21:04:06 #startmeeting reddwarf 21:04:07 Meeting started Tue May 21 21:04:06 2013 UTC. The chair is hub_cap. Information about MeetBot at http://wiki.debian.org/MeetBot. 21:04:08 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 21:04:08 hai 21:04:10 \ o 21:04:11 The meeting name has been set to 'reddwarf' 21:04:13 supposedly the bot that takes attendance only counts \o 21:04:16 lol juice i like datsun180b's the best o7 21:04:17 hola 21:04:19 \o 21:04:20 \o 21:04:29 o7 is a salute of course 21:04:32 ha a salute 21:04:34 awesome 21:04:37 o/ 21:04:40 #link https://wiki.openstack.org/wiki/Meetings/RedDwarfMeeting 21:04:44 I don't salute you hub_cap 21:04:48 #link http://eavesdrop.openstack.org/meetings/reddwarf/2013/reddwarf.2013-05-14-20.37.html 21:04:51 imsplitbit: sure u do 21:04:57 feisty crowd today 21:05:09 7o 21:05:11 heh 21:05:32 #topic Action Items 21:05:33 \o/ 21:05:49 SlickNik: archiving logs for jenkins 21:05:51 * esmute is yawning 21:05:51 So first one's mine 21:06:09 Haven't had a chance to look into that one yet. 21:06:31 And probably won't for another week. :( 21:06:51 k. is there someone else who u can give it to? 21:06:57 cuz its super frustrating 21:07:02 #agreed 21:07:23 Anyone want to volunteer? 21:07:33 (hp folks?) 21:07:52 I would but I have enough on my plate 21:07:54 * esmute steps forward 21:08:02 I'll see what I can do to get something moving on that front. 21:08:03 so they would have to scp the logs back to jenkins and then archive them eh? 21:08:16 just tell me what to do and ill be a good minion 21:08:20 Why not compress first and then offload, save some bandwidth 21:08:24 heh esmute nice 21:08:28 datsun180b: well ya fo sure 21:09:04 #action esmute work with SlickNik to figure out the archiving of the reddwarf logs for rdjenkins jobs. 21:09:10 SlickNik: u have the 2nd one too but i dont exactly understand it heh 21:09:19 Ah, I looked into that. 21:09:30 okey sup 21:09:35 And had a short conversation with mordred regarding it. 21:09:50 Basically the idea is to move to a state where we don't have rdjenkins. 21:10:02 And have openstack CI running our integration tests for us as well. 21:10:10 cool. i think that state also involves a lot of other states 21:10:17 its like a whole country 21:10:23 yup. :) 21:10:31 The first step is to have this patch land. 21:10:41 #link https://review.openstack.org/#/c/23999/ 21:10:44 we need to line up what is devstack and what is redstack.. there is a lot of "stuff" in redstack 21:11:44 id prefer to not have it call our stuff eventually too 21:11:53 but thats a whole nutha can-O-worms 21:12:00 basically mordred is putting in hooks from devstack-vm-gate that we can use to run our stuff from redstack. 21:12:10 ya 21:12:31 i figured thatd be the case, but thats a whole conversation id like to have eventualy about why we have all the setup outside of devstack 21:12:38 My thinking was that we start of this way and then make smaller patches to get our stuff integrated. 21:12:40 we do a lot of things like add users, for example 21:12:42 sure def 21:12:47 Yup me too. 21:12:53 adding users for tests, frankly, should be in tests that need those users 21:13:01 and the whole apt repo thing, that needs to go away too 21:13:05 heat will help w/ that 21:13:10 heat integration, that is... 21:13:23 anyhoo, rabbit hole quickly falling down, #movingon 21:13:26 agreed. 21:13:27 grapex: your next 21:13:43 did u talk to tempest guys to see if there is a resize issue? 