2012/03/16/#juju.txt

[00:00] <hazmat> m_3, seems like the timeout should be 5m instead of 2m
[00:00] <m_3> that's a remaininng bug I'll push tomorrow... I adjusted manually
[00:01] <hazmat> m_3, yeah.. some of those look ok, when it goes to record the state its working already
[00:01] <hazmat> but the watcher timed out
[00:01] <m_3> the current set of runs are with the new timeout
[00:01] <m_3> they seem to be passing
[00:02] <m_3> one strange hang still... hadoop-mapreduce
[00:02] <m_3> green is wrong there...an artifact of how I killed it
[00:07]  * m_3 out... back on later
[00:10] <jcastro> SpamapS, you are awesome
=== Furao_ is now known as Furao
[04:58] <_mup_> Bug #956697 was filed: Maas provider can't start second node for deployment (CloudInitError: Incomplete cloud-init: you need to call set_machine_id) <juju:In Progress by julian-edwards> < https://launchpad.net/bugs/956697 >
=== fenris is now known as Guest98977
=== almaisan-away is now known as al-maisan
=== fenris is now known as Guest77154
=== TheMue_ is now known as TheMue
=== al-maisan is now known as almaisan-away
=== almaisan-away is now known as al-maisan
=== carif__ is now known as carif
[14:25] <marcoceppi> What does "I:" items mean in charm proof? I understand W:, etc but no I:
[14:30] <bbcmicrocomputer> what's the best way of writing a charm with a generic db relation such that it can support either mysql or postgresql?
[14:31] <bbcmicrocomputer> obviously, you could have two separate relationships defined, one for each type, but then you lose the meaning of a required db interface
[14:32] <bbcmicrocomputer> it'd be nice if you could say db interface was required in metadata but then specify which implementers can satisfy that requirement
[14:35] <jamespage> bbcmicrocomputer, hmm
[14:35] <jamespage> that would be nice but is not a current feature
[14:35] <bbcmicrocomputer> I did look at the existing charms but didn't find any that have attempted this.. yet
[14:36] <jamespage> if you want to support diff database types you will have to specify different relations
[14:36] <jamespage> I'd pick one to start with and be opinionated....
[14:36] <bbcmicrocomputer> jamespage: :)
=== al-maisan is now known as almaisan-away
[15:08] <SpamapS> marcoceppi: I is just informational.
[15:08] <marcoceppi> Ah, thank you SpamapS
[15:08] <SpamapS> marcoceppi: short for FYI ;)
[15:11] <marcoceppi> hum, not sure if this is a bug
[15:11] <marcoceppi> I added a new environment to my environments.yaml file
[15:11] <marcoceppi> The name of it is "720", which throws this error: "error: Environments configuration error: /home/marco/.juju/environments.yaml: environments: expected string, got 720"
[15:12] <marcoceppi> nevermind, I wrapped it in quotes
[15:13] <marcoceppi> Okay, back to the real problem, when I bootstrap in EC2 machine 0's state is always "not-started"
[15:13] <marcoceppi> subsequent deploys don't seem to work
[15:17] <SpamapS> marcoceppi: also you may have problems with the first char not being a number
[15:17] <SpamapS> marcoceppi: that means the agent isn't starting
[15:18] <SpamapS> marcoceppi: check the console output
[15:18] <SpamapS> marcoceppi: what version of juju are you using?
[15:20] <marcoceppi> SpamapS: this is a boostrap/deployment on an environment that's all chars and not numberic, not sure which juju version since juju --version doesn't work but dpkg says it's : 0.5+bzr470-1juju2~oneiric1
[15:21] <SpamapS> marcoceppi: ok, and if you ssh to the machine directly, does it have the same juju version?
[15:22] <marcoceppi> One sec, let me check
[15:25] <SpamapS> marcoceppi: also do check the console output of the instance. I'd bet it had problems starting the machine agent.
