[02:49] <Valduare> hi guys im getting some odd behavior with juju and add-machine
[02:52] <mbruzek> Valduare, specifically what kind of odd-ness?
[02:52] <Valduare> ERROR lookup juju-openerp: no such host
[02:53] <Valduare> this is after I destroyed environment
[02:53] <Valduare> and am re-using the same vm’s
[02:53] <Valduare> I ssh’ed into the vm and rm the juju stuff there as per the wiki
[02:55] <mbruzek> I have not done the manual provider personally but I can take a shot here.  When you destroyed the environment Juju might not have cleaned up all the way.  Can you try to destroy-environment with the --force option?
[02:55] <Valduare> I have
[02:59] <mbruzek> What version of Juju are you using?
[02:59] <Valduare> 1.17.7
[03:00] <mbruzek> Yeah that is the latest.
[03:01] <mbruzek> I am sorry Valduare you seem to have stumped me.  I have a cleanup script for when the local provider does not destroy properly but I don't think that will help you in this case.
[03:02] <Valduare> hmm
[03:05] <Valduare> its not anything with the vm’s cause i’ve blown them away and reinstalled fresh servers
[03:06] <Valduare> its something on my workstation -
[03:06] <Valduare> and juju
[03:06] <mbruzek> You mentioned you were following the wiki page which page?
[03:06] <Valduare> the one on manual provisioning
[03:08] <mbruzek> What is your bootstrap-host value ?
[03:08] <Valduare> http://askubuntu.com/questions/433842/how-to-resolve-error-machine-is-already-provisioned-in-manual-provision-set-up   thats how I cleaned up one of the vm’s im trying to re-add in this new juju bootstrap environment
[03:08] <Valduare> its an ip address
[03:08] <Valduare> it bootstraps fine
[03:09] <Valduare> i have all of the vm’s setup with one of my public ip’s and diffeent ssh ports right now
[03:10] <mbruzek> All machines added with juju add-machine ssh:... must be able to address and communicate directly with the bootstrap-host, and vice-versa.
[03:11] <mbruzek> When you ssh to your machines.  you can see your workstation?
[03:11] <Valduare> yes
[03:17] <Valduare> i’ll have to figure this out tomorrow
[03:17] <Valduare> g'night
[03:17] <mbruzek> Sorry Valduare
[06:52] <rogpeppe1> mornin' all
[06:52] <rogpeppe1> axw: about "call State.SetAPIAddresses in bootstrap-state command"
[06:52] <rogpeppe1> axw: there are two reasons we want to do this
[06:54] <rogpeppe1> axw: firstly, and most pragmatically, we won't be able to enable the peergrouper worker for a little while, and having valid API hostports in state is a prereq for other branches
[06:56] <rogpeppe1> axw: secondly, if we rely on the peergrouper worker to do it, there's a window at bootstrap time when there are no API addresses, so clients that connect early won't get them
[07:14] <axw> rogpeppe1: sorry, was picking up my daughter
[07:14] <rogpeppe1> axw: np
[07:15] <rogpeppe1> axw: in #juju by accident. #juju-dev probably more appropriate
[09:44] <lseror_> Hi everyone
[09:45] <lseror_> I am Laurent and I am in charge of Outscale who is a Cloud Provider (IaaS)
[09:45] <lseror_> I just install juju and I want to test it
[09:45] <lseror_> thanks for the work, the website is great
[09:46] <lseror_> I notice that some configurations seems to be inside the juju binary, like aws region address
[09:47] <lseror_> I think that it would be cleaner to put that kind of informations in another file, with an editable format
[09:47] <lseror_> like json
[09:48] <lseror_> I did not find any forum to discuss about it but I subscribe to the ML and will use it
[09:48] <lseror_> I am just fishing here for some comments
[09:48] <ashipika> look at .juju/environments.yaml
[09:48] <ashipika> most stuff should be configurede there
[09:49] <lseror_> not the aws regions
[09:49] <lseror_> I check
[09:49] <lseror_> you can tell the regions
[09:49] <lseror_> but not the url
[09:49] <lseror_> you can put us-east-1 by example
[09:49] <lseror_> but not the corresponding utl
[09:49] <lseror_> but not the corresponding url
[09:50] <lseror_> that is in the juju file (I strings it)
[09:50] <ashipika> oh, ok.. i see..
