[00:06] <rick_h_> lazyPower: bpython :P
[00:06] <lazyPower> rick_h_: googling - hang on
[00:06] <lazyPower> sexy, does it have an integrated breakpoint thingy too?
[00:07] <lazyPower> and yes, thats the techinical definition, breakpoint thingy.
[00:08] <rick_h_> http://stackoverflow.com/questions/9892751/is-it-possible-to-use-bpython-as-a-full-debugger
[00:08] <lazyPower> boo - requires a wrapper
[00:09] <rick_h_> yea, I use it as my python repl but yea ipdb for pdb quick in
[00:09] <lazyPower> i had no idea this existed, i've been complaining about a lack of pry since i started this journey
[00:09] <lazyPower> now i can finally shut my face hole and get back to coding
[00:10] <lazyPower> and with that, its time for me to steam. o/  have a good evening everybody. See you in the am
[00:18] <lazyPower> davecheney: Hey before i jet, i saw your message. I'm not positive on which package provides charm-helpers-sh but i'm fairly certain its not up to snuff
[00:18] <lazyPower> https://launchpad.net/ubuntu/precise/amd64/charm-helper-sh - it appears there is a package for it in precise - which release are you using?
[00:20] <davecheney> lazyPower: not precise
[00:20] <davecheney> lazyPower: the one we can't announce yet
[00:20] <davecheney> running on the computer only banks can afford
[00:21] <davecheney> lazyPower: marcoceppi has a fix in progress https://code.launchpad.net/~marcoceppi/charms/precise/mysql/lp1293315/+merge/211360
[00:21] <lazyPower> well it appears the last release was for quantal, i'm not seeing it in saucy
[00:21] <lazyPower> yeah i just got that email. right on.
[00:21] <lazyPower> Glad to see its being sorted, didn't want to leave you hanging since you asked me implicitly
[00:21] <lazyPower> ta
[00:24] <davecheney> lazyPower: no worries
[00:24] <davecheney> there will be plenty more bugs
[00:24] <davecheney> while we remove the implicit assumption that there is only one series
[00:40] <hatch> is anyone else getting agent-state-info: '(error: container failed to start)' with 1.17.5 and local envs?
[00:40] <hatch> on precise
[00:43] <davecheney> hatch: can you get some details from ~/.juju/local/log/*.log
[00:44] <hatch> davecheney sure, checking
[00:45] <hatch> davecheney all-machines.log is empty
[00:46] <hatch> machine-0 has
[00:46] <hatch> 2014-03-18 00:31:59 WARNING juju.worker.instanceupdater updater.go:231 cannot get instance info for instance "": no instances found
[00:47] <hatch> 2014-03-18 00:32:19 ERROR juju.container.lxc lxc.go:211 container failed to start: container failed to start
[00:47] <hatch> 2014-03-18 00:32:19 ERROR juju.provisioner provisioner_task.go:407 cannot start instance for machine "1": container failed to start
[00:47] <hatch> 2014-03-18 00:32:29 WARNING juju.worker.instanceupdater updater.go:231 cannot get instance info for instance "": no instances found
[00:47] <hatch> and that's repeated for every instance after as well
[00:48] <davecheney> hatch: brillant
[00:48] <davecheney> enterprise grade
[00:48] <davecheney> hatch: what series are you using ?
[00:48] <hatch> precise
[00:48] <davecheney> you can't use precise
[00:48] <davecheney> local provider is not supported on precise
[00:48] <sarnold> no? aww
[00:49] <davecheney> kernel is too crap
[00:49] <hatch> lies! I've been using precise for ever
[00:49] <davecheney> lxc is too crap
[00:49] <davecheney> hatch: all i can tell you is what I know
[00:49] <davecheney> local provider is not supported on precise
[00:49] <hatch> seriously though, this worked just fine before the latest update
[00:49] <davecheney> hatch: we're both saying things that are true
[00:49] <marcoceppi> hatch: delete the files in /var/cache/lxc/cloud-precise and try again?
[00:50] <hatch> lxc really isn't supported on precise? It's just been working by fluke this whole time? lol
[00:51] <hatch> marcoceppi trying
[00:51] <sarnold> the lxc team does have a PPA you can use on precise, and maybe with a newer HWE kernel stack and that PPA it might work, but probably be just as unsupported :)
[00:52] <davecheney> sarnold: exactly
[00:53] <sarnold> heh actually I think I used pyjuju and lxc on precise way back when :)
[00:54] <hatch> marcoceppi looks like that worked!