[15:25] <marcoceppi> SpamapS: System Log from AWS doesn't show anything
[15:27] <marcoceppi> Not sure how to connect to the machine at this point, AWS Console won't connect because it has no key-pair, juju ssh 0 says "Waiting for machine to come up" and just a straight SSH timesout
[15:42] <robbiew> bcsaller: ping
[15:42] <marcoceppi> SpamapS: Did an upgrade on my machine for juju, good to go
[15:47] <SpamapS> marcoceppi: you could have just ssh'd to the public hostname ;)
[15:47] <marcoceppi> SpamapS: Tried, it didn't accept my connections
[15:50] <SpamapS> marcoceppi: did you try euca-get-console-output ? (or ec2-get-console-output)
[15:50] <marcoceppi> Nope, I didn't know about that
[15:51] <SpamapS> marcoceppi: very important commands. :)
[15:55] <_mup_> juju/refactor-machine-agent r462 committed by jim.baker@canonical.com
[15:55] <_mup_> Fix failing tests due to refactorings in other merges
[15:58] <bcsaller> robbiew: hey
[15:59] <robbiew> bcsaller: just a friendly reminder about objectives ;)
[15:59] <bcsaller> righto
[16:03] <jcastro> jamespage, thanks for that review, it's excellent
[16:03] <jamespage> jcastro, no problemo!
[16:04] <jcastro> especially the tips to the sections of the documentation, that's clutch for new charmers
[16:07] <jamespage> jcastro, I commented on your sun-jdk/hadoop bug - what do we want todo with the existing hadoop charms for oneiric?
[16:07] <jcastro> jamespage, mims is at a conference, I was going to wait for his input
[16:07] <jamespage> jcastro, sounds sensible
[16:07] <jcastro> that charm was mostly his right?
[16:07] <jamespage> yeah
[16:08] <jcastro> I have a sneaking suspicion post-12.04 that no one will care for hadoop on 11.10.
[16:08] <jcastro> so maybe it'll just start down the road to obscurity
[16:10] <jamespage> jcastro, well the charm I wrote for 12.04 works with 11.10 as well - backported the packages I did for 11.10 in the hadoop-ubuntu PPA's
[16:11] <jcastro> ah ok, so maybe just supercede it
[16:12] <jamespage> maybe
[16:14] <SpamapS> jamespage: backporting charms to oneiric is a good idea. Though I think we'll do that quite a bit less after April :)
[16:14] <jamespage> SpamapS, most likely
[16:18] <jamespage> SpamapS, that was more of a between now and april type plan from my perspective
[16:19] <jcastro> having one set is better I think.
[16:19] <jcastro> but hey, I don't want to make claims on someone's baby when they're not around.
[16:19] <jamespage> lol
[16:20] <jamespage> I'm sure m_3 is listening...
[16:20] <jcastro> "hey so welcome back, we threw away the charm you slaved on that was our premier use case for the entire project and stuff. Hope that's ok."
[16:20] <jcastro> actually, knowing mims he's probably happy that hadoop is now a james page problem and not his, heh
[16:22] <SpamapS> "I don't always do hadoop. But when I do, I do it with Sun Java". -- James Page
[16:27]  * m_3 backscroll
[16:28] <m_3> jcastro: negronjl wrote the hadoop charms... I just got the natty packages building on oneiric
[16:29] <m_3> jcastro: I'm happy with the new charms/packages... and'd love to see the oneiric cdh pkg do an end-of-life
[16:29] <jamespage> forget that - I do it with OpenJDK
[16:30] <m_3> ha!
[16:30] <jamespage> (its the only option on armhf)
[16:30] <jcastro> negronjl, mira, how do you feel about just using jamespage's hadoop charms in 11.10?
[16:33] <m_3> OMG... finally!
[16:34] <m_3> we've got consistently reproducible full-graph charm-tests.... charmtests.markmims.com
[16:35] <m_3> hazmat: whoohoo!
[16:35] <hazmat> m_3, nice!
[16:36] <SpamapS> you guys are unbelievably amazing
[16:36] <hazmat> m_3, thats incredibly awesome, nice work
[16:36] <SpamapS> http://www.youtube.com/watch?v=Vh78T--ZUxY  <-- message from the entire juju community to hazmat and m_3
[16:37]  * m_3 is in public atm
[16:37] <SpamapS> m_3: SFW
[16:37] <SpamapS> in fact, I order you to watch it
[16:37] <SpamapS> ORDER YOU
[16:37] <jamespage> lol
[16:37] <hazmat> SpamapS, got a moment to talk about the namespace-from-env branch?
