Wednesday, 2010-09-22

*** uvirtbot has quit IRC00:06
*** thom has quit IRC00:07
*** thom has joined #openstack00:08
*** uvirtbot has joined #openstack00:08
*** mattt has quit IRC00:08
*** mattt has joined #openstack00:09
*** ArdRigh has joined #openstack00:10
*** thom has quit IRC00:14
*** sparkycollier has quit IRC00:17
*** pvo has joined #openstack00:20
*** pvo has joined #openstack00:20
*** ChanServ sets mode: +v pvo00:20
*** thom has joined #openstack00:21
*** rlucio has quit IRC00:21
*** zheng_li has quit IRC00:25
*** cloudmeat has joined #openstack00:26
*** uvirtbot has quit IRC00:32
*** mattt has quit IRC00:34
*** mattt has joined #openstack00:34
*** uvirtbot has joined #openstack00:34
*** thom has quit IRC00:34
*** howardroark has joined #openstack00:35
*** thom has joined #openstack00:35
*** DubLo71 has joined #openstack00:35
*** DubLo7 has quit IRC00:36
Ryan_Laneis there currently any web frontend for openstack?00:37
*** ppetraki has quit IRC00:40
*** joearnol_ has quit IRC00:42
*** odyi` has joined #openstack00:43
*** odyi` has joined #openstack00:44
*** odyi` is now known as odyi_00:44
*** odyi_ has quit IRC00:46
*** odyi` has joined #openstack00:47
*** sophiap has quit IRC00:48
*** odyi` is now known as odyi_00:48
*** odyi has joined #openstack00:49
*** mtaylor has quit IRC00:50
*** pvo has quit IRC00:51
*** silassewell has joined #openstack00:52
*** ar1 has joined #openstack00:56
*** cloudmeat has quit IRC00:57
*** maplebed has quit IRC01:00
*** metcalfc has quit IRC01:04
*** howardroark has quit IRC01:05
*** rlucio has joined #openstack01:06
jaypipesRyan_Lane: could you elaborate on that? do you mean a web-based control panel or client?01:06
*** lmcdowell has quit IRC01:07
Ryan_Laneweb based. I think I found it though01:08
Ryan_Lanetook me like two hours to find that :(01:08
jaypipesRyan_Lane: pls prefix your replies with jaypipes: :) otherwise I don't get notified...01:14
*** DubLo71 has quit IRC01:14
Ryan_Lanejaypipes: sorry, thought you were around01:14
Ryan_Lanejaypipes: is there something better for me to use than that web control panel?01:15
jaypipesRyan_Lane: yes, the docs are pretty horrible for that piece right now.  the control panel is essentially Rackspace's internal software that speaks the RS APIs.  There's more coming in that regard, as well as the mobile client stuff.01:15
Ryan_Lanegood to hear01:15
*** DubLo7 has joined #openstack01:16
jaypipesRyan_Lane: I suppose it depends on what you want to do with things.  I would suspect the mobile apps might end up being the preferred way of managing, but in the end, it's just talking the REST APIs of the OpenStack protocol, nothing more :)01:16
Ryan_LaneI know I'm trying a little early to use this :)01:16
Ryan_Laneno. web apps would be preferred for what we are trying to do01:17
jaypipesRyan_Lane: hey, we love the input!  Please don't hesitate to post to the mailing list(s) about your experiences and bug folks to improve docs and easy of use!01:17
*** ar1 has quit IRC01:17
Ryan_Lanehell, I'll update the docs01:17
Ryan_LaneI'm writing a puppet manifest as I'm installing it as well01:17
Ryan_LaneI can post that too01:17
Ryan_LaneI'm also putting together installation documentation01:17
jaypipesRyan_Lane: so, right now, there's still a lot of integration work going on between the various openstack sub-projects (nova, swift, glance, etc)...some was Rackspace stuff, other stuff was ec2 api based original stuff from the NASA folks, other stuff is brand new.  so it's a challenge to get everything talking politely to each other ;)01:18
Ryan_Lanejaypipes: hopefully we'll be some good publicity for you guys :)01:18
jaypipesRyan_Lane: ++ on posting puppet manifest!01:18
jaypipesRyan_Lane: :)01:18
jaypipesRyan_Lane: who do you work for/with?01:18
Ryan_Lanewikimedia foundation (wikipedia, et al)01:18
jaypipesnice :)01:19
jaypipesRyan_Lane: I used to work with Domas Mituzas...01:19
*** ar1 has joined #openstack01:19
Ryan_Laneheh. domas is great01:19
jaypipesindeed :)01:19
Ryan_Laneyou were at mysql previously? or sun?01:19
jaypipesRyan_Lane: both :)01:20
Ryan_Lanemakes sense :)01:20
jaypipesRyan_Lane: got laid off from Oracle after the acquisition finalized and landed at Rackspace.  Was originally working on Drizzle but have since moved on to working on OpenStack.01:20
Ryan_Laneoracle.... not my favorite company right now01:20
Ryan_Laneall the good talent from sun seems to be going to good places though ;)01:21
Ryan_Lanewell, I must be doing something wrong:
*** sophiap has joined #openstack01:23
jaypipesRyan_Lane: hmm :) /me not too familiar with the web-console project...01:24
Ryan_Laneno worries01:24
jaypipesRyan_Lane: lemme see who can help...01:24
creihtAs far as I know, it was more at a proof of concept stage01:24
jaypipescreiht: ah, Chuck :)01:24
jaypipescreiht: what is the name of the racker who writes the client apps?  his IRC nick?01:24
creihtI thought I heard the nasa guys were trying to get their web console open sourced01:24
Ryan_LaneI think I'm missing some files01:25
*** sophiap has quit IRC01:25
* jaypipes forgotten his name...01:25
creihtMike Mayo01:25
jaypipescreiht: he hang out on irc at all?01:25
jaypipescreiht: cheers.01:25
creihttook me a while to remember his nick :)01:26
jaypipescreiht: :)01:26
jaypipesRyan_Lane: well, when greenisus hops back on IRC he'd likely be the best person to ask..01:26
Ryan_Lanecool. thanks01:26
creihthe doesn't pop on very often, and I'm not really sure what the state or goal of the control panel is at01:26
jaypipesgrowing pains... ;)01:26
creihtdendrobates is probably a good person to ask :)01:26
jaypipesyeah, me neither....01:26
Ryan_Laneok, much better:
jaypipescreiht: but I had not heard that NASA was open-sourcing theirs...but then again, I'm kinda out of the loop on stuff :)01:27
Ryan_Lanewell, if this one is anything like rackspace's interface, I'd be happy01:27
jaypipesRyan_Lane: was it a web server conf setting? :)01:27
Ryan_Lanethe public dir was in the wrong spot01:27
jaypipesOK, I'm off to finish my blueprints...see y'all.01:28
Ryan_Lanesee ya01:28
Ryan_Lanethanks for the help/info01:28
*** sophiap has joined #openstack01:32
*** sophiap_ has joined #openstack01:39
*** sophiap has quit IRC01:40
*** sophiap_ is now known as sophiap01:40
*** DubLo7 has quit IRC01:47
*** DubLo7 has joined #openstack01:47
*** Ryan_Lane has quit IRC01:50
*** adjohn has joined #openstack02:15
*** adjohn has quit IRC02:20
*** johnbergoon has quit IRC02:39
*** DubLo7 has quit IRC02:52
*** perestrelka has quit IRC02:55
*** perestrelka has joined #openstack02:55
*** Cybodog has joined #openstack02:59
*** silassewell has quit IRC03:02
*** rlucio has quit IRC03:06
*** DubLo7 has joined #openstack03:09
*** joearnold has joined #openstack03:09
*** DubLo7 has quit IRC03:13
*** DubLo7 has joined #openstack03:14
*** littleidea has joined #openstack03:18
*** kashyapc has joined #openstack03:27
*** Pentheus has quit IRC03:30
*** rlucio has joined #openstack03:38
*** kashyapc has quit IRC03:39
*** DubLo7 has quit IRC03:41
*** pvo has joined #openstack03:59
*** ChanServ sets mode: +v pvo03:59
*** rlucio has quit IRC04:02
*** krish has quit IRC04:03
*** krish has joined #openstack04:15
*** kashyapc has joined #openstack04:35
*** mtaylor has joined #openstack04:54
*** ChanServ sets mode: +v mtaylor04:54
*** omidhdl has joined #openstack04:56
*** pvo has quit IRC04:58
*** zaitcev has quit IRC05:02
*** mtaylor has quit IRC05:05
*** ArdRigh has quit IRC05:11
*** mtaylor has joined #openstack05:13
*** ChanServ sets mode: +v mtaylor05:13
*** littleidea has quit IRC05:18
*** joearnold has quit IRC05:20
*** Cybodog has quit IRC05:22
*** Cybodog has joined #openstack05:25
*** scottie has quit IRC05:31
*** littleidea has joined #openstack05:48
*** Ryan_Lane has joined #openstack05:50
*** allsystemsarego has joined #openstack05:51
*** spackest has joined #openstack05:52
spackestanyone out there?  just pondering trying to set up a distributed install of swift and am wondering if there are any docs out there05:52
