[07:32] <noodles775> jcastro: Hi! Could you schedule my "Simpler re-usable charms with ansible" session when you get a chance? Weds 14UTC would suite me best (looks like DevOps1 is free then): http://summit.ubuntu.com/uos-1406/michael.nelson/meetings
[09:50] <thumper> hi stub, got a few minutes?
[09:54] <stub> thumper: sure
[09:58]  * thumper is trying to get the storage subordinate working
[09:59] <thumper> stub: any idea what values I should put in the block-storage-broker config for endpoint?
[09:59] <thumper> stub: using the default ec2
[09:59] <thumper> us-east-1 according to the config
[10:00] <thumper> s3.amazonaws.com?
[10:00] <stub> thumper: I don't know EC2 sorry.
[10:00] <thumper> hmm... ok
[10:00] <thumper> stub: any idea what chad's irc nick is?
[10:01] <thumper> stub: the default precise postgresql charm now supports the storage subordinate?
[10:01]  * thumper tries the deployer config he set up
[10:03] <stub> csmith isn't around, but our we-bops or landscape could help
[10:03] <stub> thumper: yes, the default precise postgresql charm supports the storage subordinate (and only the storage subordinate)
[10:03] <stub> trusty... not sure if that is live yet (in which case, I should push) or still being setup
[10:04] <thumper> kk
[10:04] <thumper> yeah, gone for precise
[10:04] <thumper> likely to shoehorn everything onto machine 0 :-)
[10:15] <thumper> hmm... endpoint is wrong
[10:21]  * thumper sighs
[10:22] <thumper> bugger
[10:35] <natefinch> marcoceppi_: are you up this early?
[10:36] <marcoceppi_> natefinch: I am!
[10:36] <natefinch> marcoceppi_: cool
[10:36] <marcoceppi_> (un)fortunately?
[10:36] <marcoceppi_> what's up?
[10:36] <natefinch> hah
[10:36] <natefinch> I'm writing a charm!  WHich means I'll have a lot of questions :)
[10:37] <natefinch> the charmtool outputs a README.ex .... if I make that README.md will it still work?
[10:37] <marcoceppi_> Fantastic!
[10:37] <marcoceppi_> natefinch: well, it "will" work but you'll get a bunch of proof errors about it being boilerplate
[10:37] <marcoceppi_> also, for the sake of the rest of this conversation, what verions charm tools are you running? charm version should tell you
[10:38] <natefinch> I just mean, if I change the extension it'll still show up.  I'll modify the contents
[10:38] <marcoceppi_> natefinch: oh, yeah, that's all you need to do
[10:38] <marcoceppi_> in fact, proof will complain if you have a .ex instead of a .md in your charm
[10:38] <natefinch> ok
[10:38] <natefinch> charm versoin says it's not a subcommand
[10:38] <natefinch> version that is
[10:38] <marcoceppi_> natefinch: you're not using the ppa version
[10:38] <marcoceppi_> safe yourself a lot of time
[10:39] <marcoceppi_> add ppa:juju/stable
[10:39] <marcoceppi_> install charm-tools from there
[10:40] <marcoceppi_> its' the differences between 1.0.0 and 1.2.10 are huge
[10:40] <marcoceppi_> including the readme.ex going from RST to MD
[10:40] <marcoceppi_> so hopefully you haven't written too much code
[10:41] <natefinch> marcoceppi_: barely any :)
[10:41] <marcoceppi_> whew, I'd go so far as to saying once updated, just run charm create again
[10:41] <marcoceppi_> you'll get like an icon template a better readme template, better metadata templates
[10:41] <marcoceppi_> better boilerplates
[10:41] <marcoceppi_> and better proof command
[10:41] <marcoceppi_> basically, better everything
[10:41] <natefinch> awesome,
[10:42] <natefinch> I'll do that once apt-get update finishes
[10:42] <marcoceppi_> \o/
[10:42] <natefinch> I wish adding a ppa always ran update-get update, at least to get the stuff from the ppa, sheesh
[10:43] <natefinch> btw, is there a way to have different charms for different architectures?  I want to make the charm in Go, but you have to have a different executable for each architecture
[10:43] <marcoceppi_> yeah, like to a partial apt-get update for that source to get the delta
[10:44] <marcoceppi_> that's, no - not at the moment
[10:44] <marcoceppi_> You could just compile the hooks during the install hook
[10:44] <natefinch> yeah
[10:44] <marcoceppi_> to make the runtime for the hooks at deploy time
[10:44] <marcoceppi_> but that seems, heavy
[10:44] <marcoceppi_> otherwise you'd have to package /hooks/$ARCH/<file> and just build a light weight dispatcher for each in /hooks/
[10:45] <natefinch> yeah, that's what I was thinking.  Also kinda heavy, since you'd need an executable per architecture
[10:45] <marcoceppi_> but the execs wouldn't be very big
[10:46] <marcoceppi_> it's a trade off, "JIT"-ish compilation, or possible missing archs
[10:46] <natefinch> go execs are bigger than normal, like 4 megs generally
[10:46] <marcoceppi_> could combine the two ideas
[10:46] <natefinch> how bad is it to have an architecture-specific charm?