[16:37] <jamespage> nice one m_3
[16:38] <SpamapS> hazmat: I saw your comments. Very frustrating parsing effort in general and I'm more than happy to shorten that code to a few RE's if that will work.. I just didn't give it much thought after I got the tests passing ;)
[16:39] <hazmat> SpamapS, i'm split out the parsing to a separate function on your branch, it simplifies it a little, i'll  put a diff link on the merge, outside of that if you can finish up the coverage, i'm fine with it
[16:39] <hazmat> s/i'm/i've
[16:43] <SpamapS> hazmat: er, do you want to just do a new MP that supersedes imine?
[16:44] <SpamapS> hazmat: this has gotten a bit out of hand for me.. originally it was a 1 line change.
[16:44] <hazmat> SpamapS, fair enough, i'll finish it
[16:44] <SpamapS> hazmat: so at this point its easier for me to move forward with a wrapper than wait for this to pass review
[16:44] <hazmat> SpamapS, thanks for getting it this far
[16:48] <_mup_> juju/refactor-machine-agent r463 committed by jim.baker@canonical.com
[16:48] <_mup_> Expand comment on the reason for machine_id type assertion
[16:54] <_mup_> juju/robust-test-removed-service-unit r461 committed by jim.baker@canonical.com
[16:54] <_mup_> Merged upstream
[17:25] <jcastro> SpamapS, charmstests.markmims.com is back for me now
[17:27] <SpamapS> jcastro: yeah, ^^
[17:34]  * marcoceppi checks in on Minecraft charm
[17:38] <SpamapS> 2012-03-16 16:09:24,226 juju.service_watch:ERROR unit minecraft/3 startup failed install_error
[17:38] <marcoceppi> SpamapS: Yeah, saw that
[17:40] <marcoceppi> I really wish I could figure out the local provider, I still can't get it to bootstrap or deploy here
[17:42] <marcoceppi> SpamapS: have you ever seen this when deploying?  http://paste.ubuntu.com/886618/
[17:42] <SpamapS> marcoceppi: yes
[17:42] <SpamapS> marcoceppi: https://bugs.launchpad.net/juju/+bug/952397
[17:42] <_mup_> Bug #952397: Juju is forcing *ALL* charms in a repo to be perfect to be able to deploy *any* <juju:In Progress by hazmat> < https://launchpad.net/bugs/952397 >
[17:42] <jcastro> jamespage, would you consider the etherpad-lite charm complete enough to be demoable?
[17:43] <SpamapS> marcoceppi: you need to fix your charms
[17:43] <marcoceppi> ah, thanks
[17:45] <SpamapS> marcoceppi: if you have the latest charm-tools you can do 'charm update --fix path/to/a/dir' and it will pull all fixes in
[17:46] <marcoceppi> SpamapS: I do, but all the repos are setup against the old charm/* lp branches
[17:49] <jcastro> hey does any other charm other than phpmyadmin have a "get from upstream instead" option?
[17:50] <m_3> jcastro: jenkins does
[17:50] <jcastro> ta
[17:50] <m_3> rails, node, django do too...
[17:50] <m_3> but they're not in the store except for node
[17:50] <jcastro> node does?
[17:50] <jamespage> jcastro, no - its broken
[17:51]  * m_3 looking
[17:51] <jcastro> yeah I just need an example
[17:51] <jcastro> I'm working on my companion blog post to the store landing
[17:51] <jcastro> and I want to show an example of "don't want what's in the archive? Fine, here you go"
[17:52] <m_3> jcastro: yeah, node is using the pacakge... I thought I had npm updateing the node version, but it's only updating (node) package deps... not node itself
[17:53] <SpamapS> marcoceppi: --fix fixes that :)
[17:53] <jcastro> yeah I am waiting for the node ppa guy to get back from SXSW to talk to him about it
[17:53] <jcastro> hopefully he'll dig it
[17:53] <m_3> cool
[17:53] <jcastro> node is the perfect example of something that we can rock by having an up to date charm
[17:53] <jcastro> but do we have one that works now?
[17:53] <m_3> oh yeah
[17:53] <m_3> yeah
[17:53] <m_3> I used it in the mongodb demo
[17:53] <marcoceppi> phpmaydmin works
[17:53] <m_3> node talks to a mongodb replset
[17:53] <marcoceppi> but it's not a very fun charm
[17:54] <jcastro> marcoceppi, yeah I am hoping for something sexy and cloudy
[17:54] <jcastro> not to take away from your charm of course. :)
[17:54] <marcoceppi> *nod*
[17:54] <marcoceppi> naw, it's PMA, it's neither sexy or cloudy :)
[17:54] <m_3> jcastro: http://paste.ubuntu.com/886640/
[17:55] <jcastro> ok which part there does the upgraded node?