[09:50] <lseror_> it should have https://ec2.us-east-1.amazonaws.com as a parameter somewhere
[09:50] <ashipika> sorry
[09:50] <lseror_> no problem
[09:50] <lseror_> I am here for discussion
[09:50] <lseror_> ;)
[13:42] <weblife> say if I connect an elastic ip though aws to one of my machines managed by juju. Would there be a way to retrieve that newly attached external ip because unit-get public-address pulls the dns address (ie. ec2-54-215-195-84.us-west-1.compute.amazonaws.com) ?
[13:50] <Valduare> morning guys
[13:53] <jcastro> hi!
[13:56] <mbruzek> Good morning Valduare
[13:56] <mbruzek> Valduare, I was thinking about your problem last night.  Can you run the failing command with --debug and pastebin the results?
[13:59] <Valduare> sure
[14:00] <Valduare> it says it completes if I remember right though
[14:02] <Valduare> oh that was different debug I was thinking of
[14:02] <Valduare> one sec pastebinning
[14:04] <Valduare> http://pastebin.com/qhK1HJ9Y
[14:04] <Valduare> stripped ip and usernames out etc
[14:15] <mbruzek> Valduare, The message "ubuntu user is already initialised" makes me think there is a problem on the machine
[14:15] <Valduare> I just removed the user and removed the home dir
[14:16] <Valduare> juju had set it up on a previous run
[14:16] <mbruzek> did that help?
[14:17] <Valduare> no
[14:17] <Valduare> what I dont understand, why it asks for password after i coppied my key over already
[14:17] <Valduare> I can ssh without password to these hosts just fine
[14:18] <mbruzek> Someone from juju-core should take a look, maybe you can post this link in #juju-dev
[14:30] <themonk> marcoceppi: hi
[14:31] <themonk> marcoceppi: i am in problem with relation-set and relation-get
[14:32] <themonk> marcoceppi: need help
[14:32] <mbruzek> themonk:  What is the problem?
[14:35] <themonk> mbruzek: hi thanks, problem is in provider charms joined i am setting some values using relation-set but not getting it requirer joined/changed using relation-get
[14:38] <mbruzek> themonk, sometimes the order of things is important.  For instance sometimes the mysql charm does not set all the values so consumers of the mysql relation need to check if all values are there before they process the db relation.
[14:38] <themonk> mbruzek: may be i am missing some thing here, can u tell me when to set value and get value like joined/changed
[14:40] <mbruzek> themonk, so for instance if relation-get database is not set the related charm can not progress so some people put an exit 0 in their relation code and wait until the mysql relation returns a database
[14:40] <mbruzek> if [ -z "${db_db}" ]; then
[14:40] <mbruzek>   juju-log "No database information sent yet. Silently exiting"
[14:40] <mbruzek>   exit 0
[14:40] <mbruzek> fi
[14:40] <mbruzek> Where db_db=`relation-get database`
[14:41] <jose> that needs to be on a db-relation-changed, as it'll run once the relation data changes
[14:42] <mbruzek> jose, you are correct that is where I pulled it from.
[14:43] <mbruzek> So themonk may be getting timing issues, where the relation-set and relation-get are happening out of order
[14:43] <jose> yep, that should fix it :)
[14:44] <themonk> mbruzek: hmm, that possibility is high :)
[14:45] <mbruzek> themonk, Use juju-log to print out the data from the relations and then exit if it is not all there.
[14:45] <themonk> mbruzek: can you tell when the relation callbacks called and there order (joined,changed,departed,broken) doc is not good enough in jujudoc
[14:47] <mbruzek> *relation-joined is when a connection has been established, *relation-changed is when data on a relation changes, *relation-departed is called when you still have a connection but it is going away, and *relation-broken is called when there is no more connection.
[14:48] <mbruzek> Remember you must make he hooks idempotent because these hooks can run more than once in the lifetime of a charm.
[14:49] <themonk> mbruzek: hmm, can you point me a detail youtube video, blog or anything
[14:52] <mbruzek> themonk, have you already read this: https://juju.ubuntu.com/docs/charms-relations.html
[14:54] <themonk> mbruzek: yes
[14:54] <mbruzek> OK looking for more specific content
[14:57] <themonk> mbruzek: if i check inside *relation-changed that is data changed then my question is, is juju periodically calls it?