[00:54] <marcoceppi> when in doubt, clear the cache
[00:55] <hatch> haha yeah I totally blanked on that, thanks
[00:55] <hatch> now I can get back to working on the ghost charm again
[06:42] <jose> hey guys, any charmers around
[06:42] <jose> ?
[07:59] <vladk> jam, good morning
[08:31] <jam> vila: ping, it looks like official 1.17.5 binaries *were* uploaded to HP cloud, in the official bucket. If you unset "tools-metadata-url" I believe it will select the right bucket for you
[08:36] <vila> jam: ha ! Interesting, I've been having issues uploading to hp starting... yesterday and I'm trying to diagnose that with support. I will try to unset the tools-metadata-url as soon as I can, thanks for the feedback !
[08:36] <jam> vila: let me know if it works
[08:41] <vila> jam: trying, just got 2014-03-18 08:40:13 INFO juju.environs.bootstrap bootstrap.go:58 picked newest version: 1.17.5
[08:41] <vila> jam: sounds encouraging ?
[08:42] <jam> vila: I believe if it got there, it found the tools
[08:42] <jam> though I also see them in the official location finally: http://streams.canonical.com/juju/tools/releases/
[08:47] <vila> jam: bootstrapped, deploying
[08:47] <jam> yay
[08:48] <vila> node 0 agent-version: 1.17.5
[08:58] <vila> jam: still bitten by the 'broken pipe' on swift upload we do before the deployment :-/
[08:59] <vila> jam: juju is not involved there, that's the issue I'm tracking with hp
[13:05] <zchander> Good afternoon, everyone. Anyine around who can help me with Ceph/Juju/MaaS with physical PCs with one drive (only /dev/sda) installed
[14:29] <jcastro> sinzui, hey do we test on windows 8.1? http://askubuntu.com/questions/435928/installing-juju-client-on-windows-8-1
[14:48] <marcoceppi> mbruzek: can you review something right quick for me
[14:48] <mbruzek> yes
[14:49] <mbruzek> What  is it?
[14:49] <marcoceppi> https://code.launchpad.net/~marcoceppi/charms/precise/mysql/lp1293315/+merge/211360
[15:09] <sinzui> jcastro, the test is on win 2012 server
[15:14] <mbruzek> marcoceppi, reviewed.
[15:14] <mbruzek> marcoceppi, Can you remind me if the UnitSentry.file_contents() runs as root or as ubuntu?
[15:14] <jcastro> sinzui, I can test on win 8.1
[15:15] <marcoceppi> mbruzek: everything sentry i sroot
[15:15] <mbruzek> Ok thanks.
[15:16] <sinzui> jcastro, I believe win 2012 server is 8.1 without the gui
[16:15] <jcastro> sinzui, filed a bug, I can confirm we are broken on windows 8.1
[16:20] <sinzui> jcastro, did you use 1.17.5? Trunk is broken with all windows at the moment
[16:20] <jcastro> I am using what's on the website, 1.16.6
[16:25] <sinzui> jcastro, thank you
[16:38] <sinzui> jcastro, marking a bug confirmed just delays it from being triaged. We see the new bugs first
[16:39] <jcastro> oh sorry about that, I marked it confirmed because another person had the same problem, I'll remember that moving fwd.
[16:45] <viperZ28_> Does Juju have an ability to ensure the number of instances requested stays running?
[16:46] <viperZ28_> i..e. I have a RabbitMQ cluster with 2 units, one of the machines goes down, does Juju have the ability to start another instance or attempt to restart the downed instance?
[16:46] <sinzui> jcastro, would you be able to also test 1.16.5. I suspect a signing issue, and it may to Bug #1288873
[16:46] <_mup_> Bug #1288873: Installing juju windows binarys fails on windows 7 <windows> <juju-core:Incomplete> <https://launchpad.net/bugs/1288873>
[16:47] <sinzui> jcastro, I got signed binary differently for 1.16.6 and 1.16.5 I wonder is a proper signing will fix the issue
[16:49]  * sinzui is thinking signing is the issue because that cannot be tested
[17:01] <jose> hey guys, can anyone give me a hand on how should I use the mysql charm?
[17:01] <viperZ28_> https://www.irccloud.com/pastebin/Asm6AF5E
[17:04] <jcastro> jose, just ask!
