: sounds good
20:49:52 <flaper87> nikhil: make sure to pull me into the loop
20:49:58 <nikhil> flaper87: sure
20:50:36 <markwash> rosmaita: so for now, we could just add notifications docs to the bug list
20:50:44 <rosmaita> ok
20:50:56 <rosmaita> then can someone +2 https://review.openstack.org/#/c/33985/
20:51:31 <rosmaita> markwash: i'll add a bug and assign it to myself
20:51:36 <markwash> am I allowed to nitpick about copyright headers? :-)
20:51:43 <rosmaita> sure
20:51:49 <rosmaita> what's wrong?
20:51:55 <iccha> 2011-13 :p
20:52:19 <nikhil> markwash: no no no, you can't beat me to it this time :P
20:52:24 <markwash> lol
20:52:48 <markwash> I like how now I can pretend that I *did not* nitpick about copyright headers :-)
20:53:01 <flaper87> lol
20:53:03 <rosmaita> so is that really it?
20:53:15 <markwash> pretty much
20:53:20 <markwash> we don't need a slash
20:53:38 <rosmaita> so what should it be?
20:53:49 <markwash> its either an effectively new document (2013) or you can forgo mentioning the copyright for the recent additions by leaving it as is
20:54:10 <rosmaita> what's the consensus here?
20:54:31 <rosmaita> since i have to change it anyway
20:54:35 <markwash> the legal consensus is pretty much that it doesn't matter, but anne did put together something that said "don't bother with hyphenated dates"
20:54:53 <rosmaita> OK, missed that.  What's the glance dev consensus?
20:55:16 <ameade> booo copyrights
20:55:22 <markwash> ameade: +1 :-)
20:55:35 <markwash> I'm not willing to -1 for copyrights anymore I guess
20:55:40 <markwash> I already +2'd the change
20:55:54 <markwash> I'd slightly prefer leaving it alone
20:56:12 <markwash> but I mostly prefer trying for myself to not be a troll about copyrights
20:56:41 <nikhil> markwash: are planning on auto-generating docs like nova sometime?
20:56:51 <markwash> nikhil: oh, I don't know much about that
20:56:58 * nikhil feels that a wave of disagreement is coming his way
20:57:13 <zhiyan> jbresnah: i'm not sure your position about https://review.openstack.org/#/c/31306/ , seems your PS2-4 are all the same...so can you clear it here, maybe markwash can remove -1 there..
20:57:13 <nikhil> ah ok
20:57:47 * flaper87 read PS2-4 as Playstation 2 to Playstation 4
20:57:55 <zhiyan> :)
20:58:00 <nikhil> ya me too
20:58:05 <zhiyan> patch set 2-4
20:58:15 <iccha> = patch set -2
20:58:22 <jbresnah> heh
20:58:23 <nikhil> :P
20:58:28 <flaper87> iccha: LOL +1 for that
20:58:32 <markwash> now you all are just being silly :-)
20:58:41 <flaper87> markwash: we're trying to keep you around
20:58:51 <markwash> apparently its working
20:58:53 <nikhil> yeah
20:58:54 <flaper87> :D
20:58:58 <jbresnah> zhiyan: i did a rebase
20:59:04 <markwash> #endmeeting