[17:55] <jcastro> do you just have that local?
[17:57] <m_3> oops, sorry that was unclear... node just uses the pacakge only... I was wrong
[17:57] <jcastro> aha!
[17:58] <jcastro> hey so I see jamespage uses "lts" and "trunk" to differentiate his versions
[17:58] <jcastro> but marco does it differently in phpmyadmin
[17:58] <m_3> the node guy should update it to get npm to update node itself to the version the app actually depends on
[17:58] <jcastro> we should probably best practice one
[17:58] <m_3> yup
[17:58] <jamespage> jcastro, those are quite specific to upstream naming
[17:58] <jcastro> m_3, yea, that's what I was thinking
[17:58] <jamespage> they have LTS and trunk releases
[17:58] <jcastro> jamespage, ah ok, so we'll likely run into service-specific conventions then
[17:59] <jcastro> but for a jenkins person that totally makes sense then
[17:59] <jamespage> jcastro, take a look at the zookeeper charm
[17:59] <m_3> although it's really pretty obvious.... might not need to be so std
[17:59] <jamespage> that uses distro OR some PPA options
[17:59] <jcastro> m_3, indeed, I am overthinking
[18:02] <jcastro> jamespage, brilliant, that's exactly what I need
[18:02] <SpamapS> actually for PPA's ..
[18:03] <SpamapS> we really need to be able to override that for system policy. I think a subordinate would be able to do that.
[18:03] <jcastro> jamespage, ok so ... you can do dev, testing, or stable, what does the command look like
[18:03] <jcastro> like, do I set a config or ... ?
[18:04] <jamespage> set a config
[18:04] <jamespage> juju deploy --config config.yaml --reposistory ....
[18:04] <jcastro>     juju set config source=dev
[18:04] <jamespage> typically my charms don't support switching
[18:04] <jamespage> i.e. that information is use in the install hook only
[18:04] <jcastro> switching is a bit too sexy and scary for now I think
[18:05] <jamespage> its one more thing to test....
[18:05] <jcastro> I think just an install switch is still way awesomer than what people have to do now
[18:07] <jamespage> jcastro, I'm going to switch the etherpad-lite charm to use chris lea's PPA for node/npm
[18:07] <jcastro> jamespage, I sent chris a mail
[18:07] <jamespage> trying to make it work from upstream git for npm is to hard and unreliable
[18:07] <jcastro> I am going to try to work with him on convincing him about the node charm
[18:07] <jcastro> everyone uses his PPA for stuff
[18:08] <jcastro> we might as well make it a nice option in the node charm
[18:08] <jcastro> that would be slick.
[18:11] <jamespage> jcastro, arggh
[18:11] <m_3> jcastro: I think the node one should be done like rails... the app specifies which version of node/rails it depends on thru package.json or Gemfile
[18:11] <jamespage> etherpad-lite won't run on 0.6 node js
[18:11]  * jamespage put this down for the week
[18:11] <jamespage> I need a beer
[18:12] <jamespage> have a nice weekend folks
[18:12] <jcastro> you've had a good week james, chill!
[18:12] <jcastro> you've earned it!
[18:12] <m_3> jamespage: I can help clean that up next week
[18:12] <m_3> night!
[18:12] <jcastro> m_3, are you at a conference still? I'd like to start running my blog post past you guys
[18:40] <SpamapS> ninjix: welcome :)
[18:40] <ninjix> hey
[18:40] <ninjix> I'm trying to do some juju with virtualized servers
[18:40] <SpamapS> ninjix: excellent. That should work. :)
[18:41] <ninjix> running 11.10 orchestra and juju
[18:41] <ninjix> getting a cannot connect error
[18:42] <ninjix> when I run juju status
[18:42] <ninjix> virtual machines were PXE installed with Orchestra
[18:42] <marcoceppi> Uh, juju local is prompting for a password when I try to juju ssh
[18:42] <SpamapS> ninjix: backup.. did you run all the steps?
[18:43] <SpamapS> marcoceppi: ssh password or sudo password?