21:13:50 hub_cap: I cancelled this before the end of last meeting 21:13:57 as the resize time was said to not be an issue 21:13:57 Oh okey hehe 21:14:03 word 21:14:10 i think the issue mightve been caused by me 21:14:14 i had a similar issue 21:14:21 cuz i was trying to use more memory than was avail on the host 21:14:29 but that was manual resizes 21:14:33 so no telling if that was the same case 21:14:45 We made that action item because there were rumors the resize time was killing rdjenkins 21:14:58 ok so moving on 21:15:00 ya? 21:15:02 and then before the end of the meeting everyone agreed that wasn't really the problem 21:15:06 that's all. :) 21:15:09 yeah, sounds good. 21:15:10 cool 21:15:12 #topic Voting rules and regulations 21:15:28 so im going ot do a poll just like the polls the openstack community uses 21:15:37 the Pick Three, assign weight, voting 21:15:51 #link http://www.cs.cornell.edu/w8/~andru/civs/ 21:15:52 I like it 21:16:00 i'm for it 21:16:07 hub_cap: can we do a practice run first? 21:16:18 ++ 21:16:21 define practice run 21:16:25 to instant runoff 21:16:26 VaaS nice 21:16:34 hub_cap: just messing with you. lets vote. 21:16:44 DORK esp1 ;) 21:16:50 #link https://gist.github.com/hub-cap/5622714 21:16:59 list of names so far 21:17:01 #link https://gist.github.com/hub-cap/5622594 21:17:04 list of addresses to send to 21:17:10 only problem is 21:17:17 It's done via email isn't it? 21:17:24 yes datsun180b 21:17:31 I feel sorry for your inbox 21:17:37 vipul is not around to vote. we can wait till next friday to vote (not this friday, but th 31st) 21:17:45 but if the group says no, then we will do now 21:18:06 well let's see where we get and if it's close he can break the tie 21:18:06 how critical is the name change? 21:18:07 the vote results will be submitted to the openstack foundation for vetting 21:18:12 juice: ok 21:18:19 #link: http://en.wikipedia.org/wiki/Seshat 21:18:27 good idea juice 21:18:36 imsplitbit: we cant move to openstack github group till we change it 21:18:52 so critical 21:18:55 let's do it 21:18:57 esp1: that name just has some um... soiling implications 21:19:13 ya ps the TC guys didnt like rover ;) 21:19:15 lol@imsplitbit 21:19:29 it's "Say-shot" you goon 21:19:33 imsplitbit: haha, yeah I was kinda wondering what the long term implications might be. 21:19:52 Ask ceilometer how they deal with it 21:19:52 datsun180b: sry man, its hard to say... imagine all the people mispronouncing it 21:19:57 datsun180b: sure, but I see it and think "SHE DID WHAT????" 21:20:10 :-) 21:20:13 datsun180b: they dont, the people saying it feel dumb after they hear it correctly 21:20:16 hub_cap - you need our emails? 21:20:24 juice: SlickNik is hookin me up 21:20:27 note to self: deliberately pronounce every ambiguous word incorrectly 21:20:46 we can run the meeting while you guys setup the voting stuff 21:20:50 what's next? 21:20:55 heh 21:21:15 #topic New Meeting Time 21:21:20 oooo 21:21:22 this one should also be fun 21:21:27 I thought we were gonna do 33o 21:21:31 +1 datsun180b 21:21:31 330 21:21:33 cst 21:21:35 that doesnt help me 21:21:43 so selfish 21:21:49 :) 21:21:50 i have a 4cst and a 5cst meeting for the ptls 21:22:04 juice: ya reddwarf doesnt need any representation in the TC meeting ;) 21:22:05 how about early after our standup 21:22:07 sounds like they have too many ptl meetings 21:22:15 11:00 pst/1cst 21:22:34 I vote 830 cst 21:22:37 :-) 21:22:42 how about a different day ?? 21:22:43 ha 21:22:43 imsplitbit: your vote doesnt count 21:22:47 lol 21:22:49 OUR guys cant get in at 830 cst 21:22:56 lol 21:23:00 kagan: yes plz 21:23:04 theres a couple that are in at that time 21:23:19 any day but friday and monday 21:23:22 I don't really care about the time so long as it's not past 4 cst 21:23:28 so ill do this. there is a tool for picking days/times, like a voting tool 21:23:35 Whatever's decided, I don't want anything on monday before noon central or anything on friday after 3 central 21:23:39 although monday might not be a bad idea 21:23:40 ill organize that and send it out 21:23:45 set the stage for the week 21:24:03 juice: I'm a big fan of knocking all meetings out on mondays 21:24:10 monday is a bad day, as there are usually holidays 21:24:13 lets you have the rest of your week to get real work done 21:24:19 good point 21:24:27 yeah - I think you might be looking at Wednesdays if you change the day 21:24:29 how about same time on Wed? 21:24:31 ya and monday/friday are when people take "extra" days off 21:24:41 i vote to _not_ have it at 2/4 21:24:43 wednesday then 21:24:55 its not terribly fair to the CST'ers 21:25:02 hub_cap: is there an openstack specific tool? 21:25:06 SlickNik: nope 21:25:10 We're two hours apart, right? 