[10:46] <marcoceppi_> pretty bad considering it doesn't exist yet
[10:46] <natefinch> so.... here's the thing.  I'm writing a charm for discourse
[10:46] <natefinch> discourse says it requires 64 bit ubuntu
[10:46] <natefinch> marcoceppi_: https://github.com/discourse/discourse_docker
[10:47] <marcoceppi_> well, it's a support requirement. I definitely have run discourse on i386
[10:47] <natefinch> *nod* I know, just sayin'
[10:47] <marcoceppi_> could do both, have hooks/$arch where arch is [amd64, i386] then have the install hook attempt to compile the src hooks if the arch isn't pre-loaded
[10:48] <natefinch> marcoceppi_: that's not a bad idea
[10:48] <marcoceppi_> so, you could have arch dependant but that's going to be sad news for people running power, arm, i386, etc
[10:48] <natefinch> yep
[10:48] <marcoceppi_> and we try to avoid as many sad pandas as possible
[10:49] <marcoceppi_> can you compile go for, say i386 on an amd64 machine?
[10:49] <natefinch> yes... cross compile works 100% awesome
[10:50] <natefinch> except for power which requires gccgo
[10:50] <marcoceppi_> interesting. So, you /could/ and I mean it'd be beat up pretty badly in review, have it just for amd64 and provide instructions or a simple make file for people to pull the charm, build the hooks in their arch of choice, then deploy local
[10:50] <natefinch> not my fault juju keeps supporting all these wacky-ass platforms :)
[10:51] <marcoceppi_> hah, well we strongly frown on provider specific charms, I think only one has made it past review
[10:51] <marcoceppi_> so I can only imagine (well I can imagine, actually) we'd really beat up a charm that only supported an architecture
[10:52] <natefinch> heh
[10:52] <marcoceppi_> but, I wouldn't let that slow you down
[10:52] <natefinch> yeah, but it would work on every provider!  And you're not supposed to care about the OS
[10:52] <natefinch> just sayin' :)
[10:53]  * marcoceppi_ places head on his hands, and looks dreamily in to the distance
[10:56] <natefinch> what is charm home?
[10:56] <marcoceppi_> in relation to?
[10:57] <natefinch> charm create
[10:57] <natefinch> also, are there no help docs?  charm help create doesn't seem to do anything
[10:57] <marcoceppi_> ah, that's just the directory you want the charm to be created in, for example `charm create foo /tmp/baz` will give you /tmp/baz/foo
[10:57] <marcoceppi_> charm create -h
[10:58] <natefinch> ahh
[10:58]  * marcoceppi_ needs to make a help helper
[10:58] <natefinch> I saw the -h under charm create usage, but it didn't say what -h was for, so I didn't realize it was for help :)
[10:59]  * natefinch has chicken and egg problems
[11:01] <natefinch> gah... I want a way to spit out the charm in *this* directory, not a subdirectory :/
[11:01] <natefinch> apologies for the complain-a-thon :)
[11:01] <marcoceppi_> omit the charmhome
[11:01] <marcoceppi_> oh
[11:02] <marcoceppi_> you mean like, mkdir -p /charm/trusty/my-charm; cd $_; charm create .
[11:03]  * marcoceppi_ narrows eyes at the complain-a-thon
[11:04] <natefinch> exacrtly... I expected charm create . to create it *here* not in a subdirectory.  If I wanted a subdirectory, I'd do charm create ./foo
[11:05]  * natefinch does charm create ../
[11:05] <natefinch> ...which doesn't work because the directory already exists
[11:06] <marcoceppi_> bugs are "welcome"
[11:06] <natefinch> marcoceppi_: point me the way :)
[11:06] <marcoceppi_> though, I'm going to say this is more the operator cutting himself with paper
[11:06] <marcoceppi_> natefinch: http://launchpad.net/charm-tools :D
[11:07] <natefinch> I dunno... there's ought to be SOME way to dump a charm in the current directory
[11:09] <marcoceppi_> I can add a --natefinch-really-wanted-this flag :P
[11:10] <natefinch> it's like git clone ... git clone juju  creates a juju directory here, if you instead do git clone juju ., it dumps the repo in the current directory, and if you do git clone juju ./foo, it creates the foo subdirectory and dumps the files in there.