[18:43] <marcoceppi> SSH
[18:43] <SpamapS> marcoceppi: actually it would have to be ssh password
[18:43] <marcoceppi> ubuntu@localhost's password:
[18:43] <SpamapS> marcoceppi: most likely there was a problem during master-customize
[18:43] <SpamapS> marcoceppi: though 'localhost' sounds wrong
[18:43] <ninjix> I haven't manually copied any ssh keys
[18:43] <ninjix> I thought the juju profile was doing that
[18:43] <SpamapS> ninjix: don't get confused, I'm talking to marco about ssh
[18:44] <ninjix> ;)
[18:44] <SpamapS> ninjix: for orchestra, the steps should be: 1) add VMs to cobbler (orchestra-provisioning-server) including adding them to the available mgmt class. Did you do that?
[18:44] <ninjix> SpamapS: I followed the Dustin's blog howto
[18:45] <SpamapS> ninjix: Ok, so you did 'bootstrap' and 'deploy' already?
[18:45] <marcoceppi> Found out the problem with the Minecraft charm, the "store" doesn't have the latest code.
[18:45] <marcoceppi> Scratch that
[18:45] <marcoceppi> talking out of my butt
[18:45] <SpamapS> marcoceppi: makes sense because what you just said stinks ;)
[18:46] <SpamapS> ninjix: I'd recommend getting juju from the juju PPA. The version in 11.10 is fairly buggy.
[18:46] <ninjix> SpamapS: I added the mgmt classes but only after I ran into the "no available" error
[18:46] <marcoceppi> SpamapS: did you actually merge the changes I had in to the main charm?
[18:47] <SpamapS> marcoceppi:  not sure
[18:47] <marcoceppi> nbd if you didn't I can merge it
[18:47] <marcoceppi> but the code in charms/minecraft doesn't have my changes
[18:47] <SpamapS> marcoceppi: usually the proposer, not the approver, merges.
[18:47] <ninjix> SpamapS: do the orchestra-juju-available need to be added before the cobbler install runs?
[18:47] <marcoceppi> SpamapS: I'll keep that in mind, thanks
[18:47] <SpamapS> marcoceppi: the merge proposal would be 'Merged' if the code was merged. :)
[18:47] <marcoceppi> <3 so foolish of me
[18:48] <SpamapS> ninjix: definitely
[18:48] <ninjix> ok
[18:48] <ninjix> SpamapS: do I need both *acquired and *available enabled?
=== almaisan-away is now known as al-maisan
[18:58] <SpamapS> ninjix: no
[18:58] <SpamapS> ninjix: only available
[18:58] <SpamapS> ninjix: the sequence is to set them to 'available', then 'bootstrap', then reboot the box (so it installs w/ juju)
[18:59] <SpamapS> ninjix: can you paste the HOWTO url you're using? I want to make sure thats what it says. :)
[19:01] <ninjix> https://wiki.ubuntu.com/ServerTeam/OrchestraJuju
[19:03] <ninjix> that wiki page is missing the default-series
[19:04] <SpamapS> ninjix: so right after the 'bootstrap' step, you should reboot the server that was acquired.
[19:04] <marcoceppi> SpamapS: thanks, minecraft is updated now \o/
[19:04] <SpamapS> ninjix: I just added some clarifying notes to that in the wiki page
[19:04] <SpamapS> marcoceppi: been a while since I lost myself in minecraft.. might have to go digging for lunch.. :)
[19:05] <SpamapS> marcoceppi: favorite thing to do: dig a ridiculously deep and long tunnel, then dig back up.. then find the ridiculously tall tower I built at my old base. ;)
[19:05] <SpamapS> I'm always a lot closer than I think I should be. :-P
[19:06] <marcoceppi> yeah, that's what my first base was, a giant pillar with torches so I could find it from afar :)
[19:08] <SpamapS> I can't really get into playing on big long living multiplayer servers
[19:08] <SpamapS> Every time I do.. I come back and all my s*** is gone.
[19:09] <SpamapS> even when I hide it under sand. ;)
[19:10] <ninjix> SpamapS: restarted the kvm and it was reinstalled by cobbler
[19:12] <SpamapS> ninjix: ok, it should now have zookeeper running, and 'juju status' should work
[19:12] <ninjix> just saw the zookeeper get installed on the kvm console's echo
[19:15] <ninjix> SpamapS: ahhh... just checked the firewall and a see some blocked traffic between the orchestra kvm's VLAN and the test machines
[19:17] <ninjix> SpamapS: juju status working now
[19:17] <ninjix> SpamapS: thank you
[19:20] <ninjix> going to switch over to the ppa juju now
[19:20] <SpamapS> ninjix: you're not the first to get confused by that. Hopefully the wiki update will help.