21:25:14 but there is a voting-esque tool 21:25:16 yep 21:25:18 datsun180b: always 21:25:36 http://doodle.com/ is pretty good at helping with picking a time. 21:25:38 no, hub_cap, not on mondays ... 21:25:40 then 11 pst/13cst it is 21:25:41 What if we figure for Wednesday, 11:00/1:00 21:25:54 I'm with datsun180b 21:25:55 next 21:25:56 datsun180b: +1 21:25:56 so it's on either end of lunch for both sides 21:26:10 hub_cap: if you do wed, would that help with giving us feedback from the tc mtgs? 21:26:22 kagan: i said thats when people take days off... i didnt say yes monday ;) 21:26:29 right in the middle of the day in the middle of the week 21:26:33 amyt: ya def 21:26:57 i meant regarding not being 2 hours apart on Mondays ... 21:27:06 how about earlier on wed to give people more time to "catch up" before the week ends? 21:27:11 so tahts wed 22 utc? 21:27:16 datsun180b: ++ 21:27:32 nm thats not 2200 21:27:33 let me see 21:27:41 and both sides are motivated to keep the meeting from clobbering lunch, and keeping lunch from clobbering the meeting 21:27:49 amyt: i don't think that would work out great 21:28:08 juice: yea i forgot about the lunch thing. i GUESS we have to eat 21:28:10 I think 11:00 is the most practical time from an hp perspective 21:28:10 1900 21:28:25 swift is at 1900 21:28:32 but thats ok 21:28:37 is that.. 11 am ct? 21:28:37 and we'll all be here at 1:00 our time 21:28:38 1900! yowzers 21:28:55 amyt: 11am Pacific, 1pm Central 21:28:57 amyt that is 13:00 cst (1 pm) 21:29:03 swift is 1900utc on wednesdays (every other wednesday, currently) 21:29:09 what datsun180b said 21:29:11 gotcha, thanks guys :) 21:29:17 hub_cap: with PDT, 11PDT would be 1800UTC 21:29:32 ok crap pdt heh 21:29:33 to be fair juice suggested it moments before i did 21:29:37 so we do understand 21:29:43 that when we go off dst 21:29:48 our meeting changes by 1 hr 21:30:10 thats during moniker 21:30:13 can we just change it then? 21:30:34 not if we have other teams working w/ us 21:30:40 none of the other projects change 21:30:52 its somewhat disruptive to people in otehr countries persay 21:31:16 so it goes to 12pst/2cst 21:31:18 so if we move off dst: we become 10PST / 12CST 21:31:19 wuh oh, grapex might have a collision with 1pm 21:31:25 doh i suck 21:31:33 datsun180b: That's ok 21:31:38 just looking out for you 21:31:53 datsun180b: Thanks man 21:31:57 sound good then? 21:31:57 but if that time works I'm ok with it 21:32:00 1800 utc? 21:32:15 what is that in Earth time 21:32:20 we can always try to move it later grapex 21:32:25 hub_cap: Speak 'Merican 21:32:41 We've got daily stand up at 10 in the morning here. 21:32:42 I don't want to have to coordinate this time with the whole universe 21:32:53 hub_cap: 2:00 Wednesday would be the best. But it would cut into lunch I imagine 21:32:57 1:00 is fine if that works. 21:33:04 grapex: but it wont always be that 21:33:14 itll go from 2pm to 1pm because of dst 21:33:43 hub_cap: Argh 21:33:45 maybe we can take this offline 21:33:45 we change with dst 21:33:54 juice: prolly a good idea 21:33:59 yeah, I think we need to take this offline. 21:34:05 someone work out the kinks and get back to us with a couple of options 21:34:07 hub_cap: Ok, that changes things. :( 21:34:14 well that won't be for months, which i think should be enough time to reevaluate the meeting time 21:34:17 lets get together during the meeting time weve designated tomorrow 21:34:19 how about we do our standup at 11 that day ? 