[17:04] <jcastro> sinzui, that's interesting
[17:04] <sinzui> jcastro, I am looking for the presigned binary I sent to IS
[17:05] <jose> jcastro: so, I want to charm openphoto but uses mysql, how do I get the credentials?
[17:05]  * sinzui removed it from the milestone when the signed version was delivered :(
[17:05] <jcastro> anything I can do to help? broken on windows is bad
[17:05] <jcastro> jose, https://juju.ubuntu.com/docs/interface-mysql.html
[17:07] <viperZ28_> so it looks like Juju does not have ability to enforce runtimes only startup
[17:07] <jose> jcastro: so if I run `relation-get user` I will get the user for the db as a string? as simple as that?
[17:08] <jcastro> yessir
[17:09] <jose> jcastro: was that a yes to viper or to me?
[17:09] <jcastro> to you
[17:10] <rick_h_> jcastro: <3 that doc /me wants more!
[17:10] <jcastro> jose, if you want to debug it live, do a `juju debug-hook` into the unit
[17:10] <jcastro> and you can run the commands in the hook context
[17:10] <jose> that's awesome
[17:10] <jcastro> rick_h_, yeah I wish we had one for every interface
[17:10] <jose> thanks a bunch, going to work on that now :)
[17:10] <jcastro> viperZ28_, what are you trying to do?
[17:11] <viperZ28_> In my test I brought up a multi-node RabbitMQ cluster, I then took one of the machines out using `lxc-shutdown`,
[17:11] <viperZ28_> Juju did not try to restart the machine or spin up another one
[17:12] <jcastro> ah yes, there's been some discussion around that
[17:12] <viperZ28_> I was hoping Juju would sense the downed machine and make an attempt to restart it
[17:12] <viperZ28_> jcastro: was anything decided?
[17:12] <jcastro> I am trying to find the discussion
[17:15] <viperZ28_> I am also looking for plans to integrate with vSphere/ESX stack
[17:16] <jcastro> yeah I think one of the guys in #juju-dev would know better
[17:21] <viperZ28_> jcastro: thanks
[17:22] <sinzui> jcastro, I must congratulate myself in my effort to make sure the non-signed 1.16.6 never got out.
[17:23] <jcastro> heh
[17:23] <sinzui> jcastro, I can provide the 1.16.7 that was made when I incremented 1.16.6. I can remake 1.16.6 with the original tarball.
[17:24] <jcastro> sinzui, whatever you think is best, I only want us to work again on windows
[17:24] <sinzui> well I should remake and ask for a resigning. We can then test both
[17:27] <jose> hey, let's say I want to relate mysql and ubuntu for testing purposes, how should I?
[17:41] <lazyPower> jose: if the charm doesn't implement the expected relationship hooks, what you ask is nigh impossible
[17:41] <lazyPower> jose: pick a charm that has a :db relationship and you should be able to relate them via juju add-relation
[17:41] <jose> it's fine, as it's just for testing I'm installing mysql locally
[17:42] <lazyPower> well i dont understand what you're trying to do, there's no application on either end to support it. If you're just going to get/set the data exchange as an experiment?
[17:42] <jose> I'm trying to do a deploy of phpbb and wanted to relate it with the mysql charm
[17:43] <jose> so if it works then I just copy it over
[17:43] <lazyPower> https://github.com/charms/wordpress/blob/master/hooks/db-relation-changed - here's an example of how the wordpress charm does it
[17:44] <lazyPower> mediawiki is another good one to reference.
[17:44] <jose> so, if wp doesn't have a relation with mysql, does it still run?
[17:45] <jose> (I understand not)
[17:46] <lazyPower> it doesnt do anything, it no-ops until the relationship is present
[17:48] <lazyPower> i have a charm written in chef thats optimized for single host / scale out operation - its got a few bugs that need to be worked out though.
[17:49] <lazyPower> namely, it re-seeds the scale database with seed data on add-unit, and the secret-key is not present until first run, and chef is dealing me fits considering its pre-compiled on run, so i have to figure out how to lazy load it, but thats a horse of a different color
[18:29] <jcastro> marcoceppi, arosales: ok so I have a card, republish charm school schedule
[18:29] <jcastro> how do we feel about just declaring "the first friday of every month"? Barring holidays
[18:30] <arosales> jcastro: are you thinking once per month?
[18:30] <jcastro> yeah + any ondemand
[18:30] <jcastro> I have the topic list, we just need to pick the next topic
[18:30] <jcastro> marcoceppi, how about Juju plugins?