[19:20] <ninjix> SpamapS: it did
[19:21] <SpamapS> ninjix: I added default-series too
[19:21] <ninjix> good move
=== dvestal|away is now known as dvestal
[19:29] <ninjix> can juju do a master-master mysql relationship?
[19:31] <ninjix> i see in some references to slave instances
[19:32] <SpamapS> ninjix: yes it can, but there's no charm for that yet
[19:32] <SpamapS> ninjix: I'd like to add it to the mysql charm at some point
[19:32] <ninjix> nice.  Let me wrap my mind around these charms and maybe I can help out
[19:35] <ninjix> looks like I'm going need to write a Proxmox API extension for provisioning and controlling the kvm
[19:36] <ninjix> is that what the "virtualization" settings are for in the Cobbler profile?
[19:39] <SpamapS> ninjix: dunno. I believe cobbler has support for libvirt and/or kvm
[19:40] <SpamapS> ninjix: you might want to consider trying the 'local' provider which does things in LXC containers rather than on VMs controlled by cobbler.
[19:41] <ninjix> SpamapS: thanks but I am working on project to leverage our mucho macho server nodes
[19:42] <ninjix> we've been running proxmox for several years now
[19:42] <SpamapS> ninjix: ah cool :)
[19:43] <ninjix> just saw the first mysql charmed kvm get provisioned
[19:43] <SpamapS> ninjix: what exactly is proxmox? I only see a mail gateway at proxmox.com
[19:43] <ninjix> http://pve.proxmox.com/wiki/Category:Proxmox_VE_2.0
[19:44] <ninjix> SpamapS: mysql install is still pending
[19:45] <SpamapS> ninjix: yeah, we've been looking at ways to allow you to install all of the nodes in parallel before you know what you want them to do. ;)
[19:45] <SpamapS> ninjix: since the install takes a few minutes. :-P
[19:45] <ninjix> SpamapS: logged into the provisioned mysql charmed instance
[19:46] <ninjix> SpamapS: can see mysql there but the juju status still shows state: not-started
[19:48] <ninjix> SpamapS: kvm is idling now
[19:48] <SpamapS> ninjix: interesting. So is there a juju agent running?
[19:48] <SpamapS> ninjix: because it should have started up, and called back in to the zookeeper node.
[19:49] <ninjix> SpamapS: I see it running
[19:49] <ninjix> and it installed mysql
[19:50] <SpamapS> ninjix: ok, can you maybe pastebin your 'juju status' ?
[19:50] <SpamapS> ninjix: because that state should be 'running' or 'started' or something
[19:51] <ninjix> http://pastebin.com/Tz5UCMx9
[19:55] <ninjix> SpamapS: how do you restart the juju agent
[19:56] <ninjix> don't see juju in the /etc/init
[19:57] <ninjix> SpamapS: I see an error in the machine-agent.log
[19:58] <ninjix> SpamapS: but I am not sure if it occurred before I fixed the firewall rules
[19:59] <SpamapS> ninjix: if you're not on the PPA, you can't restart the agent
[19:59] <SpamapS> ninjix: the one that shipped in 11.10 just ran it as part of the first boot.. newer versions ship an upstart job for each agent
[19:59] <SpamapS> ninjix: dpkg -l juju , what does it show?
[20:00] <ninjix> SpamapS:  0.5+bzr398-0ubuntu
[20:00] <ninjix> that's from the mysql instance
[20:01] <ninjix> the orchestra machine is now running 0.5+bzr481-1juju3~oneiric
[20:02] <jcastro> SpamapS, hazmat, m_3, I have some stuff to run by you guys
[20:02] <jcastro> if you're feeling for a Friday afternoon hangout
[20:03] <SpamapS> ninjix: yeah thats the old distro version, and you can't restart the agent (and it dies a lot if your servers get overloaded because the timeouts aren't handled right)
[20:03] <jcastro> SpamapS, oh, kapil and ben are already hanging out
[20:04] <ninjix> SpamapS: so... I should reboot the mysql instance?