*** krish has quit IRC05:54
*** samppah has quit IRC06:03
*** ibarrera has joined #openstack06:04
*** spackest has quit IRC06:08
*** sirp1 has quit IRC06:25
*** rlucio has joined #openstack06:32
*** zheng_li has joined #openstack06:54
*** metoikos has joined #openstack07:00
*** brd_from_italy has joined #openstack07:01
*** Ryan_Lane has quit IRC07:14
*** f4m8_ is now known as f4m807:16
*** sirp1 has joined #openstack07:17
*** sirp1 has quit IRC07:26
*** sirp1 has joined #openstack07:28
*** metoikos_ has joined #openstack07:35
*** metoikos has quit IRC07:36
*** maplebed has joined #openstack07:51
*** calavera has joined #openstack07:55
*** arcane has quit IRC08:09
*** arcane has joined #openstack08:15
*** zheng_li has quit IRC08:20
*** littleidea has quit IRC08:22
*** zheng_li has joined #openstack08:39
*** kashyapc has quit IRC09:14
*** kashyapc has joined #openstack09:14
*** sureshgv has quit IRC09:24
*** maplebed has quit IRC09:24
*** maplebed has joined #openstack09:26
*** krzycoder has quit IRC09:31
*** miclorb_ has quit IRC09:34
*** sophiap has quit IRC10:27
*** sophiap has joined #openstack10:34
*** gustavomzw has joined #openstack10:35
*** cloudmeat has joined #openstack10:46
*** Podilarius has left #openstack10:56
*** omidhdl has left #openstack10:58
*** scottie has joined #openstack11:15
*** gasbakid has joined #openstack11:22
*** ctennis has quit IRC11:31
*** cloudmeat has quit IRC11:32
*** dabo has joined #openstack11:44
*** dabo has quit IRC11:46
*** msparks has left #openstack11:49
*** cloudmeat has joined #openstack11:51
*** scottie has quit IRC11:55
*** miclorb_ has joined #openstack12:04
*** dendrobates is now known as dendro-afk12:08
*** befreax has joined #openstack12:09
*** gundlach has joined #openstack12:11
*** hazmat has joined #openstack12:20
*** sophiap has quit IRC12:22
*** localhost has quit IRC12:26
*** localhost has joined #openstack12:26
*** miclorb_ has quit IRC12:30
*** sophiap has joined #openstack12:39
*** ctennis has joined #openstack12:47
*** ctennis has joined #openstack12:47
*** klord has joined #openstack12:55
*** coredump|br has quit IRC13:00
*** annegentle has joined #openstack13:05
*** coredump|br has joined #openstack13:07
*** deshantm has quit IRC13:09
*** kashyapc has quit IRC13:12
*** rlucio has quit IRC13:15
*** Cybo-mobile has joined #openstack13:16
*** prank has joined #openstack13:17
*** prank has left #openstack13:17
*** Podilarius has joined #openstack13:28
*** pvo has joined #openstack13:28
*** pvo has joined #openstack13:29
*** ChanServ sets mode: +v pvo13:29
*** DaFrog has joined #openstack13:32
*** scottie has joined #openstack13:36
*** pvo has quit IRC13:51
*** Cybodog has quit IRC13:57
*** anm_ has joined #openstack13:58
*** Cybodog has joined #openstack13:59
*** calavera has quit IRC13:59
*** befreax has quit IRC14:02
*** ppetraki has joined #openstack14:03
*** abecc has joined #openstack14:05
*** mdomsch has joined #openstack14:05
*** mdomsch_ has joined #openstack14:06
*** f4m8 is now known as f4m8_14:07
*** pharkmillups has joined #openstack14:09
*** spackest has joined #openstack14:17
*** befreax has joined #openstack14:18
*** rnirmal has joined #openstack14:20
notmynamespackest: is the best place to start14:27
*** Cybo-mobile has quit IRC14:33
*** metoikos_ has quit IRC14:34
*** metoikos_ has joined #openstack14:35
*** tobym has joined #openstack14:38
*** lmcdowell has joined #openstack14:42
creihtspackest: namely, you are going to be interested in the things under Deployment:14:46
*** littleidea has joined #openstack14:50
notmynameanyone from cloudscaling here?14:52
*** deshantm has joined #openstack14:53
*** jdarcy has joined #openstack15:03
*** befreax has quit IRC15:04
*** maplebed has quit IRC15:04
*** befreax has joined #openstack15:05
*** mdomsch has quit IRC15:05
*** littleidea has quit IRC15:12
*** gasbakid has quit IRC15:25
*** cloudmeat has quit IRC15:29
*** cloudmeat has joined #openstack15:29
*** annegentle has quit IRC15:33
*** sophiap has quit IRC15:33
*** kashyapc has joined #openstack15:34
*** annegentle has joined #openstack15:36
*** sophiap has joined #openstack15:37
rackerhackertimrc: got your cgroups ;)15:44
*** dendro-afk is now known as dendrobates15:46
*** scottie has quit IRC15:46
*** scottie has joined #openstack15:47
*** spackest has quit IRC15:48
timrcrackerhacker: music to my ears :)15:48
timrcrackerhacker: I'll play when I get home15:48
timrcrackerhacker: if only I got paid to work on this stuff :)15:48
rackerhackertimrc: hah (look for
*** scottie has quit IRC15:51
*** ibarrera has quit IRC15:55
*** befreax has quit IRC15:56
alekibangocgroups are very nice :)15:57
*** metcalfc has joined #openstack15:59
*** spackest has joined #openstack15:59
*** zheng_li has quit IRC16:01
*** abecc has quit IRC16:05
*** abecc has joined #openstack16:05
*** Mr_T has quit IRC16:06
*** Mr_T has joined #openstack16:07
*** dendrobates is now known as dendro-afk16:07
*** tobym has quit IRC16:12
*** kevnfx has joined #openstack16:12
*** brd_from_italy has quit IRC16:13
*** joearnold has joined #openstack16:17
*** krish has joined #openstack16:19
*** sirp1 has quit IRC16:21
*** zheng_li has joined #openstack16:23
*** maplebed has joined #openstack16:28
*** kashyapc has quit IRC16:34
*** joearnold has quit IRC16:36
*** burris has quit IRC16:37
*** rlucio has joined #openstack16:49
*** kashyapc has joined #openstack16:52
*** annegentle has left #openstack16:53
*** hazmat has quit IRC16:54
*** tobym has joined #openstack17:02
*** burris has joined #openstack17:07
*** hazmat has joined #openstack17:08
*** Ryan_Lane has joined #openstack17:10
gundlach_cerberus_/eday: mind reviewing my rewrite of the rbac unit tests now that rbac is done in WSGI middleware?17:12
_cerberus_I can in a bit if eday can't17:12
edaygundlach: sure, will take a look in a min17:13
gundlachthx to whomever wins :)17:13
*** spackest1 has joined #openstack17:13
*** hazmat has quit IRC17:14
*** spackest has quit IRC17:15
*** Cybodog has quit IRC17:16
*** Cybodog has joined #openstack17:18
*** Cybodog has quit IRC17:24
*** Cybodog has joined #openstack17:26
*** hazmat has joined #openstack17:27
rluciovishy: thx for the comment yesterday on deb packaging.   it will be cool to get an update, esp. since the current debs in soren's ppa are ancient now17:28
rluciothey are like.. a month old17:28
*** krish has quit IRC17:29
alekibangorlucio:  you know, there is debian-packaging. is that used as base for ubuntu packages also? can u build/use ubuntu packages using that?17:31
rlucioalekibango: well, the whole conversation started b/c the debian-packaging branch is not really usable with HOL17:32
rluciofor instance, some of the cmd line switches it uses in configuration files have changed17:33
rlucioand it doesn't know anything about nova-network17:33
alekibangouh, what is hol ?17:33
rluciosry, head of line,17:33
alekibangothere are so many strange abbrevs, there should be wiki dictionary :)17:34
rluciosry, i have a long cvs backgroung *shudder*17:34
alekibangoyesterday i scratched my head about FF -- but fortunatelly i realized it  might be Feature Freeze :)17:34
alekibangorlucio: i never enjoyed CVS, switched to svn as fast as i could. now on git mostly :)17:35
alekibangonow i am trying to be friendly with bzr and launchpad.... grr so many standards... in real word, people do not use 40 different languages when talking to each other...17:36
jdarcyalekibango: O RLY?17:37
alekibangonot if its 5 people in one office17:37
* alekibango blames Babylon for the chaos17:38
rlucioalekibango: haha funny i think my team went through the same thing about six weeks ago, we use svn in house, and know git pretty well17:38
alekibangocoming from svn i was so confused by git checkout  ....17:39
alekibangobut i must say its better....17:39
alekibangoor git reset - so similar words, so different meaning17:40