[18:31] <marcoceppi> jcastro: we can do tath
[18:31] <marcoceppi> it's a short one
[18:31] <marcoceppi> one we can probably do out of band
[18:32] <arosales> jcastro: suggest to schedule 2-3 months out if you are looking at 1 charm school / month
[18:32] <jcastro> ok
[18:32] <arosales> we may need to see how if that is too little or just right
[18:33] <jcastro> hey we never did "troubleshooting Juju"
[18:33] <arosales> as we do more and record on specific topics though the video resource library grows
[18:33] <jcastro> that could be a nice 2 parter, go deep, we've never done that before
[18:33] <jcastro> and it's on the list to do
[18:33] <arosales> +1 on trouble shooting part 2
[18:33] <marcoceppi> jcastro: we should do the troubleshooting next
[18:33] <arosales> +1 on juju plugins
[18:34] <marcoceppi> we can talk for hours on that
[18:34] <jcastro> yeah so
[18:34] <jcastro> plugins
[18:34] <jcastro> troubleshooting 1
[18:34] <arosales> osx and windows workflows
[18:34] <jcastro> troubleshooting 2
[18:34] <jcastro> and then I can put the rest of the topics on there, see what people want
[18:34] <jcastro> arosales, ooh, those are good
[18:34] <jcastro> we should do the workflow one before release
[18:35] <jcastro> maybe do 2 in april? one with marco(plugins), one with chuck(OSX workflow?)
[18:35] <arosales> jcastro: maybe sneek in an "intro to charming" every other month around the 15th
[18:35] <jcastro> lazyPower ^^ what do you think
[18:35] <arosales> jcastro: that also sounds good
[18:35]  * lazyPower thumbs up
[18:37] <jcastro> marcoceppi, ok you are 6 april
[18:37] <jcastro> lazyPower, you are 11 april
[18:37] <lazyPower> next month is going to be crazy busy
[18:38] <sarnold> welcome aboard :)
[18:38] <lazyPower> jcastro: on the eco calendar?
[18:38]  * lazyPower didn't look - because lazy
[18:38] <jcastro> no, Ubuntu Fridge calendar, since it's a public project-wide thing
[18:38] <jcastro> It will send invites
[18:38] <lazyPower> hokay
[18:38] <jcastro> I'll add it to your work calendar too
[18:38] <jose> jcastro: is it fine if you do it at uonair?
[18:39] <jose> I'd love to see it there
[18:39] <jcastro> yeah, I need to know how to use the account though, I don't remember how
[18:39] <jose> I can walk you through again, I have all the credentials and everything set
[18:40] <jcastro> actually, if you just jet me a mail that would be <3
[18:40] <jose> writing that now!
[18:41] <jose> jcastro: do you still have the password or should I resend it?
[18:43] <jcastro> I do not
[18:43] <jcastro> Sorry for the calendar spam everyone!
[18:47] <sinzui> jcastro, I added a win installer for you to test https://bugs.launchpad.net/juju-core/+bug/1294198
[18:47] <jose> you have mail, jcastro!
[18:47] <_mup_> Bug #1294198: Juju package doesn't work on Windows 8.1 <windows> <juju-core:Triaged by sinzui> <https://launchpad.net/bugs/1294198>
[18:48] <sinzui> jcastro, I am going to do the reverse. I will feed the signed juju into CI. I hope it fails, because I can add a step to retest the signed win installer to ensure we don't post crack
[18:52] <sinzui> \o/ jcastro the signed win installer is rejected by CI.
[18:52]  * sinzui tries the previous installer
[18:53] <mbruzek> I am running trusty and lxc
[18:53] <mbruzek> Has anyone seen this error
[18:53] <mbruzek> http://pastebin.ubuntu.com/7115719/
[18:53] <mbruzek> '(error: container "mbruzek-local-machine-1" is already created)'
[18:54] <mbruzek> It is preventing my amulet tests from completing.
[18:54] <mbruzek> lazyPower, marcoceppi ^ ?
[18:54] <marcoceppi> mbruzek: sudo lxc-ls --fancy
[18:55] <mbruzek> http://pastebin.ubuntu.com/7115730/
[18:55] <sinzui> jcastro, CI accepts 1.16.5's win installer. We know the signing is bad, and we can test it
[18:55] <mbruzek> stopped machines?
