davecheneyyay, thanks lp ,https://launchpadlibrarian.net/145125461/buildlog.txt.gz00:14
thumperdavecheney: hi00:20
thumperdavecheney: whazzup?00:20
davecheneythumper: everything OSCON related00:22
thumpereverything? wow00:23
davecheneythumper: https://codereview.appspot.com/11333043/00:23
thumperthat's a lot00:23
davecheney^ if this lands, does that mean we can say 1.11.3 will have a local providerr00:23
thumperit won't be perfect00:23
davecheneyor is the reality more nuanced ?00:23
thumperbut it works00:23
thumperthere are a few rough edges00:23
thumper1. containers don't restart on reboot yet00:24
davecheneyi ask because when we add providers to all.go00:24
thumper2. outside addresses for machines is broken00:24
davecheneythen they show up in juju init00:24
thumperit is functional00:24
thumperas it, you can bootstrap, deploy to it, and it works00:24
thumperand destroy00:24
bigjoolshow do you work out which goroutine caused the "fault" here: http://paste.ubuntu.com/5881651/00:26
thumperbigjools: the one that paniced L(00:26
davecheneybigjools: always the top one00:26
bigjoolsdavecheney: good to know, ta00:27
davecheneybigjools: it might look like they are sorted00:27
davecheneybut that is just a fluke00:27
bigjoolsnext question, WTF00:27
davecheneythe faulting goroutine is the one on the top00:27
davecheney0xb is SIGSEGV00:27
bigjoolshow readable :)00:27
bigjoolsis this a golang bug?00:28
davecheneybigjools: very likely, this was go 1.0.200:28
bigjoolson raring00:28
davecheneybigjools: that fault is a very 'fuck what happened'00:28
davecheneylast ditch error00:28
davecheneynormal nil derefs are less cryptic00:29
bigjoolssame code works on precise00:29
davecheneyprecise has a differnt version 1.000:29
davecheneybigjools: is it reproducable00:29
davecheneythe panic ?00:29
bigjoolsI believe so (no personal experience)00:30
davecheneyie, always crashes on raring/1.0.200:30
bigjoolsjust catching up with email and saw that rvb had this00:30
bigjoolsI am wondering how he built the binaries now00:30
bigjoolsie did he build on precise for precise deployment, raring for raring, etc00:31
bigjoolsis this a known bug? ERROR juju runner.go:200 worker: fatal "machiner": unauthorized00:32
davecheneybigjools: i do not think so00:33
davecheneyi haven't seen anything like that reported recently00:33
bigjoolsoh apparently Roger and William are investigating00:33
bigjoolsonly happens on saucy00:33
bigjoolsdavecheney: the branch of Gavin's that you just approved requires gwacl to be updated too, can you make that happen on the bot or do we need jam?00:39
davecheneybigjools: i have no access to the bot :(00:40
bigjoolsthis dependency situation is riduclous00:40
bigjoolsand other, better spelled, things00:41
thumperdavecheney: when is oscon?00:58
bigjoolsmuch better00:58
bigjoolsnext week00:58
thumperdavecheney: and are you talking about juju?00:58
thumperah, that's right00:58
thumperand mramm is there too right?00:58
* thumper goes to write some tests00:59
davecheneythumper: yes01:01
thumperdavecheney: is that a yes to you talking about juju?01:12
thumperwallyworld__: we are landing machines today01:32
thumpergo go go01:32
wallyworld__i only have one more to go but am making changes and may need another +1 from william01:33
wallyworld__thumper: since when did we decide to block imports? sadly go fmt doesn't know about that and so now i have to tediously inspect each file i change01:34
wallyworld__till now i've been able to rely on go fmt to Do The Right Thing01:34
wallyworld__i got a code review comment about it01:34
thumpera while back01:34
wallyworld__since go fmt didn't Do The Right Thing01:34
wallyworld__was there an email i missed?01:35
thumperit just mixes them01:35
thumperwallyworld__: probably01:35
thumperit was to the list01:35
thumperfrom jam01:35
wallyworld__go fmt puts things in alphabetical order01:35
thumperright, but it is better to have:01:35
thumperstandard libs01:35
thumperother deps01:35
wallyworld__sigh. sure. but the tooling doesn't support it :-(01:36
thumperit does sort each block01:36
thumperso it kinda does01:36
thumperit just doesn't split into blocks for you01:36
wallyworld__ok. if it sorts each block that's kinda ok01:36
wallyworld__too bad there isn't a tool like we had for lp01:37
wallyworld__to fix them up01:37
thumperyou could write one :)01:37
wallyworld__still is tedious without one01:37
wallyworld__i could in my copious spare time01:37
thumperthat's the spirit01:38
wallyworld__blood from stone and all that01:38
wallyworld__gotta sleep and eat sometimes01:38
wallyworld__first world problem i know01:38
thumperwallyworld__: hangout?01:44
thumperi'll start?01:45
thumperwallyworld__: https://codereview.appspot.com/11327043/01:57
thumperwallyworld__: and https://codereview.appspot.com/11327044/01:58
=== _mup__ is now known as _mup_
wallyworld__thumper: you owe me. by +1ing your 2nd mp, i've condemned myself to work fixing it up when my current branch lands02:38
* thumper hands wallyworld__ a beer voucher02:38
* wallyworld__ doesn't drink beer :-(02:38
* thumper hands wallyworld__ a generic drink voucher02:48
thumpervalue TBD02:48
thumperwallyworld__: https://codereview.appspot.com/11333043/ needs a look too03:20
thumperwallyworld__: it is the last03:20
thumperwallyworld__: the previous three are all approved and awaiting the merge bot03:20
* thumper does a little dance03:20
davecheneythumper: fuck year03:49
thumpera whole year of fuck?03:50
davecheney... berserker03:50
thumperwallyworld__: I'll make a branch that uses an environ watcher for the config03:50
thumperwallyworld__: instead of the env variables03:51
thumperwallyworld__: once the others are in03:51
thumperdavecheney: re golang backports03:51
thumpergpg: requesting key 448413DC from hkp server keyserver.ubuntu.com03:51
thumpergpgkeys: key B4CA2F7AA7F663D0B585869B1CB4303F448413DC not found on keyserver03:51