*** hazmat has quit IRC17:40
*** kashyapc has quit IRC17:42
alekibangorlucio: imho we should  take that debian-packaging  and shape it -- to have it useable.17:42
alekibangoand use it for all debian derivatives as a base17:42
*** Cybodog has quit IRC17:43
rluciooh trust me, i want to, but right now i can't code on the project -- i don't have access granted from my employer yet17:44
alekibangoi am my own employer, which makes it easier for me17:45
rlucioah that certainly cuts out the middle man17:45
alekibangoit cuts you directly most of the time :)17:45
*** sirp1 has joined #openstack17:45
rluciowell, at least you don't have to sit in the channel and beg to the devs about when they are going to finish updates and such :)17:46
*** ctennis has quit IRC17:47
alekibangorlucio: even i have boss - customer...17:48
rluciothat's a good boss to have17:48
*** kashyapc has joined #openstack17:53
*** hazmat has joined #openstack17:54
*** pharkmillups has quit IRC17:55
*** pvo has joined #openstack17:56
*** ChanServ sets mode: +v pvo17:56
*** gustavomzw has quit IRC17:57
*** ElectricBill has quit IRC17:59
*** dendro-afk is now known as dendrobates18:01
vishygundlach: having problems with your last patch18:01
vishygundlach: are you here?18:01
edayvishy: what kinds of problems?18:03
edayvishy: and is this for the ec2 api move?18:03
gundlachvishy: yep18:03
vishygundlach: seem to be related to unicode18:04
vishyso nova-api is broken because the old APIServerApplication is gone18:04
vishytrying nova-api-new18:04
vishysome weird unicode issues18:05
vishy resp.body = str18:05
vishyin _error18:05
vishycomplains that it can't body can't be set to a unicode object18:05
vishyso i wrapped that in str()18:06
vishybut it appears that all params are getting into wrong18:06
vishyfor example euca-run-instances -t m1.tiny ami-tiny18:07
vishythe params print out correctly in the log18:07
vishybut when they get into run instances, they look like this:18:07
vishy{'min_count': u'1', 'instance_type': u'm', 'max_count': u'1', 'image_id': u'a'}18:07
vishyso just the fist character of each param as unicode18:07
vishyi'm thinking of throwing apiserver back in to api/ec2/ or some such18:09
vishyso i have an interim solution, unless you have a quick fix...18:10
*** kevnfx has quit IRC18:13
alekibangovishy: paste that error please somwhere18:13
*** littleidea has joined #openstack18:14
*** dendrobates is now known as dendro-afk18:16
*** zaitcev has joined #openstack18:17
vishymessage happens to be unicode because instance_type is unicode18:18
vishyand it is in the error message string18:19
*** kashyapc has quit IRC18:20
*** sparkycollier has joined #openstack18:20
edayvishy: what if you s/body/unicode_body/ ?18:21
vishywell it isn't always unicode18:21
vishybut i could try that18:22
vishybut that issue is minor18:22
edaythat should be fine, non-unicode in a unicode type is still ok18:22
vishythe paramaters coming in wrong to is not so good18:22
*** hisaki has joined #openstack18:22
edayahh, ok18:22
gundlachsorry, had to hop away18:23
gundlachvishy, i assume this is in trunk?18:23
vishygundlach: i have a few of my patches applied18:23
vishyi can check in trunk real quick18:24
gundlachcan you point me to a place where i can suck down your code and repro?18:24
*** metoikos_ is now known as metoikos18:25
gundlachand the simplest repro you've got is the euca-run-instances thing above, correct?18:25
vishythere is a bug in trunk with nova-manage right now as well18:26
gundlachyes, eday pointed out that i left some things hanging when i removed nova/endpoint -- i didn't realize there was code uncovered by unit tests, so i just got unittests to pass18:27
gundlachwhat's the repro for the nova-manage bug?18:27
vishyany command...fix is easy18:28
gundlachah, ok so unrelated to the unicode thing18:28
vishy-from nova.endpoint import cloud18:28
vishy+from nova.api.ec2 import cloud18:28
edayvishy: well, that's not the real 'fix', I'm working a propoer cloudpipe fix (cloud pipe is busted in trunk right now)18:29
vishyeday: well, it makes nova manage run18:29
vishyeday: cloudpipe fix would be awesome18:29
*** pharkmillups has joined #openstack18:30
vishyeday, gundlach: the error is in trunk18:30
vishywith params comming in as single char unicode objects18:31
vishyi'm just sticking a print kwargs in the top of run-instances in cloud.py18:32
*** scottie has joined #openstack18:33
*** annegentle has joined #openstack18:35
vishygundlach, eday: should i throw APIServerApplication back in somewhere so the old nova-api still works?18:36
gundlachvishy: no, nova-api should be going away18:36
gundlachand nova-api-new being renamed to nova-api18:36
gundlach(if eday isn't doing it as part of cleaning up my mess, i'll be doing it)18:36
gundlachdid you find anything when you printed kwargs in run-instances?18:36
sorenvishy: Perhaps you can help me out..18:37
sorenvishy: I'm looking at some odd behaviour in the rbac checking code.18:37
gundlachsoren: fyi i may be the one you want now, as i just reworked the rbac code and tests18:37
sorenvishy: The logic seems weird, but there's a unit test that specifically tests for this logic, so I'm not sure what its intention is.18:37
sorengundlach: I noticed :)18:38
sorenAnyways, the problem is this:18:38
sorenIn nova.tests.auth_unittest.AuthTestCase.test_210_can_add_project_role..18:38
edaygundlach: I have the nova-api* swap in my branch already with cloudpipe fixes18:38
*** spackest1 has quit IRC18:39
gundlacheday: cool, i figured; i won't touch it18:39
alekibangoheh, vishy, sorry i afked - but luckily eday came in :)18:39
sorenIt creates a project, testproj, and a user, test1.18:39
sorenThen it assigns the *global* "sysadmin" role to that user.18:39
soren        self.manager.add_role('test1', 'sysadmin')18:39
sorenThen it checks whether project.has_role('test1', 'sysadmin') returns False.18:40
sorenThis makes no sense to me.18:40
vishygundlach: kwargs are all coming in as 1 character unicode strings18:40
vishyhas_role is perhaps a bad name18:40
sorenLooking at the code in the auth manager, it seems to refuse to allow you to have a role on a project if you don't also have the role globally.18:40
*** spackest has joined #openstack18:41
sorenAround line 370, it does this:18:41
sorenIt checks if the user has the role globally.18:41
sorenIf not, it returns False without checking the role on the project.18:41
vishyso your roles are defined as the intersection of your 'global' i.e. user-roles and your project roles18:41
vishyso to be a 'netadmin' you need the 'netadmin' role18:42
sorenthe docstring specifcically says union.18:42
vishyso it is a docstring error :)18:42
sorenOk, but disregarding the docs..18:42
sorenWhat's the point of this?18:42
sorenWhy mask roles like this?18:42
vishyspecific use-case18:42
gundlachvishy: i don't have nova set up and haven't run euca-run-instances before... any chance you could make a unit test showing the failure?18:42
vishygundlach: I'm not really sure how to get stuff into through the wsgi layer18:43
sorengundlach: I'm curious what we're using for our WSGI container now that tornado is gone. I could just look at the code, but I'm trying to figure this other thing out :)18:43
gundlachsoren: eventlet provides a wsgi server.18:43
sorengundlach: Ah, convenient.18:44
edayvishy: with debug logging enabled, do you see a block of "action: ...\n arg: ...=...\n, ..."? Do those look correct?18:44
vishysoren: so the idea is, project managers can set roles for people in their project18:44
sorenvishy: Makes sense.18:44
vishysoren: but a project manager can't give netadmin creds out18:44
gundlachvishy: ok so if you open nova-manage and pass stuff to directly, you don't have trouble, but instead the WSGI layer is passing unicode-single-byte data to functions in  Is that a correct summary?18:44
sorenDoes not make sense :)18:44
vishythat is a property of the user18:44
sorenvishy: I'm struggling to see how this is useful.18:45