[20:05] <ninjix> SpamapS: or can I bang on its pid?
[20:06] <_mup_> Bug #956000 was filed: 'juju' with no arguments gives confusing message <juju:Confirmed> <juju (Ubuntu):New> < https://launchpad.net/bugs/956000 >
[20:07] <ninjix> SpamapS: killed the juju pid and restarted it with the same command line args
[20:07] <ninjix> now I see the error
[20:07] <jcastro> SpamapS, invite sent!
[20:07] <ninjix> getting "No zookeeper connection configured." along with python trace
[20:11] <m_3> jcastro: sorry, no can do atm
[20:13] <ninjix> does that mean it's having trouble talking to the bootstrap?
=== al-maisan is now known as almaisan-away
[20:16] <SpamapS> ninjix: it takes stuff from the environment too
[20:18] <ninjix> SpamapS: ok
[20:19] <ninjix> going to terminate this run and try again
[20:25] <SpamapS> ninjix: definitely get on the PPA version, much better
[20:26] <bkerensa> jamespage: I just pushed a new version of subway charm...  I have not yet had a chance to deploy myself since I need to setup LXC and I am writing this charm as a favor of sorts ;p
[20:26] <ninjix> SpamapS: do I need to change a cobbler snippet to get the instances to use PPA juju?
[20:28] <bkerensa> jcastro: you did say --repository needs to point to my charm yes?
[20:28] <SpamapS> ninjix: no
[20:28] <SpamapS> ninjix: juju pushes the commands to deploy itself every time you bootstrap or deploy
[20:34] <koolhead17|away> hi all
[20:36] <bkerensa> jcastro: http://paste.ubuntu.com/886877/ :(
[20:38] <jcastro> you want
[20:38] <jcastro> juju deploy --repository= . subway local:subway
[20:38] <jcastro> but you need to be in ~Work/Development
[20:42] <bkerensa> jcastro: juju: error: unrecognized arguments: local:subway
[20:42] <bkerensa> from ~Work/Development
[20:44] <jcastro> one sec
[20:44] <jcastro> on a call
[20:47] <ninjix> SpamapS: got a little further with this run
[20:47] <ninjix> now I see a "could not internally obtain zookeeper handle" error message
[20:48] <SpamapS> ninjix: hrm
[20:55] <SpamapS> ninjix: where are you seeing that?
[20:55] <jcastro> hmm, why would  local:subway not work?
[20:55] <jcastro> clint?
[20:55] <SpamapS> jcastro: because --repository isn't pointing at something witht default-series/subway in it?
[20:56] <SpamapS> bkerensa: juju deploy --repository=~/Work/Development local:subway
[20:56] <SpamapS> actually
[20:56] <SpamapS> bkerensa: juju deploy --repository ~/Work/Development local:subway
[20:56] <SpamapS> no =
[20:56] <SpamapS> you need to let the shell expand ~
[20:57]  * SpamapS goes to get lunch
[20:58] <ninjix> SpamapS: found it. just had to read a little further up in the trace
[20:59] <ninjix> SpamapS: DNS strikes again
[21:01] <bkerensa> SpamapS: same error
[21:01] <bkerensa> =/
[21:03] <bkerensa> jcastro: no idea why its failwhaling
[21:04] <ninjix> this is turning into a very productive afternoon. Thanks for your help, SpamapS.
[21:05] <bkerensa> indeed
[21:05] <jcastro> nor me
[21:05] <jcastro> hey, I'm going to finish off my charm and then just test yours bkerensa
[21:06] <bkerensa> jcastro: sure :)
[21:06] <jcastro> you're not the first person who can't get it working locally
[21:06] <bkerensa> feel free to propose any fixes to its bad state
[21:06] <bkerensa> :D
[21:06] <jcastro> but no worries, team effort.
[21:14] <jcastro> SpamapS, znc incoming!
[21:39] <SpamapS> bkerensa: can you do ls -l ~/Work/Development | pastebinit ?
[21:39] <bkerensa> SpamapS: http://paste.ubuntu.com/886965/
[21:45] <SpamapS> bkerensa: and ls -l ~/Work/Development/oneiric ?
[22:04] <bkerensa> SpamapS: http://paste.ubuntu.com/887000/
[22:07] <SpamapS> bkerensa: they look identical.. that can't be right
[22:07] <SpamapS> bkerensa: did you forget to add /oneiric ?