wallyworld__thumper: ok03:51
thumperwallyworld__: then a branch to make containers restart03:52
davecheneythumper: how do I fix that ?03:52
wallyworld__sounds good03:52
thumperdavecheney: I don't know, normally by pushing that key to the keyserver03:52
davecheneythumper: that isn't my key03:52
davecheneyi don't think03:52
davecheneyit isn't my PPA03:52
thumperyeah, I know03:52
thumpertechnically I think it should be there03:53
thumperas the ppa thing does that.03:53
* davecheney is lost03:53
* thumper looks at bigjools03:53
thumperdavecheney: got it this time03:53
thumperseems like a transient error03:53
* thumper twiddles thumbs03:54
thumpertime for another coffee I think03:54
jtvCoffee?  Good idea.03:55
thumperwallyworld__: isn't that simple actually03:55
thumperwallyworld__: as the worker doesn't know about the config internals03:56
thumperwallyworld__: I think that the environment variables are the better way to do it03:56
* thumper moves on to restarting the containers03:56
wallyworld__hmmm. should we pass config to workers03:56
wallyworld__shouldn't we03:57
wallyworld__env variables seem out of band03:57
thumperthe workers know generic config03:57
thumperbut not environment specific config03:57
thumperas that is private03:57
thumperusing environment variables is actually very common03:58
wallyworld__sure, but here we are talking pass stuff between application internal moving parts03:59
wallyworld__if our config doesn't support that it's broken imo03:59
thumperwallyworld__: they are separate executables04:00
thumperwallyworld__: in this case, the worker is a special case04:00
thumperwallyworld__: I don't advocate using this by default04:00
thumperwallyworld__: but the local provider is always going to be a bit /special/04:00
jtvHi jam — did you hear we have another gwacl update pending?04:25
thumperjtv: jam is only on for half a day04:33
thumperjam: we need to make sure someone else knows about tarmac04:33
thumperjam: and how to poke it04:33
jtvWe'll need someone to update its gwacl.04:33
thumperjam: before you go on holiday04:33
thumperdavecheney: I wish I could write a test that checked the imports of a package04:34
thumperdavecheney: can we use magic meta-shit for that?04:34
jtvOut of curiosity: what do you want to check them for?04:36
davecheney# launchpad.net/juju-core/environs/azure04:37
davecheneysrc/launchpad.net/juju-core/environs/azure/environ.go:401: not enough arguments in call to gwacl.NewNetworkConfigurationSet04:37
davecheneythe build is broken til we can land that change via the bot04:38
jtvdavecheney: yes, we were just trying to get the tarmac updated.04:38
davecheneyoh well04:38
davecheneythe recipe should work after taht04:38
jtv"oh well": EINCONGRUOUS04:38
jtvWill the upcoming Go release finally fix this just-the-tip dependencies crap?04:39
wallyworld__jtv: what, and adopt best practice used throughout the software industry? don't count on it04:41
thumperjtv: I want to make sure that when I fix agent so it doesn't depend on the world04:41
thumperjtv: that I have a test to enforce that04:42
jtvthumper: that sounds remarkably close to one of Go's stated design goals, so you'd expect there to be a tool for that.04:43
thumperjtv: hopefully04:43
thumperI know that the go doc is able to walk the package dependencies04:43
jtvThey were very very concerned about minimizing imports.  Pike says they'd rather duplicate code than import a package for just one function.04:43
thumperso I'm hoping it should be easy enough04:43
thumperpitty we don't do that04:44
jtvIIRC I saw a strings function re-implemented in the url package.04:44
thumperthe minimizing imports thing04:44
thumperobviously we do the duplication04:44
jtv"Let's do both!"04:44
jtvOne problem is, with the "unused imports and variables are errors," the Go team seem to feel they've already done a lot to fight the problem — it may have taken the wind out of the sails of further support.04:45
jtvIME, much of the weight of pointless dependencies comes from unused types, unreachable code, and poor structuring of dependencies.  The unused variables & imports are a drop in the bucket by comparison.04:46
jamthumper: wallyworld__, dimitern, and mgz all have logged into the tarmac machine before so I know they have creds. Dimitern has done gwacl updates04:46
jtvHi jam!04:46
thumperjam: awesome, is it documented anywhere?04:47
jamthumper: I've sent emails, but I don't think there is a simple wiki page sort of thing.04:47
jtvMaybe it's time for one...  There does seem to be a lot of oral tradition in this project.04:48
thumperhmm... arse biscuits04:50
thumperjust got a failure from the merge bot04:50
thumperhas to do with looking for the lxc bridge network adapter04:50
thumperwhich it obviously doesn't have04:50
thumperdouble hmm...04:55
jamjtv: gwacl is currently on rev 179, pull will bring it to 186, does that sound correct toy ou?04:58
jtvMy local one was on 179 as well actually.  Must have had a real burst of landings.04:58
jtvI'm checking the log.04:58
jtvjam: I'll try a local test run — but if that new version breaks anything else it doesn't look like a lot of work to fix it up.05:00
jamthumper: as for the dependencies stuff, you can do a few things, 1) go has a very strong ast package that lets you introspect the source, 2) the package go/build lets you introspect packages, etc05:01
jtvjam: OK — the only build failure I'm getting is the one that that pending juju-core branch fixes.05:01
jamthumper: so it is possible to do, rogpeppe has probably done some of it05:01
jtvSo please hit the update button!05:01
* thumper nods05:01
jamjtv: update has finished please land your branch05:05
jtvIt's on the way in.05:06
thumperdavecheney: I've just updated go, how do a force a rebuild of everything?05:07
thumperdavecheney: is there a go clean?05:08
jamthumper: rm $GOPATH/{pkg,bin}  ?05:08