vishyspecific to us, there is a certification process you have to go through to get the global role18:46
vishybut just because you have the global role, as in you can have netadmin priveleges18:46
vishyyou may be on 5 projectrs18:46
gundlachsoren: maybe because some roles (user-roles) are cross-project?  e.g. maybe netadmin is used for controlling things at a deeper layer than a project.  so a project owner has no right to give that role to a user, unlike project roles which are for the project?18:46
gundlach(trying to interpret vishy)18:46
vishyand may not have netadmin priv on all projects18:46
sorenvishy: Ok, I can see how that may make sense for you guys (given the certification thing).18:47
sorenvishy: It doesn't sound useful at all to the rest of the world. Like me :)18:47
vishyyeah like i said, it may be a specific case for us18:47
vishygundlach: passing stuff to works fine18:47
vishybut it appears somewhere in the translation layer kwargs are not making the transition18:48
vishygundlach: they print out properly in the logging statement18:48
*** joearnold has joined #openstack18:48
vishyso the problem is somewhere between line 115 of ec2__init__.py18:49
gundlachvishy: what line in what file does the logging?18:49
gundlachwow, that was a fast response ;)18:49
vishygundlach: in Router.__call__18:49
vishyi was already typing it18:49
vishyso it seems like it must be in map?18:50
sorenvishy: Ok, thanks for the explanation. At least now I understand and can work around it.18:50
vishyor dec.wsgify18:50
vishysoren: there is a get_roles which you can use for actual roles18:50
vishysoren: and just avoid has_role18:50
sorenvishy: This is done in the CloudController.18:51
gundlachvishy: what happens if you add this on line 116:             _log.debug('type-arg: %s\t\ttype-val: %s' % (type(key), type(value)))18:51
vishysoren: we could even rename it has_intersected_role or something18:51
sorenvishy: So it's not local to my problem.18:51
sorenI'm just surprised I'm the only one with testcases that needed adjustment now.18:51
*** cloudmeat has quit IRC18:52
*** cloudmeat has joined #openstack18:52
soren(This came up after the checks were moved from the rbac decorators to the cloudcontroller)18:52
gundlachsince the log is printing out str() instead of repr() versions of these strings, maybe they are already unicode by that point18:52
gundlachsoren: i tried to keep the rbac functionality identical, with the exception of axeing @rbac.deny() -- are you saying that your test case in fact failed with the code change?18:53
vishygundlach: they are unicode18:53
gundlachah hh18:53
*** hisaki has left #openstack18:53
vishygundlach: but they are the whole string, does the wsgi mapping split on the unicode characters?18:54
sorengundlach: I only just sat down at my laptop again after having been away for 6 hours. Let me get my thoughts straight before I answer :)18:54
gundlachsoren: cool :)18:54
gundlachvishy: oh, i see your point.  no, when line 120 puts args into req.environ['ec2.action_args'], that's just a simple dictionary18:55
gundlachso that shouldn't be trimming/splitting/anything.18:55
*** pharkmillups has quit IRC18:55
*** pharkmillups has joined #openstack18:55
gundlachvishy: the next guy who uses them is api.ec2.apirequest.APIRequest.send; try logging kwargs[action_args] on line 54 of
vishygundlach: yeah i converted them to strings, now they are 1 char strings instead of unicode18:56
sorenWhat's your problem?18:56
vishyok gundlach, checking18:57
sorenThis sounds like a problem I had earlier.18:57
vishybb in a bit18:57
sorenOh, that.18:57
sorenI know what that is.18:57
gundlachvishy: ^^18:57
*** mdomsch has joined #openstack18:57
sorentornado used to always pass args as lists.18:57
* gundlach smacks his forehead18:57
sorenThat's no longer that case.18:57
sorenSo now, the value[0] thing just gives you a single char.18:57
sorenIt's in...18:57
sorenhang on.18:57
sorenline 73-ish.18:58
gundlachyep, i see it too18:58
sorenThe else needs to just go away.18:58
gundlachsame deal on line 79-ish?18:58
* soren stares at it for a little bit18:59
sorenI don't know.18:59
sorenI don't know how eventlet's think serialises this stuff.19:00
sorenI didn't make it that far yet.19:00
gundlachas a regular dictionary of keys to values19:00
sorenI'm still trying to get the stuff that used to work to work again.19:00
gundlachno lists.  i just can't figure out what line 79 is trying to do19:00
sorenI know that.19:00
gundlachmaybe a Bexar goal would be 100% unit test coverage and thorough code comments :)19:00
sorenSo, when the EC2 API needs a list of stuff, it takes them as Foo.1=bar1&Foo.2=bar2 etc.19:01
sorenThis used to get pased to the API methods as lists.19:01
sorenI guess that's the bit of code that handles implements that bit of magic.19:02
sorenMakes sense?19:02
alekibangosoren:  i have another question :)  -- There is a branch called debian-packaging.  You have your ubuntu branch...    isnt that a bad thing? i would think that we need common branch for all debian derivatives, so packagers will work together -- or did i miss something? I would like to have good deb packages for Debian... and i am willing to help, when i will get my head around this all.19:02
sorenalekibango: Two things:19:03
sorenalekibango: Ubuntu and Debian are similar. Not identical.19:03
sorenalekibango: and19:03
gundlachsoren: the line                 if args[key] != {} and args[key].keys()[0].isdigit():   ?   that's dealing with the .1, .2, etc?19:03
alekibangothats clear.. but also debian stable and testing can differ a lot19:03
sorenalekibango: Nova isn't in Debian. It's in Ubuntu.19:04
sorenalekibango: ..and I wanted somewhere to maintain stuff. At the very least the changelog will differ between the two.19:04
alekibangosoren: imho there should be at least common base --  and maybe even only one branch19:05
sorenalekibango: So a separate branch popped up. Once someone with sufficient determination decides to shove Nova in Debian (I already pushed all the dependencies), they're likely going to consolidate somewhat.19:05
sorenBut I'll always have a separate Ubuntu branch. I can upload to Ubuntu, not to Debian.19:05
alekibangoi am not DD either.19:05
sorenMy life it too short to wait for a sponsor every single day.19:05
sorenalekibango: The debian-branch will likely be the common base.19:06
sorenalekibango: ..but until Nova actually is in Debian, it makes no sense.19:06
sorenPut bluntly, I became an Ubuntu developer because I wanted to get shit done, not argue about shit.19:06
sorenPardon my French.19:07
alekibangowell :) i can understand that pov19:07
alekibangobut still i prefer using debian on server19:07
alekibango-> being able to build deb packages DAILY would be nice19:08
sorenGetting stuff into Debian still, at best, takes a day or so. And, again, Debian and Ubuntu are not identical. There will be differences in the packaging.19:08
alekibangoeven without DD19:08
sorenHey, feel free to build packages. It's not hard to set up.19:08
sorenI just can't be bothered, really.19:08
sorenI didn't spend all this time making Ubuntu awesome just to not make use of it.19:08
alekibangothats called emotional investment :)19:10
sorenIt's not just emotional. It's technical, too.19:11
* soren cries19:12
alekibangosoren: 'only ubuntu exists' approach will not work well in the long run, and you know  it :)19:12
sorenalekibango: Please don't put words in my mouth.19:12
sorenI said nothing of the sort.19:12
*** dendro-afk is now known as dendrobates19:13
*** burris has quit IRC19:13
alekibangowe prolly need more debian people to shape up debian-packaging branch... - then you will agree to have common base?19:14
sorenDebian is fantastic. It's a fantastic base for Ubuntu to build on. It's just that I'm on a mission, and having to spend hours and days bikeshedding over useless details gets in the way.19:14
alekibangoyou know, i am looking for synergy here...  iam not here to make distro wars19:14
sorenI never said I didn't want a common base.19:15