[18:56] <marcoceppi> mbruzek: destroy environment
[18:56] <mbruzek> I saw this error before and destroyed environment at that time
[18:56] <mbruzek> I will also run my clean up script to get rid of the lxc
[18:56] <marcoceppi> mbruzek: after you destroy environment
[18:57] <marcoceppi> sudo lxc-ls --fancy
[18:57] <mbruzek> empty after I ran the clean up script
[18:57] <marcoceppi> mbruzek: try bootstrapping again
[18:58] <mbruzek> Sure.
[18:59] <jcastro> sinzui, I will give it a shot
[19:03]  * sinzui ponders deleting the signed win installer from the release and uploading the unsigned.
[19:07] <fishy_> Is upgrading lxc on ubuntu 12 going to break everything?  My lxc-ls doesn't have a --fancy and I am jealous
[19:17] <jcastro> sinzui, works!
[19:19] <Guest25908> juju sync-tools is struck in lisint target bucket. what could be the problem?
[19:19] <Guest25908> its not showing any output..
[19:19] <Guest25908> *listing
[19:20] <fishy_> sudo lxc-ls --fancy ls: unrecognized option '--fancy' Try `ls --help' for more information. ls: unrecognized option '--fancy' Try `ls --help' for more information.
[19:21] <marcoceppi> fishy_: lxc-ls --fancy is lxc-ls on precise
[19:23] <fishy_> that just lists machines, not IPs and such
[19:25] <marcoceppi> fishy_: well, there's nothing wrong with upgrading lxc that I'm aware of, though you mihgt need a newer kernel
[19:26] <fishy_> sounds dangerous
[19:27] <fishy_> real problem:  i started a wordpress guy in a local environment and exposed him ala tutorial.. cannot get to him in browser.    Tried adding IPtables rule like so: iptables -t nat -A PREROUTING -i eth0 -p tcp --dport 80 -j DNAT --to 10.0.3.93:80     .  Still no luck..
[19:28] <fishy_> lxc is doing something weird with the bridge
[19:28] <fishy_> by default
[19:32] <lazyPower> fishy_: can you ssh to the unit with juju ssh wordpress/0?
[19:32] <fishy_> ya can get in that way
[19:32] <fishy_> just not though the external IP
[19:33] <lazyPower> ...external ip?
[19:33] <lazyPower> OH! you mean you've deployed an lxc container on another machine and you want to route that traffic into the lxc container(s)
[19:33] <fishy_>         open-ports:         - 80/tcp         public-address: 10.0.3.73
[19:33] <fishy_> no its on my box
[19:33] <fishy_> i want to just go over loopback to it
[19:33] <lazyPower> hmm, whats your bridge interface? lxcbr0?
[19:33] <fishy_> ya
[19:34] <lazyPower> and you get timeouts when pinging it? is your network on the 10.0.3.x octet?
[19:34] <fishy_> i think 10.0.3.x is used by some real machines on my network
[19:34] <lazyPower> probably having an ip collision then
[19:34] <fishy_> so pings hit that..  that was my first idea
[19:34] <fishy_> ok, back to how to change LXC network
[19:34] <lazyPower> that or the service isn't actually up
[19:34] <lazyPower> thats configurable in... 1 sec
[19:35] <fishy_> i do have 1 error in juju status I should look at too
[19:35] <fishy_>   mysql:     charm: cs:precise/mysql-37     exposed: false     relations:       cluster:       - mysql       db:       - wordpress     units:       mysql/0:         agent-state: error         agent-state-info: 'hook failed: "start"'         agent-version: 1.16.6.1         machine: "2"         public-address: 10.0.3.103
[19:35] <lazyPower> there's the issue
[19:35] <lazyPower> wordpress didn't actually do anything
[19:35] <lazyPower> it noop's on setting up the web interface until the mysql host has joined.
[19:35] <fishy_> because it couldnt talk to mysql?
[19:36] <fishy_> ok
[19:36] <lazyPower> that may be a bit misleading, it does stuff
[19:36] <lazyPower> but you wont be able to get to it unless that completes successfully
[19:36] <fishy_> 2014-03-18 19:32:30 ERROR juju.worker.uniter uniter.go:350 hook failed: exit status 1
[19:36] <fishy_> from the wordpress log
[19:36] <lazyPower> which hook?
[19:36] <lazyPower> start?
[19:37] <fishy_> yes
[19:37] <lazyPower> try juju resolved -r mysql/0 -- if its a temporary hiccup, it should resolve itself on the re-run of the hook.