thumperyeah, that worked05:09
jtvBetter make very very sure GOPATH is set though.  :)05:09
jamjtv: well I was meaning cd $GOPATH rm pkg/bin, but you're right you could do the rm via env expansion, but it is a bit risky05:09
thumperjam: did it manually05:09
thumperbut I got the idea05:10
davecheneythumper: jam that is the ticket05:10
thumperdavecheney: I'm about to land the branch that enables the local provider05:10
davecheneythumper: excellent05:10
jtvjam: No worries, it was clear that you were paraphrasing because there was no "-r" either :-)05:10
thumperjust had to mock out the interface functions for tests05:10
thumperit failed the first time05:11
davecheneyi'll come nagging about some word for the release notes for Evil Nick05:11
thumperpassed locally because I have a lxcbr005:11
* thumper waits the 15 minutes for the tests to run05:12
jamI believe someone just poked me, but my IRC window hung05:12
jamplease resend05:12
jtvIf it was me, it wasn't important.05:13
jamah it was just dave agreeing with me05:14
davecheneyjam: was me, was not important05:21
davecheneylets hope nobody figures out how I handle dependencies in the LP recipe, http://bazaar.launchpad.net/~dave-cheney/juju-core/package/changes05:22
davecheneybigjools will bollock me05:22
jtvTrunk is updated for the gwacl change.05:25
jtvPlease update your gwacl, everyone, before trying to build the latest trunk.05:25
davecheneywhoop whoop05:26
jtvAnyone available to review https://codereview.appspot.com/11409043 ?05:40
davecheney... value *golxc.Error = &golxc.Error{Name:"lxc-ls", Err:(*exec.Error)(0xf840169cc0), Output:[]string(nil)} ("error executing \"lxc-ls\": exec: \"lxc-ls\": executable file not found in $PATH")05:42
davecheneythis is going to be (lower case) run05:42
jtvMissing dependency?05:43
jtvthumper, have you got anything to do with that?  ^05:43
davecheneylucky(~) % lxc-ls05:43
davecheneyThe program 'lxc-ls' is currently not installed. You can install it by typing:05:43
davecheneysudo apt-get install lxc05:43
thumperdavecheney: there are a few bits around there05:44
thumperdavecheney: what in particular are you looking at?05:44
thumperI need to mock out lxc there too05:48
thumperforgot about that05:48
davecheneyjtv: reviewing05:48
davecheneyjtv: why is ec2 special ?05:49
davecheney(probably a silly question)05:49
davecheneyie, why doesn't the ec2 provider call ComposeUserData like the others did ?05:50
thumperand try to land it again...05:54
* thumper waits the requisite 15 minutes05:54
jtvdavecheney: did I forget a bit there!?05:56
jtvdavecheney: yup, I did!  But it's the exact same recipe as openstack.  Please stand by while I fix.05:57
davecheneyjtv: cool, I was wondering why ec2 was different05:57
davecheneyyou called it out in the description05:57
wallyworld__thumper: i've just removed InstanceId() from local env provider. i don't think it was actually used, right?05:58
davecheneybut that didn't marry with what I saw05:58
thumperwallyworld__: it was used in bootstrap05:58
davecheneythumper: 4 changes per hour, 24 hours per day05:58
wallyworld__even for local?05:58
jtvdavecheney: Absolutely right.  EC2 is the Ur-provider where it all starts.  I suspect I just assumed at one point that I'd done it before anything else.05:58
davecheneythe progenitor, the scion05:59
jtvThank you.  I knew there were better terms but couldn't think of any.05:59
jtvExcept Urvater, but that's not English.05:59
wallyworld__thumper: so local bootstrap does call jujud --bootstrap-state? i didn't see it using that or cloud init anywhere05:59
davecheneyjtv: ooh, nice word06:01
jtvOh good, now I have a conflict!06:01
jtvPlease bear with me.06:01
davecheneyyou need the überlagern06:01
wallyworld__thumper: in fact, local bootstrap just writes the state file directly from what i can see06:02
thumperwallyworld__: it was used in one and only one place06:02
wallyworld__so provider.InstanceId() is not used unless i am on crack and can't see it06:02
thumperwallyworld__: if you removed it, you've fixed it06:02
wallyworld__used != defined06:02
wallyworld__it was implemented06:02
wallyworld__but i was asking if anything called it06:02
jtvdavecheney: überlagern...  isn't that a verb for to spend the night in camb?06:03
davecheneyjtv: internets says it means to overlay one on top of another06:04
thumperwallyworld__: yes, one thing called it06:04
jtvAh.  Well if you're going to noun it, you'll need to capitalize it as well.  It's a bit like Go's exporting rules.  :)06:04
thumperwell it did when I grepped the code for it06:05
wallyworld__in local?06:05
wallyworld__right, i already removed that then06:05
wallyworld__thanks :-)06:05
davecheneyI can't even type the ü, i had to copy pasta it06:05
wallyworld__jtv: i'm removing InstanceId() from the Aszure provider. I'll tweak the bootstrap code to make the tests pass. tomorrow after i land can you do a live test for me?06:05
jtvwallyworld__: isn't InstanceId() required?06:06
wallyworld__jtv: not with my latest code06:06
jtvWe went through a lot of pain to implement that...06:06
wallyworld__jtv: the fact that it was there at all sucks06:06
* thumper twiddles thumbs some more06:06
thumperabout 4 minutes to wait06:06
jtvI can try to do a live test tomorrow, yes.  It'll be my first one.06:06
jtvYou're right, it was always a pain and a disappointment.  Can you do the maas side as well then?06:07
wallyworld__jtv: sorry about the implementation pain. we had similar pain with openstack06:07
thumperif it doesn't land this time...06:07
thumperit can wait for tomorrow06:07
wallyworld__jtv: maas already done06:07
wallyworld__in my branch06:07
wallyworld__tests pass fwiw :-)06:07
jtvThere's probably a lot of code that can disappear from  the Azure provider without this.06:08
jtvAll the WALA stuff AFAIK.06:08
thumperwallyworld__: btw, \o/ to removing the dumb method06:08