sorenI even want it to be the debian branch.19:15
alekibangosoren: btw many times i loved that debian did such obstacles and policies -> they have things really documented and systematic :)19:15
sorenIẗ́'s just a complete and utter waste of time for me to make my changes in the debian branch and merge them into the Ubuntu branch when the debian branch isn't being used for anything. What would be the point?19:15
*** joearnold has quit IRC19:15
*** cloudmeat has quit IRC19:16
alekibangoi miss such pedantic work on ubuntu :)19:16
*** cloudmeat has joined #openstack19:16
alekibangoyou know, the 'fascism' of policies19:16
alekibangosoren: i see what you mean....19:16
alekibangoi had to ask, because there is not good description on branches19:17
sorenIf someone actually wants to upload nova to Debian, more power to them.19:17
*** joearnold has joined #openstack19:17
sorenUntil then, it's *completely* pointless (and would be confusing) to go through the Debian branch.19:17
alekibangook, thanks for you explanation19:18
*** joearnold has quit IRC19:19
*** Ryan_Lane is now known as Ryan_Lane|food19:21
alekibangoi will look for some debian developers later, after the freeze... and i will try to push nova into debian...19:21
gundlachsoren: thanks for your help.  vishy: i'm going to write a failing unit test, fix this, and merge it.  in the meantime, you should be able to follow soren's instructions above re to proceed19:21
*** brd_from_italy has joined #openstack19:25
*** tonywolf has joined #openstack19:26
sorengundlach: Sure thing, man.19:26
sorengundlach: Oh, I missed your question earlier. Yes, I think that code is meant to deal with the .1, .2 etc.19:27
gundlachthx.  really hard to wrap my head around what it's doing.19:27
*** gustavomzw has joined #openstack19:29
*** dendrobates is now known as dendro-afk19:29
gundlachvishy: could you send me the URL and headers generated by euca-run-instances that ended up causing that failure?  I want to encode that specifically as a unit test.19:30
*** dendro-afk is now known as dendrobates19:31
*** brd_from_italy has quit IRC19:37
*** brd_from_italy1 has joined #openstack19:37
sorengundlach: I already have a unit tests that fails as a result of that change.19:37
sorengundlach: That's how I found it :)19:37
sorengundlach: (it's in a not-yet-merged branch)19:37
sorengundlach: Just fyi.19:37
gundlachsoren: oh, sweet... then um do you want to just drop the else clause and the [0] and merge your branch?  or is it not ready yet?19:38
sorengundlach: I think I just finished merging with trunk, so it should be ready in a couple of minutes.19:38
sorenIt does not cause additional unit tests to fail.19:39
gundlachok i will abandon my branch; thanks for the fix.19:39
sorennova.tests.scheduler_unittest has a bunch of weird stuff going on on my box, but that's true on trunk, too, so I'll look at that later.19:39
edaygundlach: it looks like you dropped the ec2 /latest, /1.0, and /<API date> url mappings as well.. any reason? I'm guess that may break some ec2 client libraries19:39
soreneday: What's left without those?19:40
edaysoren: /services19:40
gundlacheday: no reason except that i assumed they were legacy nasa stuff.  i can put them back in if we do in fact find problems19:40
sorengundlach: Clients (when doing the right thing) access a URL specific to the API version they were written for.19:41
gundlachagain, without unit tests, as far as i knew/was concerned it worked.19:41
edaygundlach: Hmm, those are part of the ec2 API than client libs will probably expect, I wouldn't just drop them19:41
sorengundlach: Does the current code just drop the first path element?19:41
edaygundlach: a lot of things are missing unittests, we wouldn't have much if we removed all things that didn't :)19:41
gundlacheday: i'm coming to realize that :)19:41
gundlachsoren: not sure what you mean, but the ec2 endpoint is served off of <app root>/services19:41
sorenAll the time I'm not spending actually fixing bugs after Feature Freeze will be spent on writing tests.19:42
gundlachperhaps it should be served off of <app root>/<whatever API date it is conforming to>/services19:42
sorengundlach: Ok, that answers my question.19:42
edaysoren: and then fixing the bugs the tests expose :)19:42
gundlachsoren: so you are sure that /v1.0 and /latest are part of the EC2 api?19:42
soreneday: Sure. Otherwise it can't get merged :)19:42
sorengundlach: Positive.19:42
gundlachsoren: k, thx.  i'll go look at the old code to see what they should return :)19:43
gundlach[and will add them back forthwith]19:43
edaygundlach: they all map to the same metadata request, so should be easy to add back in19:43
sorengundlach: Oh, by the way... One of your commit messages suggested that my updating eventlet on Hudson hadn't worked. Is that true?19:49
edayof course there is a conflict on GET /, so we may want to detect hostname to serve the appropriate one ( or something). Could also add a new top-level path element19:49
gundlachsoren: correct; but mtaylor took care of it19:50
gundlachsoren: re eventlet on hudson19:50
sorengundlach: What was the problem?19:50
gundlachsoren: no idea -- mtaylor swooped down and fixed that and the "deleting parent endpoint" mystery merge conflict :)19:50
sorengundlach: 2010-09-17 08:09:07 status installed python-eventlet 0.9.12-0ubuntu1~lucidppa119:50
gundlachi, a mere mortal, didn't hear the secret to his success19:51
sorengundlach: From dpkg.log on the hudson box.19:51
sorenmtaylor: I can haz enlightenment, plz?19:51
gundlachsoren: don't ask me; ask mtaylor.  he said "ok, so eventlet 0.9.10 is still installed" or something.19:51
sorenThis was yesterday, right?19:51
gundlach2010-09-21T18:59:33  <mtaylor> ok. no - eventlet 0.9.10 is still installed19:51
gundlach2010-09-21T18:59:37  <gundlach> WHEW19:51
gundlach2010-09-21T19:00:00  * gundlach steps back from the abyss19:51
gundlach2010-09-21T19:00:10  <mtaylor> gundlach: ok. 0.9.12 is now installed19:51
* soren grumbles19:52
sorenSep 21 19:01:11 openstack-hudson sudo:  mordred : TTY=pts/0 ; PWD=/home/mordred ; USER=root ; COMMAND=/usr/bin/easy_install eventlet19:52
* soren loathes easy_install19:52
mtaylorsoren: me too19:52
gundlachyeah, pip -U fwt19:52
gundlachspelling ftw19:52
sorenspelling wtf19:52
mtaylorsoren: but that's what was already on there - I wasn't sure if there was an 0.9.12 package anywhere19:52
sorenmtaylor: It was even installed. For days :)19:53
mtaylorsoren: 0.9.10 was sitting in /usr/local when I went and looked at the problem19:53
*** Ryan_Lane|food is now known as Ryan_Lane19:53
mtaylorsoren: heh19:53
mtaylorsoren: then easy fix - we can just delete the stuff from /usr/local :)19:53
sorenmtaylor: So you removed the easy_installed one?19:53
mtaylorsoren: no19:53
mtaylorsoren: but we can do now19:53
sorenmtaylor: What does this do, then: /usr/bin/easy_install -U eventlet ?19:53
sorenWould you mind? I'm not easy_install savvy.19:53
mtaylorsoren: essentially, I did "print eventlet.__file__"19:53
mtaylorno problem - doing now19:54
sorenI'd like to use the packaged version so that everything is in sync.19:54
mtaylorcompletely agree19:54
sorenWhat was the other problem?19:54
sorenOr rather, what was the solution to the other problem?19:54
mtaylorno clue - it went away magically19:54
gundlachsomething re "deleting parent endpoint" -- cerberus has seen a similar error in his work.19:55
sorenYeah, I had seen it to and could reproduce and fix it, but my fix didn't take on the hudson box.19:55
mtaylorsoren: there is also a copy of webob installed via easy_intsall - should I delete that too?19:55
sorenmtaylor: There's a packaged version, so I'd say yes.19:56
mtaylor>>> print eventlet.__file__19:56
mtaylorsoren: ok. all fixed19:57
*** tonywolf has quit IRC19:58
rluciosoren: i was reading the earlier conversation.... do you have an updated branch of debian-packaging available?20:01
