[09:52] <jcverdie> Hi all, does anyone know how/if I can deploy a juju framework on a pre-existing amazon instance? I have reserved instances which I'd like to use for this
[09:52] <jcverdie> e.g. instead of adding machines which are created by juju, using my own machines...
[10:15] <axw__> jcverdie: yes, using manual provisioning. it works well in the latest dev release, it's a bit flakey in the stable release
[10:16] <axw__> jcverdie: I have to run, but the docs are here: https://juju.ubuntu.com/docs/config-manual.html
[10:16] <axw__> (a little out of date, but mostly still true)
[10:17] <jcverdie> thanks Im gonna look
[13:36] <tomixxx> hi, when i try to assign multiple charms to the same machine, i get an error in agent-state-info. it seems that the machine cannot download some things from http://cloud-images.ubuntu.com. Please have a look on the output of the terminal when i call "juju status": http://pastebin.ubuntu.com/6873041/ As u can see in the output, i have tried it now several times but every time the lxc container failed to create
[14:09] <tomixxx> can i install "juju" if i have installed "juju-core"?
[14:20] <lazyPower> tomixxx: juju is a metapackage, It won't harm anything.
[14:21] <tomixxx> lazyPower: i still have the problem that i cannot assign multiple charms to the same machine
[14:21] <lazyPower> do you have the juju-local package installed?
[14:23] <tomixxx> lazyPower: no i have not installed this package
[14:23] <lazyPower> tomixxx: you will need to destroy your environment, install the juju-local package, then rebootstrap your local environment
[14:24] <lazyPower> tomixxx: https://juju.ubuntu.com/docs/config-LXC.html
[14:25] <tomixxx> lazyPower: ok i will do this
[14:25] <lazyPower> Best of luck. Ping me if you get stuck
[14:25] <tomixxx> ty
[15:03] <tomixxx> lazyPower: i have followed your guide and then i have called "juju deploy mysql --to lxc:0" again but i got the same error
[15:04] <lazyPower> tomixxx:  shoot me a pastebin of your juju-status output again please
[15:05] <tomixxx> lazyPower: http://pastebin.ubuntu.com/6873459/
[15:06] <jcverdie> Hi I'd like to deploy to a EC2 instance which I already have, but as it's a regular EC2 i have to use PEM to connect to it, and I can't find how to do it with juju ?
[15:07] <lazyPower> tomixxx: does the machine you're bootstrapping juju on have internet connectivity?
[15:07] <tomixxx> lazyPower: no
[15:07] <lazyPower> jcverdie: what you're attempting to do is manual provisioning
[15:07] <lazyPower> jcverdie: thats still very much in beta phases at this point
[15:07] <jcverdie> lazyPower: I did it (juju switch null)
[15:07] <jcverdie> but juju bootstrap fails : ERROR failed to enable bootstrap storage: failed to create storage dir: exit status 255 (Permission denied (publickey).)
[15:07] <lazyPower> tomixxx: you will need to get that machine to have connectivity before creating the containers will work. Its dependent on the cloud-tools to bootstrap the node.
[15:08] <jcverdie> because when i ssh it i have to do ssh -i mykey.pem
[15:08] <tomixxx> lazyPower: My architecture is as follow: my maas-server is connected to two nodes via switch. the maas-server has to network interfaces: one interface is connected to the switch and the other interface connects me to the internet
[15:08] <lazyPower> jcverdie: ok, so its an ssh auth error. Can you add your sshkey to the authorized_keys on the host?
[15:08] <jcverdie> and I don't know how to tell juju to do it
[15:09] <jcverdie> i'll try
[15:09] <lazyPower> tomixxx: The only resolution here is to get internet connectivity to the host that is bootstrapping those lxc containers, i'm sorry.
[15:10] <lazyPower> jcverdie: ssh-copy-id -i $PATH_TO_KEY ubuntu@host
[15:10] <tomixxx> lazyPower: lazyPower: ok
[15:11] <jcverdie> lazyPower: i have the public .pem and my own public rsa in the authorized keys on the server
[15:12] <lazyPower> Ok that should be good. When you ssh to that host, you no longer have to specify the key, correct?
[15:12] <jcverdie> lazyPower: right, but juju bootstrap still fails :(
[15:13] <lazyPower> jcverdie: still on object storage?
[15:13] <jcverdie> yep, exactly the same error
[15:13] <lazyPower> hmm.. is juju using a different keypair than what you have provided to the host?
[15:14] <jcverdie> no, i've put my 3 keys on the server
[15:14] <jcverdie> could it be somethign wrong in my environments.yaml ?
[15:14] <lazyPower> well, one second
[15:14] <lazyPower> is juju trying to use your personal user and not ubuntu? there's a config flag for the user
[15:14] <lazyPower>     # bootstrap-user:
[15:15] <jcverdie> i've set bootstrap-host, bootstrap-user
[15:15] <lazyPower> I'm not sure then :| that should have been a green light to juju
[15:16] <jcverdie> and there's a storage-auth-key which I haven't set but it's there
[15:16] <lazyPower> yeah, you should be fine without that. I haven't used the configuration field in my trial tests of the manual provider
[15:16] <jcverdie> :(
[15:18] <lazyPower> jcverdie: i'll keep brain bending on this, but i'm not sure why that's not working. can you pastebin me all the non-sensitive stuff relating to the null provider from your environments.yaml?
[15:18] <jcverdie> sure
[15:19] <jcverdie> did you get it ?
[15:19] <lazyPower> tomixxx: another thing, make sure your lxc bridge device is pointed at the proper ethernet port to get internet connectivity too. The bootstrap process will be downloading bits from the itnernet as well.