jtv../gwacl/storage.go:14: import /home/jtv/go/pkg/linux_386/launchpad.net/gwacl/logging.a: not a package file06:08
wallyworld__jtv: i removed some maas stuff too already06:08
jtvwallyworld__: cleanups.  Always feels good.06:09
wallyworld__yes. indeed06:09
wallyworld__that particular method was nasty06:10
jtvIt came as an unpleasant surprise...  You don't expect part of the EnvironProvider suddenly to run on the instance and have to figure out stuff it received from the Environ on the server.06:10
thumperMERGED!!!! \o/06:11
jtvdavecheney: diff should be updated...  this drives the line count for my branch even further into the negative.  :)06:12
jtvWow but the Rietveld part of "lbox propose" takes a long time.06:12
davecheneydelete everything for the win!06:12
davecheneythumper: did you do it ?06:12
davecheneydo we have an lxv provider ?06:12
thumperyes we do06:12
* davecheney golf clap!06:12
jtvThe delay is especially annoying without DCVS because you basically can't get any work done while waiting for lbox.06:13
* thumper is done06:13
thumpersee ya later06:13
wallyworld__jtv: from what i can see, you could have returned "" for that InstanceId() method cause azsure is not using cloud init06:14
wallyworld__and the only thing that used provider.InstanceId() was jujud bootstrap-state which was called from cloud init06:14
wallyworld__whereas the azsure provider is saving the bootstrap state directly06:14
jtvWait...  we didn't have an *Ubuntu  image* for Azure that supported cloudinit.  But AIUI we do use cloudinit on Azure.06:15
jtvIn some highly modified way, I'm sure.06:15
wallyworld__ah i think you are right06:18
jtvIf we'd done all that work for nothing, that would be a bit of a wet blanket.06:18
wallyworld__but i'm almost sure it doesn't require or use InstanceId()06:18
wallyworld__hence i can just delete that method, but i'll keep checking to be sure06:19
jtvWell with any luck we'll never know.  :)06:19
davecheneyawwww shit06:27
davecheneycan someone moderate my message to juju-dev https://lists.ubuntu.com/mailman/confirm/juju-dev/cffced6f12ee7bde9b7a60b938cd49ece99e601d ?06:27
jtvIt seems I'm not privileged.  :(06:31
davecheneybradm is fix0ring for me06:32
jtvbtw davecheney, did you get the diff update for the branch you're reviewing?06:32
* jtv reboots06:33
davecheneyjtv: just looking now06:44
davecheneymight have to wait a bit06:44
davecheneyabout to walk out the front door06:44
rogpeppe1mornin' all08:00
rogpeppe1i'm going to be unavailable for a little while as i'm reinstalling this laptop from scratch08:00
rogpeppe1which, hopefully, will fix lots of stuff. crossed fingers.08:00
=== tasdomas_afk is now known as tasdomas
davecheneybzr: ERROR: bzrlib.errors.InvalidHttpResponse: Invalid http response for https://xmlrpc.launchpad.net/bazaar/: Unable to handle http code 502: Bad Gateway09:04
davecheneythanks for nothing lp09:04
davecheneythat is twice today a build hsa failed becuase lp soiled itself09:04
bigjoolsdavecheney: any idea what would cause this?  we switch from 1.0.2 to 1.1 in the PPA and get failing tests: http://paste.ubuntu.com/5883536/09:11
bigjoolsthe panic is.... interesting09:12
bigjoolsor anyone?09:14
* davecheney looks09:16
davecheneybigjools: is it repeatable ?09:16
bigjoolsdavecheney: so far yes09:17
davecheneybigjools: got a branch I can poke ?09:17
bigjoolsdavecheney: lp:gwacl09:17
bigjoolswe were using 1.0 until your email09:18
davecheneyok, it's the gwacl branch09:18
TheMuebigjools: cleaned the pkg dir after upgrade and before testing?09:18
bigjoolsit bombed out much quicker when I didn't :)09:18
davecheneythe compiler would have refuse to link to old packages09:18
davecheneyit might have given a confusing error09:18
davecheneybut it would not have worked09:18
bigjoolsit looks like a bug in the 1.1 runtime09:19
bigjools[fp=0x2aaaaac6f4f8] runtime.sigpanic()09:19
bigjools        /usr/lib/go/src/pkg/runtime/os_linux.c:239 +0xe709:19
davecheneybigjools: maybe09:20
davecheneybigjools: yup, i see a panic09:26
bigjoolsdavecheney: ok ta09:26
=== ChanServ changed the topic of #juju-dev to: https://juju.ubuntu.com | On-call reviewer: fwereade | Bugs: 7 Critical, 79 High - https://bugs.launchpad.net/juju-core/
=== jtv1 is now known as jtv
davecheneybigjools: as well as the crash09:34
davecheneyare you also getting other test failures ?09:34
bigjoolswell we were but trivial09:34
bigjoolsxml formatting stuff09:34
davecheneybigjools: http://paste.ubuntu.com/5883598/09:35
davecheneythis is what I see09:35
bigjoolsdavecheney: pull again09:35
bigjoolsthey're fixed09:35
davecheneybigjools: confirmed. other failures have ceased, i'm down to the segv now09:40
rogpeppeall reinstalled, hopefully this machine will behave better now.09:41
jtvOh, was part of the failure just an omission to mangle calling conventions into the linkage?09:43
davecheney // libcurl go bingding09:49
davecheney^ what can possibly go wrong09:49
davecheneybigjools: please run go test github.com/andelf/go-curl09:51
bigjoolsdavecheney: /o\09:52
davecheney=== RUN TestCallbackFunction09:54
davecheneyunexpected fault address 0xfffff8250c8b486409:54
davecheneywell, there's your problem09:54
bigjoolsdavecheney: given the reluctance on https://code.google.com/p/go/issues/detail?id=5742 to implement renegs, I am at the point of grabbing chunks of what hair I have left09:54
davecheneybigjools: understood09:55
davecheneyi know your on that thread with gavin09:55
davecheneyplease save your pate09:55
bigjoolsbelieve me I want to - it's easier to get sunburn here09:56
bigjoolsdavecheney: how do we work out where go-curl is breaking? the trace is useless09:58
bigjoolsdebug build of Go?09:58
davecheneybigjools: psychic debugging09:59
davecheneyi started with a bias against go-curl09:59
davecheneyran their tests09:59
* bigjools ohmms09:59
davecheneysaw there was the same failure09:59