[11:11]  * marcoceppi_ nods
[11:11] <marcoceppi_> we can make charm create smarter
[11:17] <natefinch> marcoceppi_: btw, I love the charm tools logo.  That's clever.
[11:17] <marcoceppi_> natefinch: hah, thanks!
[11:18] <natefinch> btw, what does this mean: Failed to find discourse in apt cache, creating an empty charm instead.
[11:19] <marcoceppi_> natefinch: it attempts to look the charm name up in apt, if it finds it charm create will pre-fill metadata.yaml with some details
[11:19] <marcoceppi_> it's something that was moved over from the original charm-tools. i'll probably drop it eventually because it just flat out doesn't work on mac or windows machines
[11:19] <marcoceppi_> and it's not a huge value add
[11:20] <natefinch> yeah.... I could see it being something that happens with a flag, like --apt or something, if you know that's what you want it to do, but I could see names that happen to match things in apt that would not be the correct behavior
[12:11] <natefinch> the instructions to install docker involve piping the output of wget https://get.docker.io to sh... seems legit
[12:17] <marcoceppi_> natefinch: yeah, I hate those. Same with RVM
[12:18] <marcoceppi_> there's not much you can really do. Security is a thing fo the pass when you got a `wget | sudo sh -` one liner
[12:18] <marcoceppi_> s/pass/past/
[12:18] <natefinch> yep
[12:18] <natefinch> le sigh
[12:43] <lazyPower> security? thats still a thing?
[12:45] <marcoceppi_> no, nooo...
[12:45]  * marcoceppi_ pulls the wool over lazyPower's eyes
[12:46] <lazyPower> ahhh ignorance is bliss
[12:49]  * hazmat adds some bliss to morning coffee
[12:49] <hazmat> natefinch, you can also grab from their deb repo..
[12:51] <marcoceppi_> that's a way friendlier idea
[13:19] <natefinch> hazmat: that sounds much nicer
[13:30] <hazmat> natefinch yeah.. its wget | gpg --verify --sign | sudo sh - one liner ;-)
[13:33] <lazyPower> haha
[13:33] <lazyPower> #trolledhard
[14:17] <mhshams_> hi, got some basic questions about juju, i hope here is the right place to ask.
[14:17] <mhshams_> Does juju provides an API, (REST for example), so that one can use it from another application ?
[14:19] <marcoceppi_> mhshams_: yes, it's a websocket
[14:21] <mhshams_> @marcoceppi : where can I found some document or other info about it ?
[14:22] <marcoceppi_> mhshams_: https://github.com/juju/juju/blob/master/doc/api.txt
[14:22] <marcoceppi_> Not the friendliest of docs
[14:22] <marcoceppi_> mhshams_: if you're using Python, there's a jujuclient python package which does all that ofr you
[14:38] <mhshams_> @marcoceppi thanks for the hints.
[14:57] <jamespage> someone was asking about the ceph charms and using different networks for client access vs cluster resyncs
[14:58] <jamespage> I have some branches ready for review that implement this if anyone wants to test
[14:58] <lazyPower> designated: ^
[14:58] <marcoceppi_> jamespage: ah, awesome, yeah I told himto ask on the list
[14:58] <marcoceppi_> oh good he's around!
[15:20] <nuclearbob> can anybody here help me with something in the juju-tools branch?
[15:31] <marcoceppi_> nuclearbob: juju-tools?
[15:54] <nuclearbob> marcoceppi_, sorry, got distracted.  Can I use canonistack-sshuttle from a canonistack instance?  That seems to be causing me problems
[16:24] <rbasak> jamespage: looks like there are about ~21 bugs for Trusty SRU 1.18.1 -> 1.18.4.
[16:24] <rbasak> jamespage: do we want to individually do SRU paperwork for each of those then?
[16:38] <designated> jamespage, where can i find those charms you wrote?
[16:41] <jamespage> designated, one sec
[16:41] <jamespage> designated, top three branches here - https://code.launchpad.net/~james-page
[16:42] <jamespage> designated, I have further work todo to enable all things that relate to ceph to use the public network addresses
[16:42] <jamespage> but that should give you a feel on what's possible
[16:43] <jamespage> rbasak, I'd really hope we don't need todo that
[17:01] <designated> jamespage, thank you
[17:34] <marcoceppi_> jose: I need ubuntu on air stuff setup
[18:32] <jose> marcoceppi_: I'm here now, what's up?
[18:34] <lazyPower> jose: troubleshooting I on charm school schedule today
[18:34] <lazyPower> can we get UoA setup for it?