[15:19] <lazyPower> i did
[15:20] <jcverdie> brb
[15:20] <tomixxx> lazyPower: how do i check this?
[15:21] <lazyPower> tomixxx: lxcbr0 (the default bridge device created by juju) is by default attached to eth0 ( i'm pretty sure ) and if you need to change that
[15:22] <lazyPower> the configuration of the bridge interface is found in /etc/lxc/default.conf - but it doesn't look like it has the bridged interface there...
[15:23] <tomixxx> lazyPower: ok, in my case, eth0 has no internet, but eth1 has
[15:23] <lazyPower> tomixxx: ah its in /etc/default/lxc
[15:24] <lazyPower> you can configure the LXC virtual bridge device / vnet in that file. it will require a restart to take hold however. I don't believe that sudo service networking restart will affect the LXC bridge device.
[15:25] <tomixxx> lazyPower: hmm, LXC_ADDR="10.0.3.1"
[15:26] <lazyPower> if you don't want to reach the containers that LXC is spinning up on your network thats fine. Juju's local provider occupies that ip range by default
[15:26] <lazyPower> if do you DO however want to reach your LXC containers, you will need to do some configuration magic
[15:27] <tomixxx> lazyPower: i have to use lxc-containers because i have only 2 nodes
[15:28] <lazyPower> tomixxx: I understand. I have an intranet deployed in my house using juju and lxc containers.
[15:28] <lazyPower> it works really well, but I don't know that its something I would want to use in production. It does get a bit wonky here and there.
[15:57] <tomixxx> lazyPower: do i need to change my "interfaces" file in order to enable the cloud-nodes to access the internet?
[15:57] <lazyPower> I don't believe so
[15:58] <tomixxx> lazyPower: Ok, so there is no need to configure a "bridge" or sth like that?
[15:58] <lazyPower> thats what i pointed you at, the LXC bridge configuration
[15:59] <tomixxx> lazyPower: k, but what do i have to change there?
[16:01] <lazyPower> tomixxx: not knowing how your network is configured, or the host machine, thats difficult to say.
[16:02] <tomixxx> lazyPower: my network is configured as follow: http://pastebin.ubuntu.com/6873735/
[16:03] <tomixxx> lazyPower: eth1 connects me to the internet and eth0 connects me to the cloud-nodes
[16:03] <lazyPower> tomixxx: there's a discussion on this in the forums, read through this - http://ubuntuforums.org/showthread.php?t=2137446
[16:05] <tomixxx> lazyPower k, ty
[17:24] <tomixxx3> i have tried to NAT the traffic from the LXC bridge to the internet-capable network interface eth1: http://pastebin.ubuntu.com/6874102
[17:24] <tomixxx3> the problem is it does not work
[17:25] <tomixxx3> i get the same error when i call "juju status" -> "failed to get https://cloud-images.ubunt.com/..."
[17:33] <tomixxx3> lazyPower: do i have to rebootstrap juju if i have changed lxc bridge settings?=
[17:34] <marcoceppi_> tomixxx3: possibly
[17:34] <tomixxx3> marcoceppi: i have only modified the "interfaces" file, in detail. i added some port-forwarding lines
[17:35] <tomixxx3> marcoceppi: i have already restarted the maas-server but it seems still not work because i cannot create lxc containers
[17:35] <marcoceppi_> tomixxx3: didn't realize you were driving lxc with maas
[17:36] <tomixxx3> marcoceppi: i have deployed juju on top of maas because i want to use openstack
[17:36] <tomixxx3> marcoceppi: in order to deploy juju charms, i need to have multiple charms on a single machine
[17:36] <tomixxx3> marcoceppi: and therefore, i need lxc containers
[17:37] <marcoceppi_> tomixxx3: right, but are the LXC containers in maas, or are you using juju deploy --to lxc:# ?
[17:37] <tomixxx3> i use "juju deploy XXX --to lxc:x"
[17:37] <marcoceppi_> tomixxx3: you won't need to re-bootstrap
[17:38] <marcoceppi_> but I'm not sure if anything else is required
[17:39] <tomixxx3> k, the problem is, when i try to deploy sth i call "juju status" after it and then i can see that the creation of the lxc container failed because the container was not able to download sth
[17:39] <marcoceppi_> right, so some networking issue
[17:39] <marcoceppi_> what version of juju are you using?
[17:40] <tomixxx3> if its "agent-version" it is 1.16.5.1
[17:41] <tomixxx3> marcoceppi: the output of "juju status" looks like the following: http://pastebin.ubuntu.com/6874176
[17:41] <marcoceppi_> tomixxx3: yeah, I think lxc bridge stuff was patched in 1.17.2, but the person who would know that is not online right now
[17:41] <tomixxx3> marcoceppi: as u can see, i have tried it multiple times
[17:42] <tomixxx3> marcoceppi: 0/lxc/4 was last try ;)
[17:43] <tomixxx3> maroceppi: so, do you mean it make sense to go to version 1.17.2?
[17:45] <tomixxx3> marcoceppi: i mean latest, not last
[17:49] <marcoceppi_> tomixxx3: for LXC stuff, yes
[17:49] <marcoceppi_> it's something thats actively beinig worked on in juju at the moment
[17:54] <tomixxx3> marcoceppi: i fear that the main problem is that i have made some mistake when configuriering or installing sth
[18:08] <cjohnston> Is there a way to get juju to automatically clean up the security group spam it creates?
[18:17] <marcoceppi_> cjohnston: no, but there is a tool you can compile with go to do it
[18:18] <marcoceppi_> cjohnston: though it's a bit...over zealous