davecheneyran go test -v10:00
davecheneysaw it was related to callbacks from C to Go10:00
davecheneythen I went to get another beer10:00
bigjoolsI gather there's some game of rugby league on tonight10:00
davecheneytranslatoin: i was equal parts pesimistic and lucky10:00
davecheneybigjools: https://github.com/andelf/go-curl/issues/1510:05
bigjoolsdavecheney: I'm positive there's a solution hidden behind the, errr, whatever language that is10:06
bigjoolsgoogle translate FTW10:07
davecheneywhat is mandarin for 'eat a dick'10:08
davecheney"Suspected cgo callback go when blocked go running. Consider other ways to make trouble. . ."10:08
davecheneymy favorite is number 7 followed by a number 310:09
bigjoolsthis is kind of a blocker then.  Which pot of piss shall we consume?10:10
davecheneythat is a number 610:10
davecheneyunrelated, https://twitter.com/davecheney/status/35744301406448025910:14
davecheneyi think you mean10:17
davecheneytrolololo, boom, splash, *debris noises*10:17
davecheneyrogpeppe: still not awesome ?10:30
rogpeppedavecheney: some things are now working again10:31
rogpeppedavecheney: others are broken10:31
rogpeppedavecheney: like currently i can't raise the side menu10:31
rogpeppedavecheney: oh, now i can10:31
rogpeppedavecheney: and acme is now interpreting a left button click as a right  button click, so i can't select anything10:32
davecheneybigjools: calling in favors now10:34
TheMuerogpeppe: that really sounds evil10:38
TheMuerogpeppe: done a fresh install?10:38
rogpeppeTheMue: yes, from total scratch10:39
TheMuerogpeppe: then it's even worse10:39
davecheneybigjools: ok, sitrep10:43
davecheneythe package is broken because it depends on an exact representation of a function pointer in Go 1.010:44
davecheneythat was changed in Go 1.110:44
davecheneythe last line of that sectoin may need some revision10:45
jamdavecheney: probably no "Go" code is affected, but "cgo" code is.10:45
davecheneyjam: semantics semantics10:46
mgzjam: the bot seems unhappy? I can't ssh to it.10:46
jammgz: there are 1000s of files in /tmp/mongodb*.sock10:46
jamI'm trying to "ls" all of them10:46
jamand it is bringing the machine to a halt10:46
davecheneyjam: that is standard behavior for our test suite :)10:46
davecheneytmpwatch ftw10:46
jammgz: I sent a mail to the list, but can you check your own machine?10:46
jamI have 535 of them10:46
davecheneyi have /tmp on tmpfs10:47
mgzI have a few hundred10:47
davecheneyi konw it's time to clean them out when my machine starts to page10:47
mgzmy machine is pretty disposable though, so tmp doesn't last that long10:47
bigjoolsunreadable variable names in Go doc examples obfuscates things10:49
jtvOn an unrelated note, is Tarmac stuck again?10:49
jamjtv: it is "running" but I'm doing stuff like ls too-many-files*10:49
jamjtv: it is currently trying to merge your patch10:49
jtvGood to know, thanks.10:50
jamjtv: but it was stuck trying to run mine10:50
jamand timing out10:50
jambecause of aforementioned "too-many-files*"10:50
jamjtv: apparently the test suite leaks /tmp/mongodb-*.sock files10:50
jtvJust saw the email...  New behaviour?10:50
jammy guess being we start and stop mongo a lot and it may not clean up the sock file10:50
jamjtv: from what davecheney said, no10:50
jamjust bot has been running long enough to be a $SERIOUS problem10:50
jambot has no swap file, so it can't page anything :)10:51
jtvMaybe the test suite should run with a TMPDIR that gets cleaned up from out-of-process afterwards?10:51
davecheneyjtv: it does, c.Mkdir does that10:53
davecheneyobviously whatever is creating the mgo files isn't using it10:53
jamdavecheney: I don't think it sets TMPDIR for mongo10:53
jtvdavecheney: c.Mkdir makes a directory *in* TMPDIR, but does it set TMPDIR!?10:54
jtvIt would surprise mme.10:54
jamjtv: It doesn't set TMPDIR, but IME most tests are good about using c.Mkdir for the actual testing10:54
jtvMost.  :)10:54
jtvMy point is that the thing that runs the test suite should try to insulate itself a bit from the test suite, rather than rely on the tests' good behaviour.10:55
jtvBecause this kind of thing will just happen from time to time.10:55
jamjtv: interesting thought. The question there is: TMP, TMPDIR, TEMP, .... ?10:56
jtvThe beauty of standards...10:56
jtvI thought TMPDIR was more or less the Unixy standard, but I could well be wrong.10:57
jtvAnother option might be chroot.  Not as any kind of security measure, but to facilitate cleanup.10:57
jamjtv: I'm going to have to kill your process so that I can cleanup the machine. forgive me and please re-approve in a while10:58
jtvNo worries.  I hope it helps!10:58
jtvNo, Rietveld, this is *not* what I was asking you to do.11:00
jtvWhy does Rietveld error out so often?11:00
jtvInvalid XSRF token.11:01
bigjoolsdavecheney: feck. So 1.0.2 doesn't work with ec2, 1.1 doesn't work with Azure.  Fucking score.11:01
davecheneybigjools: yup, we hit the jackpot there11:01
hazmathow much work is implementing the tls renegotiation?11:02
jtvAny other reviewers for a cleanup job that'll eliminate some duplicate code across providers?  https://codereview.appspot.com/11409043/11:02
davecheneyhazmat: step back11:02
davecheneyis it known that TLS renegotaion is what we require ?11:02
davecheney[citation needed]11:02
jtvIt seems likely, but we're not actually certain.11:03
jtvThat's my assessment.  Anyone disagrees, I would be very very happy to hear the rest of the story.  :)11:03
* hazmat digs around for a ref11:04
jamok, we had *only* 8000 mongodb sock ifles11:05
jtvWhat a relief!11:05
davecheneyjam: i have some spare if you need 'em11:06
hazmatjtv, davecheney most of the refs seem to refer to our own usage, or the renegotiation attack (for which recommends are disable renegotation or augment with extension from rfc 5746)11:10
* jam heads to put gas in the car, I should be back in time for standup11:11