21:34:20 and discuss 21:34:20 :) 21:34:29 damn the pseudoephedrine is kicking in 21:34:38 lol@hub_cap 21:34:55 moving on 21:35:06 hang in there, juice 21:35:10 #topic Backups 21:35:18 done! 21:35:22 nice 21:35:23 moving on 21:35:25 * juice cheers 21:35:29 cheers! 21:35:36 SlickNik: are they merged? 21:35:40 halleluya 21:35:43 you did it wrong kagan ;) 21:35:53 * hub_cap laughs at juice 21:35:57 meaning second ? 21:36:03 /me cheers 21:36:03 yes, they are. Thanks for all the good work everyone! 21:36:07 sweet perfect 21:36:11 #topic Notifications 21:36:22 That's merged too. 21:36:30 * robertmyers cheers 21:36:33 well we still need the hourly right? 21:36:37 exists 21:36:42 err *daily* 21:36:42 well not completely done on my end 21:36:46 Juice was working on that one 21:36:50 right 21:36:52 juice even* 21:36:56 there's a bunch of crap they want but we are pushing back on it 21:37:16 they want some stuff out of nova that I just don't think we can do now 21:37:33 other than that I think it's oke 21:37:50 seems kind of silly to duplicate all of nova exist events 21:38:02 since they can just use them 21:38:03 robertmyers: agree 21:38:10 robertmyers: ++ 21:38:22 they should be aggregating this stuff on instance id's et 21:38:26 tc 21:38:34 well exists nova != exists reddwarf tho right? 21:38:40 what if mysql has been offline for 24hrs 21:38:47 hub_cap: that is true. 21:38:53 true but they have the details if the instance is up 21:38:53 but compute has been hummin that entire time 21:39:12 sure the nova compute instance is up 21:39:19 + there are periods of time where instance is up and not usable. 21:39:20 but to me that doenst mean jack 21:39:42 not usable = mysql is not up for whatever reason. 21:39:46 right 21:39:49 what I was saying is that the can create a composite set of info IF we tell them mysql is also up. We don't need to repeat the base nova info 21:39:54 and does that "exist" if mysql is not up? 21:40:11 we don't send exists if mysql is down 21:40:17 we only send exists if mysql is up :) 21:40:40 ok i guess im a bit confused... yall can work it out 21:40:46 :P 21:41:00 i thought u said we werent gonna send exists cuz nova does already but that sounds like its not the case 21:41:43 no there was just some extra info that nova has that we don't 21:41:48 AHH ic 21:41:49 cool 21:41:52 so my argument was why do we also have to send it 21:41:56 hub_cap: we need exist events because of nova != reddwarf. But we don't want to send duplicate nova info in our events. 21:41:59 maybe we just need a hook to optionally run exist evetns 21:42:13 robertmyers: that is the plan. 21:42:14 it will be a conf flag so you can turn it off 21:42:30 we are using it so that if multiple ™ are running only one sends it 21:42:38 ™ = taskmanager 21:43:04 ic what u all mean now. perfect 21:43:30 okie 21:43:38 good that's all cleared 21:43:48 * juice sips his coffee 21:43:51 juice: yup great 21:43:56 moving on 21:44:05 what is API Validation? 21:44:10 do we need to cover it? 21:44:11 that one is me 21:44:14 ok 21:44:19 I bet it's about the username/database name evaluation 21:44:20 #topic API Validation 21:44:24 I do need to ask you guys a few qestionss 21:44:32 shoot 21:44:51 go for it 21:44:55 so our sec folks say we need to clamp down on the data coming it at all levels but are ok with focusing on api for now 21:45:16 so I was looking at schemes for validating field length, type, etc 21:45:30 relevant to https://review.openstack.org/#/c/28850/ or a more general approach? 21:45:39 I found the validator in nova that seems like it would do that trick and we can just plug it into paste 21:45:53 juice: Is this the idea of validating everything according to a JSON or XML schema? 