edaygundlach: did you test to make sure the rpc calls worked after moving to wsgi? it appears it still attempts to attach to tornado...20:02
vishysoren, gundlach: thanks for the suggestions, had to run out to a lunch meeting20:02
vishyeday: i was wondering that myself, if ioloop isn't started they will probably fail20:02
edayvishy: yeah20:02
gundlacheday: nope!  not at all.20:02
gundlachi haven't ever run nova on my box, i think -- just the unit tests.20:03
vishyeday: there is a branch from termie moving things to twisted20:03
edaygundlach: ugh, there is still a lot of work todo to get ec2 running again20:03
gundlachblech.  well, if we need to revert the branch or something, let's do that.20:03
vishyeday: we probably need to merge that and switch call to use twisted?20:03
edaygundlach: almost wondering if we should back it out until all these things get sorted20:03
edayvishy: we don't need twisted at all, eventlet will let you use carrot without messing with deferreds or anything20:04
*** tonywolf has joined #openstack20:04
edayvishy: twisted will only complicate things if we want to stick with WSGI20:04
vishyeday: ok we need a version of call for twisted and eventlet then20:05
gundlacheday: what do you say to my counterargument: we keep trunk as it is, and in any case where ec2 fails, we write a failing test and fix it.20:05
* gundlach is loathe to back out, and frustrated that the unit tests missed so much, but will back out if it's blocking people's work20:06
vishyeday: termie said that the call should still work in twisted, because it doesn't interfere with ioloop, but i don't know about ioloop combined with eventlet20:06
edayvishy: yep, and the eventlet one is simple, just use it as you would a simple app with no ioloop/deferreds20:06
edayvishy: we shouldn't use ioloop at all with eventlet20:07
edayvishy: we should jsut do the normal carrot.consume.wait() call20:08
edayvishy: it won't actually 'block', since eventlet wraps the socket calls underneath to be non-blocking, and context switches the different eventlet tasks automatically20:09
vishysoren: did you get my merge patch to your branch?20:09
edayvishy: this is different from twisted, where you manage the deferred stack manually20:09
vishyeday: I understand, but we need this working asap20:10
vishyeday: or we need to put the tornado version back in in the interim20:10
edayvishy: yup. so assuming this will take a day or two to sort out, would you rather back it out or wait till we can get the patches in to fix20:10
vishyafk 2 min20:10
edaymtaylor: how much of a PITA would it be to back a rev out from nova trunk?20:11
gundlacha big rev...20:11
edayat least it's still the tip, makes it a lot easier20:13
eday(no one approve anything for a few minutes :)20:13
mtayloreday: not too bad - want me to remove the tip rev?20:16
sorenvishy: Just merged it now, thanks.20:16
gundlacheday: exactly what is remaining to do to get ec2 done, that you're aware of?20:16
mtayloreday: rev 290?20:17
sorenvishy: I'm not sure how it happened, but I seem to have had a bunch more changes that didn't get pushed to launchpad, so your patch was against an old version of my branch, and... well, a bunch of stuff just blew up in my face.20:17
edaymtaylor: one sec...20:17
edaygundlach: not sure yet till we run some end-to-end tests20:17
mtayloreday: k. it's no problem to do if that's what you decide you want20:17
edaygundlach: cloud pipe + the rpc stuff is certainly broken, I just don't want to hold up vishy and folks since they need it now20:18
edaymtaylor: cool, thanks20:19
gundlachok, as i'm not familiar with the code in question, having spent my time in nova/endpoint, i guess let's pull it out, and i'll track down what's not covered by tests.20:19
gundlachmtaylor: let's revert r29020:19
*** gustavomzw has quit IRC20:20
edaygundlach: so, the thing is we're not really talking about just unittests here, it's a bit close to integration tests... making sure a HTTP request produces a valid rpc call for the various works and such20:21
edayerr, workers20:21
mtayloreday: we really need to get hudson to run integration tests too...20:21
mtayloreday: sorry, I've been meaning to get that set up for a while now20:21
edaymtaylor: I think we need better tests written too (both unit and integration) :)20:22
gundlacheday: cool, unit tests, functional tests, integration tests -- whatever we're missing, let's get them written and working :)20:22
edaymtaylor: awesome, thanks20:22
edayvishy: the endpoint remove rev is reverted from trunk now :)20:22
gundlachvishy: /bin /etc /var /boot20:23
vishysoren: ok, i cleaned up db a little bit20:23
sorenvishy: I saw.20:23
*** joearnold has joined #openstack20:24
*** burris has joined #openstack20:24
sorenvishy: I think what happened was that I did half a merge with trunk (my copy of trunk was out-of-date), and then everything just blew up in my face after that.20:24
vishygundlach, soren, eday: lets get it fixed quickly though20:24
edayvishy: thats the plan20:26
sorenSorry, what? Get what fixed?20:26
*** dizz|away has joined #openstack20:26
*** ambo has left #openstack20:26
vishysoren: nova-api-new20:26
edaygundlach, vishy: of course, this does bring up the question of priority of moving ec2 off of tornado vs just getting rackspace API bits done20:26
sorenIt's broken?20:27
* soren hasn't noticed yet.20:27
edaysoren: we just reverted the ec2-wsgi branch20:27
sorenI hate you all.20:27
vishysoren: it works for you?20:28
sorenMy unit tests work.20:28
edaysoren: unittests are fine, there is very limited testing though20:28
gundlacheday: wasn't there some argument about how moving ec2 off of tornado was required in order to serve rs on wsgi?20:28
sorenIncluding all the ones that exercise the ec2 security group api.20:28
soren...unit tests that I just added.20:28
edaygundlach: not if we run them on different ports20:28
soren...and that worked fined in the tornado days.20:28
* gundlach thought he remembered being convinced of the importance of getting this done20:29
sorenMy gripe is that I spent half a day merging this branch with trunk and now you say that I'll have to rewind this again.20:29
vishysoren: are you successfully able to use call usning nova-api-new20:29
edaysoren: anything that will issue an rpc call (run instances) will break, as well as a few things like cloudpipe not being in there. plan is to get it back in ASAP, so hopeflly won't be long20:29
gundlachsoren: you can merge it with lp:~gundlach/nova/controllers-in-api if you wish, it should be equivalent20:29
sorenSheesh, slow down, people :)20:30
sorenvishy: "use call"?20:30
gundlach== rpc.call20:30
*** joearnol_ has joined #openstack20:30
sorenOh. Maybe not.20:30
edaysoren: we need mroe tests that, for example, run a fake worker that just verifies the API sends a proper message via rpc.call20:31
*** anm_ has quit IRC20:31
sorenWhat are the current symptoms?20:31
*** anm_ has joined #openstack20:31
edaysoren: it's just due to nova.rpc assuming you are using ioloop/tornado20:32
sorenIs this carrot now playing well with eventlet or something?20:32
edaysoren: carrot will be fine, we just need to simplify it for eventlet20:32
*** joearnold has quit IRC20:33
edaysoren: just call conumer.wait rather than setup a deferred with ioloop20:33
edayconsumer.wait that is20:33
sorenSounds simple enough.20:33
edaysoren: yup.. we wanted to back out until we get this and a few other things cleaned up since vish and company need trunk working with ec220:34
edayI'm going to finish up the cloudpipe cleanup and will propose into gundlach's branch. I'll look at rpc stuff after unless someone else gets to it first20:35
sorenwhat the..20:36
sorenHow did you revert this?20:36
sorenbzr push --overwrite to the trunk or something?20:36
vishyeday, soren: we don't *need* to revert trunk20:37
sorenNevertheless, lp:nova is missing gundlach's commit from yesterday.20:38
vishywe can leave it in and just have a workaround to make nova-api still work20:38
*** jaypipes has quit IRC20:39
edayvishy: probably cleaner this way.. I think mtaylor already did it anyways20:41