jtvYes... we figured that was one of the reasons why the Google folks thought renegotiation was too messy to implement.  But the question was really: is renegotiation really what's needed to solve this problem?  Or is the renegotiation only happening to work around an avoidable condition?11:11
rogpeppeweird, X is sending mouse events with Mod1Mask set.11:13
TheMuejtv: you've got a review11:14
TheMuejtv: a real good cleanup11:14
TheMuejtv: Bitteschön11:14
jtvAh yes, forgot about the underscore...11:14
dimiternjam: btw i noticed that in addition to mnogo-*.sock files (1751 before i removed them) my /tmp contained also quite a few test-mgo* subdirs11:15
wallyworld__dimitern: mgz: i will missing the meeting. HUGE game of football on right now11:15
dimiternwallyworld__: enjoy ;)11:15
wallyworld__i am :-)11:15
wallyworld__we are winning but only just11:16
hazmatif i had to get a working version of juju-core for a demo with openstack today. what's the recommend? go 1.1 rel branch + juju + trunk?11:17
mgzhazmat: what's in saucy should be fine11:17
hazmati'm on precise for this11:18
mgzthen you can download the tarball and build with the go backport, also the devel ppa is probably okay11:20
hazmatmgz, thanks.. i'll try tarball with a dl of go1.1.111:22
mgzppa:juju/golang should be fine for you11:24
hazmathmm.. hasn't juju init been around for a while.. a build of the tarball says its not a valid command  http://pastebin.ubuntu.com/5883893/11:33
mgzhazmat: what does `juju version` say?11:35
hazmatdoh.. pyjuju11:36
hazmatdoing  pyjuju for maas -> openstack and then juju-core for openstack workload..11:36
hazmatmgz, is there a way to get the openstack provider to ignore invalid certs? http://paste.ubuntu.com/5883925/11:50
mgzho ho ho11:51
mgzI think that's another thing we didn't port over from pyjuju11:51
* hazmat files a bug11:51
mgzrogpeppe: ^have a look in a sec11:52
hazmatits bug 1202163 for ref12:03
_mup_Bug #1202163: openstack provider should have config option to ignore invalid certs <juju-core:New> <https://launchpad.net/bugs/1202163>12:03
rogpeppehazmat: you can configure the http stack to always ignore invalid certs, or ignore them for just a single request12:05
rogpeppehazmat: for the former, you can set http.DefaultClient.Transport to &http.Transport{TLSClientConfig: tls.Config{InsecureSkipVerify: true, (maybe more here)}}12:06
rogpeppehazmat: for the latter, you can create a new http.Client and call Do (or Get or whatever) on that12:07
* rogpeppe goes for lunch12:07
hazmatrogpeppe, thanks12:08
jamjtv: your maas loggo patch has landed.12:13
jtvYup, just saw it thanks.12:13
jtvDoes that mean I can land the next one too?12:13
jamjtv: there is currently a fair backlog of things to land, but you should be able to mark it approved and the bot will get to it12:17
jamto whom it may concern, we should add to the release notes that we can upgrade a 1.10 deployment to 1.11.313:10
jamBut I couldn't find the 1.11.3 release notes in my quick search13:10
mgzjam: dave linked them in his message in the "local provider revies" thread13:14
=== tasdomas is now known as tasdomas_afk
=== TheRealMue is now known as TheMue
gary_posterevilnickveitch, arosales, jcastro this is the proposed replacement for the hooks tab.  AOK?  http://ubuntuone.com/1jW3lOgQLDNRup1bOX5xWf16:22
jcastrogary_poster: so "source" instead of "hooks"?16:24
gary_posterjcastro, yes per evilnickveitch bug 120184016:24
_mup_Bug #1201840: hooks section lists more than just hooks <charmbrowser> <juju-gui:Triaged by lucapaulina> <https://launchpad.net/bugs/1201840>16:24
evilnickveitchgary_poster, cool, where does that link go to?16:34
evilnickveitch(the Contribute to this charm one)16:34
gary_posterevilnickveitch, https://juju.ubuntu.com/docs/authors-charm-store.html16:35
gary_posterevilnickveitch, AOK?  need to run16:35
evilnickveitchgary_poster, yeah, that's fine16:36
gary_postercool, will land after lunch16:36
arosalesgary_poster, taking a look16:37
arosalesgary_poster, my only fear is that is is "different" that what we have from the existing charms16:38
marcoceppigary_poster: what about just "Files"16:41
jcastroI would just call it "Code"16:44
jcastrocharms are code, so just call it code16:45
arosalesgary_poster, talking with eco team16:45
arosalesgary_poster, seems it is ok to rename that tab to something else, consensus is looking like "code" but we wouldn't be heart broken if you kept source16:45
jcastrogary_poster: what do you think about having the Juju home button go to the browse mode and not the canvas?16:52
_mup_Bug #1202306: We need an "all" category <juju-gui:New> <https://launchpad.net/bugs/1202306>16:56
rogpeppervba: which revno of gwacl should we be using currently?17:17
rogpeppervba: sorry, ignore me, i hadn't pulled17:19
rogpeppedone for the day. see y'all tomorrow.17:36
=== benji___ is now known as benji
marcoceppiHey guys, compiling from source, but I already compiled juju-core a while ago. So I ran `got get -u launchpad.net/juju-core/...` which exited 0, ran go install -v launchpad.net/juju-core/... no output exit 0 but my GOPATH/bin/juju has the old version still19:14
marcoceppianything I'm missing?19:14
ahasenackmarcoceppi: try deleting it19:20
marcoceppiahasenack: cool, I was just doing that actually19:21
marcoceppiahasenack: http://paste.ubuntu.com/5885337/ libcurl3 is installed on the system, ran apt-get dep-build for juju-core as well same response19:36
ahasenackmarcoceppi: hm, no clue, I run go get -v -u everyday19:37
ahasenackand ran it just now19:37
ahasenackmarcoceppi: trash $GOPATH maybe19:37
marcoceppiahasenack: I did, found the issue. libcurl4-gnutls-dev was needed19:37
marcoceppiahasenack: thanks!19:38
ahasenackmarcoceppi: have you tried lxc? I just get "connection refused" when I bootstrap, but telnet to that host/port shows no problems19:38