[14:57] <themonk> mbruzek: sorry for bad english :)
[14:58] <mbruzek> Juju will not call it randomly.  It gets called when information on that relation changes
[14:58] <jose> you need to consider that information may change, so you may have to update it at any point
[14:59] <jose> make sure you take your precautions when writing the code
[14:59] <mbruzek> themonk, jose is correct.  Here is a youtube video that talk about relations http://www.youtube.com/watch?v=yRcqSjOGweo
[15:02] <themonk> mbruzek: Ok. what i am doing is this, my provider charm *relation-joined is setting some data then requirer charm *relation-changed should called by juju automatically and my method will check the data and make changes, is it ok?
[15:03] <mbruzek> themonk, Yes that is the pattern.
[15:08] <themonk> thanks jose and mbruzek
[15:09] <themonk> mbruzek: i am watching it now
[15:12] <weblife> say if I connect an elastic ip though aws to one of my machines managed by juju. Would there be a way to retrieve that newly attached external ip because unit-get public-address pulls the dns address (ie. ec2-54-215-195-84.us-west-1.compute.amazonaws.com) ?
[15:22] <mbruzek> weblife, I don't know of a way to do that.  Could you find the ip by using dig or some tool like that?
[15:25] <mbruzek> our use route53 to look it up?
[15:33] <weblife> mbruzek Okay thank you.
[15:37] <mbruzek> weblife, actually nslookup might give you a better parsable answer
[15:38] <mbruzek> weblife  nslookup ec2-54-215-195-84.us-west-1.compute.amazonaws.com
[15:45] <weblife> thanks good idea
[15:45] <weblife> try it out in my next launch
[16:46] <bloodearnest> anyone know a way to fake a public ip for a unit with the local provider?
[16:48] <Flint_> hi mbruzek
[16:49] <mbruzek> Hello Flint_
[16:50] <Flint_> just checking connections, seems to only work on one of my 4 computers...
[16:51] <mbruzek> Flint_ Well I can see you there so one of the works.
[16:51] <Flint_> btw juju brings up some interesting photos by default
[16:51] <mbruzek> Yes I have seen some of those
[16:52] <mgz> beware the image search
[17:01] <timrc> marcoceppi, FYI, upgrading to 1.17.7 fixed the issue I had with upgrade-charm where it was losing the service config values
[17:09] <tvansteenburgh> inside of a config-changed hook, i'd like to check "am i in a mongodb relation, and if so, what's the ip and port of the remote side?" is this possible with charm-helpers?
[17:10] <tvansteenburgh> there are lots of hookenv.relation_* functions...trying to figure out which ones i need for this (if it's possible)
[17:14] <jose> tvansteenburgh: what I could do in that case is check if touch a file called '.mongodb-configured' once configuration is done, and then check if the file's there, maybe there's a charm helper that can do that more easily
[17:15] <tvansteenburgh> looks like i could borrow the code from `is_relation_made()` so i can get the context
[17:35] <marcoceppi> timrc: \o/
[19:14] <cargill> marcoceppi: going through the amulet documentation vs. code, could tthe doc somehow hint which parts are not implemented yet? Maybe with a note, or being greyed out?
[19:15] <marcoceppi> cargill: what's not implemented in the docs?
[19:15] <marcoceppi> with the exception of the bash interface, everything in the docs should be in the code
[19:15] <cargill> picking up a file from the unit
[19:16] <cargill> https://bazaar.launchpad.net/~marcoceppi/amulet/master/view/head:/amulet/sentry.py
[19:16] <cargill> oops? maybe I'm reading too fast?
[19:16] <marcoceppi> cargill: it's implemented in UnitSentry
[19:16] <marcoceppi> yeah
[19:17] <marcoceppi> cargill: you can get to it from d.sentry.unit['unit/#'].file()
[19:19] <cargill> sorry, I was getting weird errors, and waiting 20 minutes for a test to get to the failure and not being able to run pdb must have gotten the best of me
[19:47] <lemao> I am trying to figure out "An error occurred: watcher error: [Errno 111] Connection refused". This after I try to access the juju-gui from the browser
[20:32] <mbruzek> lemao Did you run:  juju expose juju-gui    first?
[20:34] <mbruzek> lemao, What command are you trying to run?
[20:40] <lemao> mbruzek: yes, I did expose.
[20:40] <mbruzek> lemao, What command are you getting the error on?