[19:37] <fishy_> 2014-03-18 19:32:27 INFO juju.worker.uniter uniter.go:348 running "start" hook 2014-03-18 19:32:27 INFO juju.worker.uniter context.go:255 HOOK mysql stop/waiting 2014-03-18 19:32:30 INFO juju.worker.uniter context.go:255 HOOK start: Job failed to start 2014-03-18 19:32:30 ERROR juju.worker.uniter uniter.go:350 hook failed: exit status 1
[19:38] <fishy_> same, need to find real logs
[19:39] <lazyPower> fishy_: better yet, lets interactively debug it
[19:40] <lazyPower> juju resolved -r mysql/0; juju debug-hooks mysql/0
[19:40] <lazyPower> it'll be a race to see if it attaches to the tmux session before the hook re-runs, but 90% of the time, it works
[19:40] <fishy_> awesome
[19:40] <lazyPower> otherwise, you just re-run juju resolved -r mysql/0 after you've attached (in a sep. terminal)
[19:41] <lazyPower> then you can re-run the hook that failed, eg: `hooks/start`
[19:41] <lazyPower> if you get no output, add the xtrace to the top of the hook "set -eux"
[19:41] <lazyPower> and then re-run it and see where it turned into a bowl of chunky salsa
[19:42] <fishy_> mysql/0:start % hooks/start stop: Unknown instance:  start: Job failed to start
[19:43] <lazyPower> ah
[19:43] <lazyPower> alter line 5 to read as follows
[19:43] <lazyPower> service mysql start || service mysql restart
[19:43] <fishy_> line 5 of what?  in the tmux?
[19:43] <lazyPower> looks like it never started up in the config-changed hook
[19:43] <lazyPower> the start hook in hooks/start
[19:44] <fishy_> mysql/0:db-relation-joined % ./start  start: Job failed to start stop: Job has already been stopped: mysql
[19:44] <lazyPower> interesting
[19:45] <lazyPower> marcoceppi: this happened before, it was due to a memory tuning bug right?
[19:45] <lazyPower> or am i remembering some ancient workaround?
[19:45] <fishy_> was ./start the right way to execute the start hook?
[19:45] <fishy_> or do I need to do some more magic
[19:45] <lazyPower> nah thats it if you're in $CHARM_DIR/hooks
[19:46] <fishy_> ya
[19:46] <lazyPower> however, typically you path it with hooks/start from $CHARM_DIR usually, as charms written in bash assume they are being called from $CHARM_DIR
[19:46] <lazyPower> just a bit of info to keep in mind moving forward
[19:46] <fishy_> what did set -eux do
[19:47] <lazyPower> sets explicit, xtrace, and i forget what u does
[19:47] <fishy_> i know e
[19:47] <fishy_> cool
[19:47] <lazyPower> set -u makes Bash check whether you have initialised all your variables. If you haven't, Bash will throw an error about unbound variables.
[19:47] <lazyPower> ah! thats right. strict mode
[19:47] <fishy_> nice
[19:47] <sinzui> jcastro, One more test. IS resigned the installer. https://launchpad.net/juju-core/1.16/1.16.6 has the signed version that CI liked. Does your machine like it too?
[19:48] <lazyPower> fishy_: well, i dont remember the exact cause of this bug but there is a mailing list post about it
[19:48] <lazyPower> let me try to fish it up for you
[19:49] <fishy_> The hook im "stuck on" is db-relation-joined
[19:49] <fishy_> could it be something that hook did that broke me
[19:49] <fishy_> that now is making start break
[19:49] <lazyPower> wait, "stuck on"
[19:50] <fishy_> my tmux window title
[19:50] <lazyPower> i dont kno what youv'e done so far, i was under the impression the service was stuck on the "start" hook
[19:50] <fishy_> is db-relation-joined
[19:50] <lazyPower> did you exit from the start hook? as db-relation-joined would be the next hook in the sequence.
[19:50] <fishy_> = thats the step im on?
[19:50] <lazyPower> that debug-hooks interactive tmux session is like a queue, when you complete one hook context it moves into the next automatically
[19:51] <lazyPower> and db-relation-joined is what creates the myql user, database, password and hands off the information to the requesting charm, so without an active mysql daemon on the host, it will fail to do anything
[19:51] <fishy_>       mysql/0:         agent-state: error         agent-state-info: 'hook failed: "db-relation-joined"'
[19:51] <fishy_> makes sense
[19:51] <jose> hey guys, how can I make a loop which ends when a statement is true?