ahasenackit's the mongo port19:38
marcoceppiahasenack: I just got it compiled, about to give it a go19:38
ahasenackI get19:38
ahasenack2013-07-17 19:36:23 ERROR juju open.go:89 state: connection failed, will retry: dial tcp connection refused19:38
ahasenackand tcpdump indeed shows a syn and a rst19:38
ahasenackbut when I telnet to the same address and port, I get a connection established19:39
ahasenackso no clue what is going on19:39
ahasenackI also don't know which mongo to use, I grabbed the raring one19:39
ahasenack2.2.4, seems to be listening in ssl mode at least19:39
ahasenackmarcoceppi: it also complained there was no secret in the env, the juju init template for local: needs admin-secret it seems19:40
marcoceppiahasenack: I just copied the one from cheney's email19:41
ahasenackmarcoceppi: it's the same as juju init, let's see if it will complain in your case19:41
marcoceppiahasenack: ah, yeah it complains19:43
marcoceppimakes sense since that's needed for the api19:43
ahasenackmakes you wonder if they really tried it ;)19:44
ahasenackthen you need sudo19:44
ahasenackand install mongodb19:45
ahasenackand that's where I'm stuck now, bootstrap fails to connect to mongo, even though it's running19:45
marcoceppiahasenack: well, it's a first cut, I'm not expecting perfection. I'm here to report bugs!19:46
marcoceppitechnically you had to use "root" to bootstrap the last local provider19:47
marcoceppiit just prompted for sudo password during bootstrap19:48
marcoceppiahasenack: seems to have worked for me, waiting for status to return19:49
ahasenackmarcoceppi: what did you get for bootstrap -v?19:49
marcoceppiahasenack: http://paste.ubuntu.com/5885365/19:49
marcoceppiahasenack: I didn't -v that command :(19:50
ahasenackmarcoceppi: ok, that it was probably full of connection refused too19:50
ahasenackbecause it exits silently19:50
marcoceppiI'll destroy and try again in a second19:50
ahasenackthen status shows what you pasted for me too19:50
marcoceppiwell I'm getting connection estabilished19:50
marcoceppifinally errors with 2013-07-17 19:50:38 ERROR juju supercommand.go:235 command failed: cannot log in to admin database: auth fails19:51
marcoceppierror: cannot log in to admin database: auth fails19:51
marcoceppimongodb is 1:2.2.4-0ubuntu119:51
ahasenackso, the mongodb package starts mongo19:52
ahasenackjuju creates another upstart job19:52
ahasenack /etc/init/juju-db-<something>.conf19:53
ahasenackI think it starts mongo on another port, so it doesn't conflict19:53
ahasenackthat job starts a mongo on port 3701719:53
ahasenackwith ssl19:53
ahasenackmarcoceppi: did you try bootstrap -v?19:53
marcoceppiahasenack: bootstrapping now19:53
marcoceppiahasenack: yeah, getting conn fail19:54
ahasenackand my juju status also fails with a ton of "connection established"19:54
ahasenacklike several per second19:54
* marcoceppi changes port19:54
ahasenackI stopped the main mongodb service19:54
ahasenackso just the one started by juju is in play19:54
ahasenackeven it I let it run, it's on a different port19:55
ahasenacktcp        0      0 *               LISTEN      10851/mongod19:55
ahasenackhm, wait19:55
ahasenackit didn't start19:56
ahasenackthat's the one from juju19:56
marcoceppiahasenack: well we're in the same boat now19:58
ahasenackso when juju's mongo is running, the packaged mongodb won't start, I don't know why19:59
ahasenacknot sure if it's even relevant19:59
marcoceppiahasenack: it looks like juju mongo runs both 38017 and 3701720:00
ahasenackmarcoceppi: maybe one is the web admin port20:00
ahasenackI'm trying ssl on it, but chrome is being stupid20:00
ahasenack"You attempted to reach localhost, but the server presented an invalid certificate."20:01
ahasenackand doesn't allow to continue20:01
ahasenackok, firefox works20:01
ahasenackso 38017 is some sort of web status page20:01
ahasenacklike apache's /server-status20:01
ahasenack37017 is the client port20:02
ahasenackwhat clients should use20:02
ahasenackit shows some logs, that's good20:02
ahasenack           17:02:12 [conn26]  authenticate db: admin { authenticate: 1, nonce: "dd1208d655bae7f3", user: "admin", key: "770f106419a82fc861b9d03103b0cda6" }20:02
ahasenack           17:02:12 [conn26] auth: couldn't find user admin, admin.system.users20:02
ahasenackso something needs to create that user somehow20:02
ahasenackmaybe that's the bootstrap step that failed20:02
ahasenack(this was as a result of juju status)20:02
ahasenackI think it's a timing issue20:05
ahasenackit starts juju-db-andreas-local (upstart job)20:05
ahasenackand I kept trying that port in another terminal, also getting connection refused20:06
ahasenackafter a while, it worked20:06
marcoceppiahasenack: yeah, so I'm running Umongo against it, it seems like core stops trying to connect too soon20:06
marcoceppiabout 45 seconds after bootstrap stops trying to connect, it works20:06
ahasenackit's an aggressive retry even20:07
marcoceppithis is a little disappointing20:07
ahasenackneed to stash a sleep in there somewhere, or increase the number of retries20:08
ahasenacki counted 3820:08
marcoceppiI'm going to try to piss it off, see if I can past the retries20:11
ahasenackmarcoceppi: worked here, I increased the timeout 10x20:16
ahasenackmarcoceppi: https://pastebin.canonical.com/94586/20:16
ahasenackjuju status works now20:17
marcoceppiahasenack: brilliant20:18
marcoceppiI need to figure out how to patch that in my install20:18
ahasenackI ran go install launchpad.net/juju-core/... after that, it rebuilt the binary20:18
marcoceppiahh, just patch it in the src20:19
marcoceppiahasenack: I ended up putting it to 60, a minute seems like a fair amount of time20:21
marcoceppiahasenack: have you opened a bug yet?20:21
ahasenackso far it's that (too short a timeout) and the admin-secret missing from the template20:22
ahasenackdoing a deploy now, I see wget action in the process list20:22