[20:42] <lemao> mbruzek: well, I am trying to get my OSX Mavericks working with Juju in a VagrantBox using the local provider. I am using JuJuBox + precise lxc guests. I basically used the Vagrant IP to access the juju-gui :80 and got a juju web page with the error above
[20:45] <AskUbuntu> Error when using Juju Client on Windows 8.1 juju bootstrap to aws get AWS Acess key id you provided | http://askubuntu.com/q/441663
[21:00] <lazyPower> lemao: are you using the official basebox or did you build your own?
[21:01] <lemao> lazyPower: this specific case I am using the JujuBox
[21:01] <lemao> (error: template container "juju-precise-template" did not stop)
[21:01] <lemao> lazyPower: I also have a trusty vagrant box but ran into other issues
[21:02] <lemao> lazyPower: (sorry the error message came first)
[21:02] <lazyPower> lemao: interesting. I haven't seen that, and i ran the demo vid on mavericks.
[21:02] <lazyPower> but we have have updated the box recently which would have other implications. Whats the juju version in the box?
[21:02] <lemao> lazyPower: Ok, let me review my steps.
[21:02] <lazyPower> s/have have/may have/
[21:03] <lemao> lazyPower: 1.16.6-precise-amd64
[21:03] <lazyPower> ok that seems like its the same version in the box i demo'd
[21:04] <lazyPower> lemao: Can you pack up your logs of the vagrant run and post thems omewhere so I can see whats going on?
[21:08] <tvansteenburgh> anyone know of an existing charm that uses the charm-helpers @hook() decorator?
[21:09] <tvansteenburgh> my hooks aren't failing, but i don't think they're executing either - none of my log msgs are showing up
[21:10] <mbruzek> tvansteenburgh, I don't see @hook() in the ganglia charm that I am working on
[21:10] <tvansteenburgh> here's my code: http://paste.ubuntu.com/7186887/
[21:11] <tvansteenburgh> so, line 214 for example
[21:13] <lemao> lazyPower: Let me start once more from scratch just to make sure it is not a stupid error from my part
[21:13] <lazyPower> lemao: ok ping me if you want me to lookat logs
[21:13] <lemao> lazyPower: I will. thanks
[21:13] <mbruzek> tvansteenburgh, I haven't seen any python hooks like that.  The ones I have seen have a key value section that calls the hooks
[21:14] <lazyPower> @tvansteenburgh the decorator is displayed in the charm helpers youtube video ~ 3/4 of thew ay through if that helps
[21:14] <lazyPower> @tvansteenburgh aslo i think cassandra uses the decorator
[21:15]  * tvansteenburgh looks at cassandra
[21:15] <mbruzek> cassandra hooks are bash
[21:15] <tvansteenburgh> i don't see it being used in cassandra
[21:16] <tvansteenburgh> oh well, i'll see if i can debug it.
[21:18] <tvansteenburgh> oh i see
[21:19] <lazyPower> hmm
[21:19] <lazyPower> maybe i'm thinking of mongo
[21:19] <tvansteenburgh> you still have to execute the hook with `hooks.execute(sys.argv)`
[21:20] <tvansteenburgh> i missed that part
[21:20] <lazyPower> nope
[21:20] <lazyPower> right on
[21:21] <lazyPower> ok i'm heading out for a bit
[21:21] <tvansteenburgh> o/
[21:21] <lazyPower> o/
[21:33] <lemao> lazyPower: stupid me... I had a custom bootstrap.sh being called by vagrant that was probably messing up JujuBox. I removed it and started from scratch and now I get the juju-gui on my OSX
[21:36] <lazyPower> lemao: bueno.
[21:49] <lemao> I have deployed mysql but get "ERROR juju.worker.uniter uniter.go:350 hook failed: exit status 1" how should I debug this in juju?
[21:51] <jose> lemao: ssh into the machine and look at the logs, they're stored in /var/log/juju
[21:56] <tvansteenburgh> debug-hooks also very useful: https://juju.ubuntu.com/docs/authors-hook-debug.html
[22:03] <lemao> jose: I did that but the above message was most of what I could find
[22:05] <lazyPower> lemao: its typically due to the pool memory size
[22:05] <lemao> lazyPower: I ran 'sudo route -n add -net 10.0.3.0 172.16.250.15' in OSX and was able to access LXC nodes inside Vagrant without any port forwarding. If this can be hooked into Vagrant so the route is added when vagrant is up'ed and deleted from halt'ed/destroy'ed that would be quite nice
[22:05] <lazyPower> its a known bug
[22:05] <lazyPower> lemao: we've been using sshuttle
[22:05] <lemao> lazyPower: sshuttle doesn't work for me
[22:05] <lazyPower> that way it doesn't alter your pc's routing table after sshuttle is terminated
[22:06] <lemao> lazyPower: "IOError: [Errno 24] Too many open files: '/etc/resolv.conf'"
[22:06] <lazyPower> lemao: thats coming from sshuttle?