mtayloreday: done20:41
mtaylorcan always be re-merged in20:41
edaygundlach: want to build off of lp:~gundlach/nova/controllers-in-api?  should merge with latest trunk (just in case) and merge your rbac branch from this morning in there20:44
*** allsystemsarego has quit IRC20:44
*** dendrobates is now known as dendro-afk20:49
soreneday: Maybe I'm doing something wrong, but it works for me :)20:50
*** hazmat has quit IRC20:50
soreneday: I just made a RunInstances call that gave me a response back.20:50
soreneday: That means it has succesfully retrieved an IP from the network worker.20:50
soreneday: Over rpc.20:50
sorenOh, that may be rpc.cast, though.20:50
*** kevnfx has joined #openstack20:51
Ryan_Laneare the ubuntu packages in ppa:nova-core/ppa recent, or fairly out of date?20:51
sorenAre the problems specific to .call() or should cast() be affected, too?20:51
sorenRyan_Lane: Recent. They're built daily.20:51
sorenRyan_Lane: Don't install them right now, though.20:51
Ryan_LaneI installed them yesterday20:51
*** dendro-afk is now known as dendrobates20:51
sorenRyan_Lane: There's a problem with the sudo integration that'll be fixed by tomorrow.20:51
*** jaypipes has joined #openstack20:52
sorenRyan_Lane: Ah. Same problem yesterday. I only just fixed it today, so will be in the package tomorrow.20:52
Ryan_Lanewell, I'm just testing right now20:52
sorenRyan_Lane: sudo is likely hosed.20:52
Ryan_LaneI'm doing everything as root, so it shouldn't be a problem20:52
Ryan_Lanesudo is hosed ;)20:52
edaysoren: just call() (anything that needs to wait for a response), but run_instances calls that and cast20:52
sorenchmod 440 /etc/sudoers.d/nova_sudoers20:52
Ryan_LaneI noticed that when I got flooded with sudo emails :D20:52
gundlacheday: sure, controllers_in_api is a fine base.  will merge from trunk and merge rbac branch.20:52
sorenRyan_Lane: ^^ fixes it.20:52
Ryan_Lanealready did that :)20:52
soreneday: Well, run_instances works for me (when I fix the kwargs deserialisation thing).20:53
soreneday: Like so:
vishycast should work fine20:56
vishysoren: call is used the first time you run an instance frome a bare intstall to set_network_host20:57
* soren blows away the db and tries again20:58
chmouelsoren: sweet i'll test that (cf: packaging branch)20:58
edaygundlach: and make sure you get a trunk without r290 :)20:59
gundlacheday: yeah, i'll merge from lp:nova20:59
*** mdomsch has quit IRC21:01
vishysoren: any ideas on the libvirt error i emailed you about?21:01
sorenvishy: I think I asked you which version you were using?21:01
sorenvishy: (Here on IRC)21:01
vishysoren: ah: libvirt 8.3 from nova-core21:02
sorenThat should be fine.21:02
edaysoren: a few other things along with the rpc bits still need some fixing, even if the tornado/ioloop does actually work still (those need to be removed). Just didn't want to keep trunk broken while we work through the issues21:02
sorenvishy: I'll have to take another look.21:02
vishysoren: ok, let us know if call is working...perhaps we can merge in your fixes and use that21:02
vishythen fix the outstanding stuff after21:03
sorenI didn't fix anything.21:03
sorenWell, nothing rpc-like.21:03
sorenJust what is in that pastebin. ( )21:03
*** sparkycollier has quit IRC21:03
vishysoren: so it may work, but your fixes21:03
sorenWith a clean db, euca-run-instances (against nova-api-new) gives me : INSTANCEi-1ami-tiny10.0.0.310.0.0.3schedulingNone (soren, None)0m1.small2010-09-22 21:01:00.87399121:04
vishyfor the value[0] thing is what i meant21:04
vishysoren: and it actually starts?21:04
sorenYeah, without it, it tries to find an instance type called "m".21:04
sorenLet me check.21:04
sorenI don't know, actually. I just saw I got an IP back which meant that rpc was working.21:04
sorennova-compute is suspiciously silent.21:05
sorenAh, wasn't running.21:06
gundlacheday: wait, the rbac branch hasn't yet been merged to trunk -- so why merge it into controllers_in_api?21:07
sorenvishy: nova.scheduler.driver.NoValidHost21:08
sorenvishy: Not sure how to interpret that.21:08
*** jdarcy has quit IRC21:08
edaygundlach: guess we don't need to, but was thinking to consolidate the ec2 changes21:11
*** krzycoder has joined #openstack21:15
*** tonywolf has quit IRC21:15
*** metoikos has quit IRC21:16
sorenvishy: Oh.21:16
*** tonywolf has joined #openstack21:16
sorenvishy: Differnet problem.21:16
sorenvishy: Timezone problem, it seems.21:16
sorenvishy: The scheduler rejects hosts that haven't had a heartbeat in the last 60 seconds. It thinks it's been two hours for my compute service. I'm two hours ahead of UTC. I smell a rat.21:17
mtaylorsoren: install ntp?21:18
sorenHow will that change anything.21:18
mtaylorsoren: might fix the 'I'm two hours ahead of ... oh, tits21:18
mtaylornevermind, totally read the words wrong21:19
sorenall you guys are behind UTC, so the scheduler thinks your last heartbeat is several hours in the future which is more recent than 60 seconds ago.21:19
mtaylorsoren: so really the scheduler should not be trying to use timezoned values21:19
sorenmtaylor: Right. I'm fixing that right now.21:20
mtaylorsoren: I will go back to lurking quietly21:20
_cerberus_Can someone help me figure out why didn't merge? All tests passed locally for me (this time :-D ) It seems to be blowing up on the LDAP stuff21:23
soren_cerberus_: Cascade error.21:23
soren_cerberus_: It's not really the ldap stuff.21:23
soren_cerberus_: The LDAP stuff is fallout from the first test failing.21:23
soren_cerberus_: it doesn't get to clean up the user it creates for testing. This makes the rest of the tests very sad.21:24
_cerberus_soren: any suggestions as to what I can do about it?21:24
soren_cerberus_: exceptions.AttributeError: 'AccessTestCase' object has no attribute 'project'21:24
sorenis your actual problem, probably.21:24
_cerberus_I'm not sure how to classify that, since it's not something I touched in my branch21:25
gundlach_cerberus_: i think i remember seeing similar errors yesterday in one of your branches -- i suggested that maybe setUp() hadn't been called.  did you ever figure that one out?  maybe it's the same prob21:25
*** cloudmeat has quit IRC21:26
_cerberus_gundlach: I never saw anything related to it, and like I said, now all of these tests pass locally21:26
_cerberus_So I'm not really sure what the issue is21:26
gundlach_cerberus_: do they still pass if you "bzr branch lp:nova trunk" and then merge your branch into your local trunk copy?21:27
_cerberus_I'll check21:28
sorengundlach: Fixed the .1, .2 stuff too.21:28
gundlachsoren: thanks -- looks like everybody's cleaning up after me today :/ :)21:28
sorenvishy: Yes, it runs the instances, too.21:28
gundlacheday: to make sure we're not duplicating effort -- i'm currently adding the metadatarequesthandler back into the api.21:29
*** scottie has quit IRC21:29
*** scottie has joined #openstack21:29
sorenIn summary, I really don't see this brokenness you all speak of.21:29
sorenI really don't enjoy how this just got yanked from the trunk. It's going to be a mess putting stuff back.21:30
sorenI want to get some sleep. This is the magic someone needs to apply somewhere for gundlach's branch to support EC2 properly:
_cerberus_gundlach: re above: yes, they pass after merging my branch back into trunk :-/21:32
gundlach_cerberus_: then i think it's time to invoke mtaylor :)21:32
gundlachmtaylor: ^^21:32
gundlachsoren: thanks and good night21:32
sorengundlach: If you add that little patch to your stuff, everything seems fine.21:32
edaygundlach: you want to apply soren's pastebin fixes to your branch?21:32
sorenI can do RunInstances and everything works.21:33
gundlachsure will do now and send to eday for review21:33
sorenAnd TerminateInstances, as it happens.21:33
sorenI needed the last bit for TerminateInstances to work.21:33