[22:08] <bkerensa> SpamapS: add it where?
[22:08] <bkerensa> it exists
[22:08] <bkerensa> ~/Work/Development/oneiric/subway is where the charm resides
[22:12] <SpamapS> bkerensa: I was asking for ls -l ~/Work/Development/oneiric
[22:12] <SpamapS> bkerensa: also is the name set to 'subway' in metadata.yaml .. that can mess it up too
[22:14] <bkerensa> SpamapS: http://paste.ubuntu.com/887014/ ^
[22:15] <SpamapS> bkerensa: ok cool, and 'cat ~/Work/Development/oneiric/subway/metadata.yaml' ?
[22:15] <SpamapS> bkerensa: you are almost as good as sshd at running my commands btw.. though I think the connection has a bit more latency than I'd like. ;)
[22:15] <bkerensa> SpamapS: http://paste.ubuntu.com/887015/
[22:16] <SpamapS> bkerensa: and in environments.yaml, do you have 'default-series: oneiric' ?
[22:18] <bkerensa> SpamapS: where is this located?
[22:23] <SpamapS> bkerensa: ~/.juju/environments.yaml
[22:23] <SpamapS> bkerensa: you wouldn't have been able to bootstrap without that
[22:25] <bkerensa> SpamapS: http://paste.ubuntu.com/887026/
[22:29] <SpamapS> bkerensa: interesting. You have two 'environments' sections. You should only have one.
[22:29] <SpamapS> bkerensa: if that fixes it, we have a bug to fix :)
[22:32] <bkerensa> SpamapS: Nope I removed the EC2 copy and tried again
[22:32] <bkerensa> http://paste.ubuntu.com/887037/
[22:34] <SpamapS> bkerensa: AHA!
[22:34] <SpamapS> provides: subway: interface: http
[22:34] <SpamapS> well, irc ate that
[22:34] <SpamapS> bkerensa: interface needs to be indented
[22:34] <SpamapS> what a terrible error message.
[22:35] <SpamapS> 2012-03-16 15:31:02,246 WARNING Charm 'subway' has an error: MetaDataError('Bad data in charm info: /home/bkerensa/Work/Development/oneiric/subway/metadata.yaml: provides.subway: expected unicode or utf-8 string, got None',) Bad data in charm info: /home/bkerensa/Work/Development/oneiric/subway/metadata.yaml: provides.subway: expected unicode or utf-8 string, got None
[22:35] <SpamapS> bkerensa: I'm opening a bug to make charm validation more friendly. :)
[22:36] <SpamapS> actually bug 899433 is already open :)
[22:36] <_mup_> Bug #899433: YAML errors in charms should be obvious to users <juju:Invalid> < https://launchpad.net/bugs/899433 >
[22:36] <SpamapS> hm no
[22:37] <SpamapS> That wasn't even logging an error
[22:37] <SpamapS> bkerensa: so, yeah, indent interface: http one more level, and it should work fine
[22:40] <_mup_> Bug #957498 was filed: Metadata errors should be user friendly <juju:Confirmed> < https://launchpad.net/bugs/957498 >
[22:51] <bkerensa> SpamapS: That fixed it
[22:51] <bkerensa> jcastro: My charm works fully now
[22:51] <bkerensa> I just deployed it succesfully so its working fine here and I just pushed the final changes
[22:51] <SpamapS> bkerensa: *woot*
[22:51] <SpamapS> bkerensa: whats your new-charm tagged bug #? I'll review it first since you've been such a good sport working this out. :-D
[22:54] <bkerensa> SpamapS: Bug #944246
[22:54] <_mup_> Bug #944246: Charm Needed: Subway IRC client/server <new-charm> <Juju Charms Collection:Fix Committed by bkerensa> < https://launchpad.net/bugs/944246 >
[22:55] <SpamapS> bkerensa: ok, I have one other thing blocking me.. then I'll start review in about 30min
[22:55] <bkerensa> SpamapS: no problem :) I have to go run errands I just wanted to sort this for review before I went out and did this stuff
[22:55] <bkerensa> :D
[23:31] <_mup_> Bug #929187 was filed: juju.control.tests.test_remove_unit.ControlRemoveUnitTest.test_zookeeper_logging_default is non-deterministic and should be removed <juju:In Progress> < https://launchpad.net/bugs/929187 >