[19:51] <jose> that's in bash
[19:52] <fishy_> you can ram your statement in a while block
[19:52] <fishy_> like its 1996
[19:53] <lazyPower> jose: http://tldp.org/LDP/Bash-Beginners-Guide/html/sect_09_02.html
[19:53] <lazyPower> fishy_: https://lists.ubuntu.com/archives/juju/2014-February/003421.html
[19:53] <lazyPower> here's the thread
[19:53] <jose> uh, looks neat, thanks!
[19:55] <fishy_> hum my innodb_buffer_pool_size is set to 20 GB, that seems suspicious
[19:55] <fishy_> investigating mysql
[19:55] <mbruzek> fishy_, http://tldp.org/HOWTO/Bash-Prog-Intro-HOWTO-7.html
[19:56] <mbruzek> sorry that was for jose.
[19:57] <mbruzek> But it looks like lazyPower was all over that one
[19:57] <jose> mbruzek: thanks! btw, any chance you already checked the mailman charm?
[19:58] <mbruzek> Not yet, there are a few other charms ahead of it on the queue
[19:58] <jose> ok then :)
[19:58] <mbruzek> I will try to get to it as soon as possible.
[19:58] <fishy_> where do hooks live inside a container
[19:58] <lazyPower> fishy_: are you still in your tmux session
[19:59] <jose> fishy_: /var/juju/ I think
[19:59] <fishy_> nope ;/
[19:59] <lazyPower> /var/lib/juju/agents/unit-<service>-<unit#>/charm/hooks
[19:59] <lazyPower> if you're in hook context, you have an environment variable $CHARM_DIR
[20:00] <lazyPower> so, use either/or depending on context
[20:00] <fishy_> i dont see any env vars set
[20:00] <fishy_> after a juju ssh
[20:01] <lazyPower> right, you *have* to be in the context of an executing hook to get those env variables
[20:01] <lazyPower> juju is very judicious in keeping its environment to itself.
[20:03] <fishy_> and debug hooks is how you do that?
[20:03] <fishy_> got mysql up:  mysql    23066  0.4  0.5 1485292 129720 ?      Ssl  20:01   0:00 /usr/sbin/mysqld
[20:03] <lazyPower> correct. so long as you're in one of those titled tmux sessions, you're in a hook-execution-context
[20:04] <lazyPower> otherwise, you have none of the amenities, and its just a plane ole tmux session
[20:04] <fishy_> fancy
[20:04] <lazyPower> cool - what was the fix for you? changing the innodb size?
[20:05] <fishy_> yah from 20GB down to 1GB
[20:05] <fishy_> and a restart of mysql
[20:05] <fishy_> however that value is populated seems wrong
[20:05] <fishy_> my machine has 24GB ram.. but giving 20 to mysql seems excessive
[20:06] <fishy_> i guess on a 1 GB vm, giving it 756MB wouldn't be bad
[20:07] <lazyPower> fishy_: i hate to be a bother, but could you file a bug with those findings for me? +1 that it affects you and give us a brief summary of what you did to correct the problem? the more feedback we have on this the better the charm will be for future users.
[20:07] <lazyPower> https://bugs.launchpad.net/charms/+source/mysql
[20:08] <fishy_> yes
[20:08] <fishy_> while I type that up..  any pointers on changing my LXC IP range to not conflict?
[20:08] <lazyPower> Excellent, ping me with the bug URL after its filed.
[20:08] <lazyPower> ah right - i forgot about that 1 moment
[20:09] <lazyPower> fishy_: you edit the network in /etc/default/lxc
[20:10] <lazyPower> fisy_: http://paste.ubuntu.com/7116089/
[20:10] <lazyPower> aside from having the bridge be br0 (which unless you create that bridge, wont be available to you) - that's what the config will look like.
[20:11] <lazyPower> and apparently some stray backslashes got in there when copy/pasting, strange...
[20:11] <lazyPower> so yeah, disregard the bridge and the random backslashes.
[20:13] <fishy_> reported https://bugs.launchpad.net/charms/+source/mysql/+bug/1294334
[20:13] <_mup_> Bug #1294334: mysql charm blows up on out of memory error <mysql (Juju Charms Collection):New> <https://launchpad.net/bugs/1294334>
[20:16] <lazyPower> fishy_: thanks! You should be emailed when a fix lands and the bug is resolved.