ahasenackcontainer running20:23
marcoceppiwow it's fast as hell20:24
marcoceppiIt's a shame you still have to download the cloud image on first run20:26
ahasenackhow big is that?20:26
marcoceppiI thought there was going to be a way to seed the cloud images before dpeloyment20:26
marcoceppiit's used to build the template, so once it's downloaded (as it is during the first deploy command) all furture deploys are fast because it's cached20:27
ahasenackthat's good enough20:27
marcoceppiin pyjuju the cache was cleared during each destroy, I wonder if it'll keep this cache around longer20:27
jcastroI remember bashing out the workflow with thumper on the cloud image sync20:29
jcastroand we asked specifically for a sync-like command to preload the image20:29
ahasenackanyway, worked, got wordpress up (with mysql) on lxc20:33
ahasenackgot a 502 from nginx20:33
ahasenackwill debug later20:33
jcastromarcoceppi: thumper is online in a few, I'm going to snag him on G+ and ask a buncha questions20:38
marcoceppijcastro: excellent20:38
marcoceppiahasenack: got it running over here20:39
* marcoceppi shrugs20:39
marcoceppijcastro: dude, local provider is awesome already. But I too have a few qs for thump20:40
=== tasdomas_afk is now known as tasdomas
jcastrohey do we still need sudo? maybe we can answer that dude's question20:42
marcoceppijcastro: you need sudo for local juju-core provider20:43
marcoceppito bootstrap and destroy20:44
marcoceppijcastro: but I don't think his question is about using juju local provider20:44
marcoceppithumper: \o/20:56
thumperhi marcoceppi20:57
=== tasdomas is now known as tasdomas_afk
marcoceppithumper: the other thing that ahasenack found while using local provider is admin-secret is not put in the local privider init template. Would you like me to open a bug for that as well?21:47
thumpermarcoceppi: yes please21:47
thumperI'll look to address ASAP21:47
marcoceppithumper: ack21:47
ahasenackthumper: congrats thumper, consider me impressed21:47
thumperahasenack: uh... ok21:47
thumperjust doing my job :)21:47
hazmatthumper, local provider is pretty nice22:51
thumperhazmat: thanks22:51
* thumper is working on surviving reboot22:51
thumperauto-restart containers22:51
sidneithumper: there was a question from marcoceppi earlier about why go get -u + go install doesn't update the built version. i have the same question. :)22:53
sidneii still have 1.11.1-saucy-amd64 after doing that22:54
thumpersidnei: I'm not entirely sure I understand your question22:54
jcastrothumper: I like that you're swamped with people asking about local. :D22:55
thumperperhpas because GOBIN isn't in your path?22:55
jcastrothis is a pretty awesome milestone dude22:55
thumperjcastro: shouldn't you be not working/22:55
sidneithumper: did you mean GOPATH?22:55
jcastroI'm back22:55
jcastrogotta play with local. :D22:55
thumperGOPATH/bin even22:56
thumperthere are several different go env vars22:56
sidneiit's there yeah22:56
sidnei$ which juju22:56
sidneithe timestamp is fine even22:57
sidnei$ ls -la /home/sidnei/src/go/bin/juju22:57
sidnei-rwxr-xr-x 1 sidnei sidnei 13907104 Jul 17 17:05 /home/sidnei/src/go/bin/juju22:57
sidneiso i guess it got updated after all, just reports an odd version?22:57
sidneiuhm, no, it just got rebuilt but still no local22:58
thumpersidnei: haha22:59
thumperI know what it is22:59
thumpersudo ~/go/bin/juju bootstrap --debug22:59
thumperyour GOPATH isn't in root's path22:59
thumperthis goes away when we have it packaged22:59
sidneiuhm, i don't think that's it23:00
sidneijuju init | grep local doesn't show anything23:00
sidneiso it smells like it wasn't updated after all23:00
sidneisource is at the right revno23:01
sidneisidnei@sidnei-laptop:~/src/go/src/launchpad.net/juju-core$ bzr revno23:01
thumperthat isn't tip23:02
thumpernot even close23:02
thumperbzr info?23:02
sidnei  parent branch: http://bazaar.launchpad.net/~juju/juju-core/trunk/23:02
sidneiseems like the branch changed location23:03
thumpersidnei: bzr pull lp:juju-core --remember23:03
thumperthat'll use bzr+ssh as well23:03
sidneifun that go get -u didn't think of looking at the new location23:04
sidneii guess it just did a pull locally23:04
sidneithanks thumper! eager to try local :)23:06
* sidnei dinners23:06
* marcoceppi continues to throw papercut fixes23:10
davecheneyhow the bloody hell am I supposed to do this ?!?23:16
* davecheney throws table23:16
davecheneybigjools: jtv how do you create an x509 cert for azure with openssl, is it possible ?23:26
davecheneycan you get ms to create one for your ?23:26
=== flaviamissi is now known as flaviamissi_
sidneithumper: 2013-07-17 23:45:58 ERROR juju runner.go:200 worker: fatal "machiner": unauthorized23:47
sidnei am i missing something?23:47
davecheneyjtv: bigjools oh hey, it's in the README, thanks heaps!23:47
davecheneysidnei: nope, sounds like a bug23:49
sidneidavecheney: this is with the local provisioner fwiw, i saw a mention of missing settings in the juju init -spitted template23:50
davecheneyshit you guys are keen23:50
davecheneythat code only landed yesterday23:51
sidneicouldn't help it, im subscribed to commits :)23:54
sidneiit keeps looping around with: 2013-07-17 23:45:58 ERROR juju runner.go:200 worker: fatal "machiner": unauthorized23:54
sidnei2013-07-17 23:54:14 INFO juju runner.go:246 worker: restarting "state" in 3s23:54
sidneilooks like it might fill the disk if i leave it :)23:55
* thumper poks23:57
thumpersidnei: where are you seeing that?23:57
thumpersidnei: mine seems to be working fine23:57
sidneithumper: locally here, in ~/.juju/local/log/machine-0.log23:58
sidneii ran: sudo /home/sidnei/src/go/bin/juju bootstrap -v -e local23:58
thumperhmm, I don't get that23:59
thumpersidnei: missing the secret?23:59
thumperI have to fix that23:59

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!