edaysoren: we're still missing some API endpoints like the metadata handlers and cloud pipe. ALso, even though the RPC code may work, it's still relying on tornado. This needs to be cleaned up properly21:33
edaysoren: we didn't want to block vish and guys from having a working ec2 api, since they need it fully functioning still21:34
soreneday: Sure, but is that a reason not to merge it?21:34
edaysoren: yes, it broke trunk, and we need to keep trunk working :)21:35
soreneday: As I said: It works.21:35
edaysoren: cloudpipe doesn't work for you21:35
soreneday: So the only downside is the fact that it hasn't completed its Tornadorectomy.21:35
gundlacheday: scratch that, i just went ahead and applied it to controllers-in-api.21:35
soreneday: Oh, perhaps not.21:35
edaysoren: as well as the metadata api requests21:35
soreneday: I only heard about the rpc stuff and I can verify that that works.21:36
soreneday: Fair point.21:36
*** brd_from_italy1 has quit IRC21:36
sorenOk, I'm off to bed.21:38
*** annegentle has left #openstack21:42
_cerberus_mtaylor: Any chance you can look at my merge problem?21:42
mtaylor_cerberus_: sure!21:43
_cerberus_mtaylor: much appreciated. All tests seem to pass locally, and blow up during the merge21:45
mtaylor_cerberus_: uh, hrm. so I don't _fully_ know - other than it seems that running the tests from your branch on the hudson machine is unhappy21:46
mtaylor_cerberus_: which could be due to library version differences or to perhaps something else weird...21:46
mtaylorlike files being left around someplace21:46
_cerberus_I added a model to the orm models, could that do it?21:46
_cerberus_That's the only possibly out of the ordinary thing (I think) I did21:47
mtaylor_cerberus_: well, for _one_ thing...21:48
soren_cerberus_: Oh. I think I know what's up with your tests.21:48
eday_cerberus_: any reason why you removed ramdisk_id?21:48
eday_cerberus_: from the file21:48
mtaylor_cerberus_: your branch contains gundlach's branch that we just removed ... :)21:48
gundlachoh, yeah, that'll do it21:49
sorenOf course it does. It was (like we're supposed to do) branched off trunk.21:49
mtayloryes. indeed.21:49
* mtaylor blames eday 21:49
sorenEveryone's checkout of trunk even still has gundlach's commit on there.21:49
gundlachi'll take the blame, for not knowing enough about nova to realize that its unit tests are incomplete.21:50
sorenIf you want to revert stuff, revert it.21:50
sorenDon't pretend it never existed :)21:50
_cerberus_Well crap21:50
soren_cerberus_: I have patches that fix your tests, I think.21:50
_cerberus_Oh, was that the thing you pasted above? Ok21:51
_cerberus_Also you fail at going to bed ;-)21:51
soren_cerberus_: Absolutely.21:51
mtaylorsoren: I applied your patch to _cerberus_ branch and it still fails on the hudson box21:51
sorenmtaylor: Which patch is that?21:51
sorenNot that.21:51
eday_cerberus_: line 238 of your merge diff concerns me (sorry for not reviewing sooner)21:52
sorenAmong other things, that branch fixes the tests.21:52
soren...and does about a million other things.21:52
eday_cerberus_: although I don't think thats your issue here21:52
*** cloudmeat has joined #openstack21:53
_cerberus_eday: I wouldn't mind an extra review21:53
_cerberus_Oh, that's the ramdisk line21:53
*** Cybodog has joined #openstack21:53
*** annegentle has joined #openstack21:54
eday_cerberus_: any reason why you removed it?21:54
mtaylorsoren: still fails with your security groups patch21:54
_cerberus_At best an accident21:54
*** ppetraki has quit IRC21:54
*** ctennis has joined #openstack21:55
sorenmtaylor: You applied the whole thing?21:55
_cerberus_eday: Yeah, I accidentally yanked that line jumping around the file21:55
mtaylorsoren: yeah. (just in a local branch on that box)21:56
* soren distinctly remembers an erorr just like that... I forget what the fix was.21:56
_cerberus_eday: fixed and pushed21:57
* soren will have to look tomorrow, if not sorted out before then.21:57
sorenI'm guessing it's got to do with an authentication failure which ends up breaking everything else due to not having cleaned up properly after itself.21:58
sorenAnyways, /me really heads bedwards21:58
*** annegentle has quit IRC21:59
*** lmcdowell_ has joined #openstack21:59
dendrobatessoren: what?  it's only midnight!22:01
*** lmcdowell has quit IRC22:03
*** lmcdowell_ has quit IRC22:04
*** lmcdowell has joined #openstack22:05
Ryan_Lanewhere does nova-volume try to store its database files? I'm getting the following error:  OperationalError: (OperationalError) unable to open database file ...22:07
*** cloudmeat1 has joined #openstack22:14
*** cloudmeat has quit IRC22:17
Ryan_Lanefor some reason, nova is writing its sqllite files at: /var/run/nova.sqlite22:19
Ryan_Lanewell, trying anyway22:19
Ryan_Lanesince /var/run isn't owned by the nova user, it is failing miserably :(22:19
*** dendrobates is now known as dendro-afk22:22
*** gustavomzw has joined #openstack22:29
*** zheng_li has quit IRC22:30
*** pvo has quit IRC22:30
*** Podilarius has left #openstack22:31
*** cloudmeat has joined #openstack22:31
*** joearnol_ has quit IRC22:32
*** joearnold has joined #openstack22:32
*** cloudmeat1 has quit IRC22:33
*** burris has quit IRC22:34
Ryan_Lanenow I see the problem. before the daemon is run from the init script, it cds to /var/run22:36 defines the sql directory to be ./22:36
Ryan_Lanethis seems like a bug22:37
*** silassewell has joined #openstack22:38
*** rnirmal has quit IRC22:40
*** hazmat has joined #openstack22:42
*** tonywolf has quit IRC22:43
*** dendro-afk is now known as dendrobates22:51
uvirtbotNew bug: #645626 in nova "sqllite databases created in /var/run" [Undecided,New]
*** scottie has quit IRC22:52
*** spackest has quit IRC22:54
*** kevnfx has quit IRC22:55
*** spackest has joined #openstack22:58
*** lmcdowell has quit IRC23:00
*** tonywolf has joined #openstack23:01
*** lmcdowell has joined #openstack23:02
*** DubLo7 has joined #openstack23:07
*** tonywolf has quit IRC23:09
*** ChrisAM` has joined #openstack23:09
*** ChrisAM has quit IRC23:10
*** ChrisAM` is now known as ChrisAM23:10
*** lmcdowell has quit IRC23:13
*** lmcdowell has joined #openstack23:13
*** howardroark has joined #openstack23:14
gundlacheday: if you're still around,
vishyRyan_Lane: looks like a bug, debian packaging hasn't been updated to respect the new code yet23:20
Ryan_Laneyeah. I entered one :) #64562623:21
Ryan_Laneit's also a problem with some of the nova-manage commands23:21
Ryan_Lanefor instance: nova-manage project zip admin admin /root/nova.zip23:22
Ryan_Lanesqlalchemy.exc.OperationalError: (OperationalError) unable to open database file None None23:22
Ryan_Laneif I cd into /var/run, that error goes away23:22
*** burris has joined #openstack23:23
vishyyeah that is nasty23:23
vishywe should have a better default23:23
gundlachvishy: mind reviewing the branch i linked to above?  i'd like to get trunk back to normal ASAP23:23
Ryan_Laneyeah. the default should be set via /etc/default/nova, or something along those lines23:24
vishysure gundlach23:24
gundlachvishy: tyvm23:24
vishyhow does it fit in with sorens branch?23:24
*** pvo has joined #openstack23:24
*** ChanServ sets mode: +v pvo23:24
gundlachvishy: mm, not sure which branch yo umean?23:30
gundlachhis 3 line change to fix
gundlachthat's already rolled into controllers-in-api23:30
edayhm, looks like cloudpipe has been broken for a while. was using methods removed in orm branch23:36
*** perestrelka has quit IRC23:40
edayvishy: I don't see a method for getting an instance by it's IP address anymore, this hidden someone or was it just removed?23:40
*** perestrelka has joined #openstack23:40
*** rlucio has quit IRC23:45
*** spackest has quit IRC23:45
*** ArdRigh has joined #openstack23:45
*** pvo has quit IRC23:47
*** jtdowney has joined #openstack23:50
*** klord has quit IRC23:52
*** ArdRigh has quit IRC23:54
*** pvo has joined #openstack23:54
*** ChanServ sets mode: +v pvo23:54
*** joearnol_ has joined #openstack23:59

Generated by 2.14.0 by Marius Gedminas - find it at!