[18:34] <jose> lazyPower: yeah, I know
[18:34] <jose> erm, you want me to host the hangout?
[18:34] <jose> sure, I can do that
[18:34] <lazyPower> i know nothing of this UoA sorcery
[18:34] <lazyPower> you are the holder of the magic beans on that project sir
[18:34]  * mbruzek wants to learn
[18:35] <jose> lemme grab a quick snack and I'll be around
[18:35] <mbruzek> I want to learn how to do the UoA
[18:35] <jose> you gotta have some secret codes for that ;)
[18:50] <lazyPower> jose: we g2g?
[18:50] <jose> lazyPower: I was having lunch, almost ready
[18:50] <mbruzek> Ubuntu on Air
[18:51] <mbruzek> With Jose !
[18:52] <mbruzek> we need a url
[18:57] <marcoceppi_> mbruzek: looks like lazyPower is doing this one solo ;)
[18:57] <lazyPower> marcoceppi_: you got an invite
[18:57] <marcoceppi_> lazyPower: can you pm me the URL, it's not ringing
[18:58] <axisys> is there a view only url?
[18:58] <lazyPower> axisys: ubuntuonair.com
[18:58] <axisys> http://ubuntuonair.com/ is showing the one from yesterday
[18:59] <axisys> I guess it will refresh when ready?
[18:59] <lazyPower> You will need to refresh the page, but it will update with teh current session when its active.
[18:59] <axisys> I see it now.. thanks
[19:01] <mbruzek> marcoceppi_, are you OK?
[19:03] <lazyPower> stream is live
[19:03] <axisys> yep
[19:03] <lazyPower> Feel free to ask questions!
[19:03] <arosales> stream looks good
[19:16] <axisys> I had to run this to make juju lxc work
[19:16] <axisys> https://gist.githubusercontent.com/allomov/11373534/raw/bd430c2c92116340f2fa5d124b769c555946609d/solution.sh
[19:16] <axisys> on Apr 28
[19:16] <lazyPower>  axisys interesting. what was the use case that installing juju-local didn't resolve?
[19:17] <axisys> right
[19:17] <axisys> I am going to try again after this hangout
[19:17] <axisys> with may be newer juju
[19:18] <axisys> referring to this thread
[19:18] <axisys> https://lists.ubuntu.com/archives/juju/2014-April/003749.html
[19:22] <axisys> so what does bootstrap do? do you need to run bootstrap before doing any juju <environment> ?
[19:23] <mbruzek> https://juju.ubuntu.com/docs/config-LXC.html
[19:25] <axisys> would be nice if you can start a hadoop local?
[19:25] <axisys> I was never been successful
[19:27] <axisys> here is my thread from apr
[19:27] <axisys> https://lists.ubuntu.com/archives/juju/2014-April/003737.html
[19:27] <axisys> I have 8G ram
[19:27] <axisys> and 1TB disk
[19:27] <axisys> running trusty
[19:28] <lazyPower> asanjar: axisys was asking about hadoop on local
[19:28] <asanjar> axisys: how may I help
[19:28] <jose> axisys: that's a fixed bug
[19:30] <axisys> ok, I will try again.. thanks for your help
[19:31] <axisys> good to know.. I use vagrant a lot
[19:43] <lazyPower> Do any of our viewers have any burning issues they've run into that they would like to see how a ~charmer would debug?
[19:52] <axisys> thanks a lot!
[20:02] <lazyPower> Thanks everyone for attending. Great success of a charm school today.
[20:04] <mbruzek> jose you still there?
[21:05] <dpb1> Hi -- I'm trying to give a user permission to bootstrap and destroy a local env through sudo.  what command do I need to allow?
[21:17] <roadmr> dpb1: hm, my tests show that for the initial bootstrap, juju basically does "sudo /bin/bash -s"; so maybe you'll have to allow bash invocation like this
[21:17] <roadmr> dpb1: given that, the user will essentially be able to do anything as root, so you may as well give him access to everything I think :/
[21:18] <dpb1> roadmr: :(
[21:18] <roadmr> dpb1: I may be wrong, though, my investigation was a bit primitive
[21:18] <dpb1> roadmr: ok, thanks.  I will probably go another route then. :0
[21:18] <roadmr> dpb1: (I just did a juju bootstrap and looked in /var/log/auth.log to see which commands get run by sudo)
[21:18] <dpb1> roadmr: thankfully the jenv files are portable (with some catches), so that works
[21:19] <roadmr> dpb1: cool :) sorry I couldn't provide a nicer answer
[21:19] <dpb1> roadmr: np, appreciate the help
[22:15] <themonk> where should i put lock file so that between restarts of lxc it will persist
[22:57] <themonk> lazyPower: hi