21:46:09 good luck validating json to a schema 21:46:16 datsun180b - they want us to do it across all requests on on fields 21:46:20 juice: we shodul adopt http://json-schema.org/example2.html 21:46:27 datsun180b: I think a more general approach across the API - not just database name 21:46:36 grapex: i was thinking that but have since let that go 21:46:46 Probably a good idea to attack it broadly instead of piecemeal 21:46:49 How's the Nova validator work? 21:46:54 juice: ^^ 21:47:20 grapex: they have a validator class that you specify the rules. then they run the request from the context through that 21:47:29 so you are specifying the schema in python 21:47:42 json schema! 21:47:43 this is consistent with another approach I found which is called schematics 21:47:50 no not json 21:47:54 juice: Is this schema some custom thing they made up? 21:48:01 Does the same schema work for JSON and XML? 21:48:04 #link http://json-schema.org/ 21:48:05 I believe so 21:48:19 it is agnostic since the response has been parsed by that point 21:48:19 its using that special xml definition 21:48:35 so I think it something like a dict or kwargs or something 21:48:48 let me send you a link to the class 21:48:54 Here's another thing to consider... it's selfish, but we use Repose. I feel like if used generic things like JSON and XML schemas, we could use that with Repose. People creating client libraries could also reuse them. 21:49:10 Yes, juice a link would be helpful. thanks! 21:49:17 we cant say that grapex :( 21:49:22 repose is _not_ openstack compatible 21:49:38 hub_cap: Yes, but json and xml schemas are open source compatable. ;) 21:49:48 for sure 21:50:17 ok so do we have any resolution on this? 21:50:28 #link https://github.com/openstack/nova/blob/master/nova/api/ec2/__init__.py 21:50:38 I mean I'll give this thing in Nova a fair shake, but if its 100% custom its worth thinking about if it's a worthwhile approach. If the idea is to validate all input and output I'm sure there's existing tools and libs out there in Python, not Repose, that would also work on typical schemas. 21:51:02 grapex: we just need input validation 21:51:33 grapex: feel free to take a look at the python space for libs. the best I found was schematics 21:51:40 there is not much 21:51:41 juice: Ok. 21:51:44 lol 21:51:48 validation is not terribly pythonic it seems ;) 21:51:54 #link https://github.com/j2labs/schematics 21:51:56 not to hijack this topic but i think https://review.openstack.org/#/c/28850/ is related to this 21:51:58 hub_cap: *sigh* 21:51:58 yeah its not so much 21:52:02 #link https://github.com/openstack/glance/blob/master/glance/api/v2/images.py#L205 21:53:06 by the time it reaches us we can validate w/ json schema id guess 21:53:06 we can talk about this later in another topic or right now 21:53:13 good call esmute 21:53:24 moving on 21:53:42 #topic open discussion 21:53:43 thanks please get back to me in the reddwarf if you want any input on this 21:53:51 juice: ya lets chat after 21:54:29 hub_cap: The glance thing looks pretty good, seems to support a standard, though its just JSON 21:54:50 Something to note, we transform XML stuff into dictionaries 21:55:03 i want to talk about ephemeral drive 21:55:03 which means we could probably run it through the JSON schema at that point 21:55:10 grapex: sure but by the tiem its gets to us its dict's 21:55:11 exactly 21:55:28 I can't wait for the day when we say "we support xml" and the world snickers and says "who still does that anymore???" 21:55:31 it just means we could have a delta between what our xml "says" and "does" 21:55:40 esmute: chat away 21:56:26 grapex: https://github.com/openstack/glance/blob/master/glance/api/v2/images.py#L482 21:56:35 hub_cap: I have incorporated your suggestion, which is using the same exisiting flags to determining ephemeral support 21:56:35 * SlickNik can hear esmute typing away 21:56:38 whens the vote? 21:56:51 yeah I'm waiting for my vote email 21:56:54 juice: what vote? 21:56:58 a pull requests is already up for review.. Im expecting robots to like it 21:56:58