[20:18] <lazyPower> sorry that was such a long winded fix too - but its intermittant in when it decides to rear its ugly head. I've been tracking that one on and off for about a month now.
[20:20] <fishy_> i still never saw a good error anywhere
[20:20] <fishy_> should have ended up somewhere
[20:21] <fishy_> but at least i have set up mysql enough to know when I see a crazy param
[20:21] <lazyPower> yeah, that was the proposed work around in the mailing list too was to reduce the innodb buffer pool
[20:22] <lazyPower> thing is, when i deploy mysql in lxc, i dont see that behavior, and haven't since i was on the 1.16 series of juju
[20:22] <fishy_> on a local deploy?
[20:22] <lazyPower> then again, i'm tracking the devel releases now - are you on the ppa:juju/stable release?
[20:22] <lazyPower> right
[20:22] <fishy_> do you have a ton of ram on your machine
[20:22] <lazyPower> 8gb, fairly normal size for a light weight dev rig
[20:23] <fishy_> ya it may only manifest with a lot of ram
[20:23] <fishy_> like 24
[20:23] <fishy_> i am on whatever tutorial said, assume stable
[20:23] <lazyPower> Well, my juju farm in the closet has 32gb of ram, and doesn't do that either
[20:23] <lazyPower> yeah, you would be on stable, we dont recommend people use devel unless you want shiney features and the iniability to safely upgrade (or so i've been told, ymmv, i haven't done it)
[20:23] <fishy_> i think i need   a) machine with 24g ram.   b) swap sapce disabled.     c) about 15gb of other stuff open
[20:24] <fishy_> could be part of my setup
[20:24] <lazyPower> well, by default when deploying to lxc i dont see the cause/effect for having a 20gb innodb cache - you're not exactly serving up the wikipedia database - however - there are those using lxc in production
[20:24] <lazyPower> so thats a slippery slope
[20:25] <fishy_> well and if i deploy it to a 20GB machine with nothing else on it
[20:25] <fishy_> nothing better to use it for
[20:25] <lazyPower> fishy_: if you need anything else dont hesitate to ping! I'm going to stuff my head back in a terminal
[20:25] <fishy_> assume most 20GB machines want a big cache
[20:25] <fishy_> ok thanks
[20:32] <marcoceppi> lazyPower: I've got 32gb, but mysql doesn't have any issues on my machine
[20:33] <lazyPower> marcoceppi: its strange how intermittant it is. we've confirmed 3 cases of it being silly now
[20:33] <lazyPower> sebas, daniele, and now fishy - my issues resolved when i jumped to 1.17.4 - not sure WHY that made a difference, and may be coincidental
[20:47] <varun123> i am trying to install juju, in https://help.ubuntu.com/community/UbuntuCloudInfrastructure inside install juju it says "juju requires ssh keys to be able to access the deployed nodes" here which node does deploy nodes refer to?
[20:48] <lazyPower> varun123: everything deployed with Juju uses key based authentication. You will not be given accounts with username/password combinations.
[20:48] <varun123> okay.
[20:49] <varun123> so we need to create ssh keys for all nodes?
[20:50] <lazyPower> Juju will handle that for you. When you run juju init on the station you will be performing your orchestration from, the keys are generated and placed in ~/.juju/ssh
[20:50] <jcastro> I think in the past we didn't generate keys automatically
[20:50] <jcastro> which is why we say that
[20:50] <lazyPower> its a good callout too - i can see where people would be confused. Not everybody does keybased authentication, even though they should be.
[20:51] <lazyPower> </my 2 cents>
[20:51] <varun123> okay but juju init u meant juju bootstrap? because juju-quick-start doesnt say juju init
[20:51] <varun123> *by
[20:56] <jose> guys, should I use db-relation-joined or db-relation-changed for when a DB starts a relation with my charm?
[20:57] <lazyPower> jose: authors choice. Do you want to run the code on everytime something changes on the db unit?
[20:57] <jose> not at all, just run the code when the relationship is done to set up some config files
[21:09] <lazyPower> jose: sounds like a joined hook to me
[21:09] <jose> ok then :)
[21:50] <timrc> juju destroy-service really needs --force and --with-fire options.  The --force option is self-explanatory the --with-fire option is basically an alias of --force but conveys a higher level of frustration :)
[21:50] <timrc> Especially important while developing and testing a charm :)
[21:53] <marcoceppi> timrc: ;)