[22:07] <lemao> lazyPower: yeah, that would be ideal. It just barfs as soon as I hit one of the ip's proxied by sshuttle
[22:07] <lemao> lazyPower: yes
[22:07] <lemao> lazyPower: after tons of ': DNS request from ('192.168.0.104', 55405): 31 bytes'
[22:07] <lazyPower> lemao: that has something to do with the max number of open apps/file streams you have going on your system.
[22:07] <lazyPower> i dont think theres anything I can do to resolve that
[22:08] <lemao> lazyPower: yes, that's what I thought, but even after a reboot I still get the same error.
[22:10] <lemao> lazyPower: for now I will stick with manually adding/removing the route. I need to experiment with charms now.
[22:21] <lemao> On a different note, does it make sense to use juju to deploy tomcat wars? Tomcat will definitely be a charm, but what about the 1 or more .war's that are deploy there? Would that be a charm that is related to the tomcat charm?
[22:45] <Valduare> when is 1.18 due to release
[22:50] <davecheney> Valduare: this week
[22:50] <Valduare> neato
[23:21] <lemao> humm, found 'servercloud-r-juju-appserver-support'. Is this basically already supported with subordinate charms (web app deployments)?
[23:21] <lemao> (https://blueprints.launchpad.net/ubuntu/+spec/servercloud-r-juju-appserver-support)
[23:30] <davecheney> marcoceppi: http://paste.ubuntu.com/7187355/
[23:30] <davecheney> manual provider on ppc
[23:31] <marcoceppi> davecheney: sweet!
[23:32] <davecheney>       mysql/0:
[23:32] <davecheney>         agent-state: pending
[23:32] <davecheney>         agent-version: 1.17.8.9001
[23:32] <davecheney>         machine: "1"
[23:32] <davecheney>         public-address: 10.245.67.7
[23:32] <davecheney> marcoceppi: it's over 9000
[23:32] <marcoceppi> hahak I noticed
[23:33] <andreas__> how do I bootstrap a trusty env, just "juju bootstrap --series=trusty"?
[23:33] <davecheney> andreas__: good question
[23:34] <davecheney> andreas__: do you want a trusty boottrap machine
[23:34] <davecheney> or service units running trusty
[23:34] <andreas__> right now I wanted bootstrap only because it was going to be the fastest way to get a clean trusty machine "out there"
[23:34] <davecheney> andreas__: if you want a clean trusty macine
[23:35] <andreas__> but if that doesn't work, I can do having bootstrap on whatever an a "ubuntu" unit on trusty
[23:35] <davecheney> juju deploy cs:trusty/ubuntu
[23:35] <andreas__> yeah, I just didn't want to waste a machine
[23:35] <davecheney> andreas__: sorry, i didn't understan that last comment
[23:35] <andreas__> so I figured bootstrap could be trusty
[23:35] <andreas__> davecheney: juju deploy cs:trusty/ubuntu will in the end need two machines
[23:35] <andreas__> bootstrap + ubuntu
[23:35] <davecheney> andreas__: yes
[23:35] <davecheney> you could try
[23:35] <andreas__> I just wanted one, and figured bootstrap could be it
[23:35] <davecheney> juju bootstrap --constraints="series=trusty"
[23:35] <davecheney> it might work
[23:36] <andreas__> ok, because juju bootstrap --series=trusty doesn't
[23:36] <davecheney> --series isn't a flag
[23:36] <andreas__> ah, it's just about the tools?
[23:36] <andreas__>     upload tools for supplied comma-separated series list
[23:36] <davecheney> andreas__: nope
[23:37] <andreas__> I don't know what it is for then, clearly my assumption was wrong
[23:37] <davecheney> andreas__: juju makes it really hard to control the series of the bootstap machine
[23:37] <andreas__> and the "series" constraint isn't known
[23:37] <davecheney> andreas__: you could try default-series in config
[23:38] <davecheney> but we're getting rid of that
[23:38] <andreas__> I think that doesn't exist anymore either
[23:38] <andreas__> I'll go with bootstrap + ubuntu