[09:23] <ehg> hi. how would we go about upgrading from goju 1.10 to 1.11.4? would there be anything to keep in mind?
[09:26] <bbcmicrocomputer> ehg: sudo add-apt-repository ppa:juju/devel; sudo apt-get update; sudo apt-get upgrade
[09:27] <ehg> bbcmicrocomputer: is anything likely to break? :)
[09:27] <bbcmicrocomputer> ehg: I believe upgrading deployed 1.10 environments work now
[09:27] <bbcmicrocomputer> ehg: no, more things are likely to work :)
[09:28] <ehg> awesome
[09:30] <raywang> hello, I just try to juju-core (1.11.4-1~1514~raring) on aws, when I bootstrap, and check its status, it always report "error: no instanerror: no instances found", what is that? thanksces found
[09:59] <ehg> bbcmicrocomputer: is juju upgrade-juju the command we should run once our client is at 1.11.4?
[10:00] <bbcmicrocomputer> ehg: yes, I believe so.. although I've never tested the upgrade path personally
[10:01] <ehg> bbcmicrocomputer: thanks, we'll try it!
[10:10] <bbcmicrocomputer> raywang: sounds a bit like https://bugs.launchpad.net/juju-core/+bug/1176961.. you may need to clear out your Juju S3 bucket and start again
[10:10] <_mup_> Bug #1176961: Wedge an environment by using the other management tools <juju-core:Triaged> <https://launchpad.net/bugs/1176961>
[10:11] <bbcmicrocomputer> raywang: alternatively changing the 'control-bucket' parameter in .juju/environments.yaml file and bootstrapping should work
[10:11] <raywang> bbcmicrocomputer, thanks, but how do I clear my S3 bucket out?
[10:12] <raywang> bbcmicrocomputer, thanks, but i have no idea change  ' 'control-bucket' to what? :)
[10:13] <bbcmicrocomputer> raywang: I personally use package 'libnet-amazon-s3-tools-perl' to delete things from S3 at the cli
[10:14] <bbcmicrocomputer> raywang: a combination of s3ls, s3rm and s3rmbucket
[10:15] <bbcmicrocomputer> raywang: e.g 's3ls juju-xxxxxxxxx' (control-bucket from .juju/environments.yaml)
[10:15] <bbcmicrocomputer> raywang: just alter the last digit on the value to another hex value (0-9, a-f) :)
[10:16] <raywang> bbcmicrocomputer, ah, i see, it's very helpful :)
[10:19] <raywang> bbcmicrocomputer, hit another error :(
[10:19] <raywang> error: cannot start bootstrap instance: cannot set up groups: cannot revoke security group: Source group ID missing. (MissingParameter)
[10:20] <bbcmicrocomputer> raywang: what's the output when you run 'juju -v bootstrap' ?
[10:20] <raywang> bbcmicrocomputer, http://pastebin.ubuntu.com/5924749/
[10:22] <bbcmicrocomputer> raywang: hmm.. have you tried clearing out security groups before bootstrapping?
[10:22] <raywang> bbcmicrocomputer, let me check
[10:22] <bbcmicrocomputer> raywang: sometimes previous bootstrap junk is left
[10:23] <raywang> bbcmicrocomputer, destroy-environment doesn't do that for me?
[10:23] <bbcmicrocomputer> raywang: it's not perfect :)
[10:31] <raywang> bbcmicrocomputer, oh, just check, i don't have a secgroup  :(
[10:33] <bbcmicrocomputer> raywang: hmm, I'm not sure then if it's failing on a clean aws environment
[10:33] <bbcmicrocomputer> raywang: I'd suggest filing a bug
[10:33] <raywang> bbcmicrocomputer, ok, i'm try to see if I can find something new about the secgroup
[10:37] <bbcmicrocomputer> raywang: what AWS region are you deploying to?
[10:37] <raywang> bbcmicrocomputer, ap-southeast-1
[10:37] <bbcmicrocomputer> raywang: https://lists.ubuntu.com/archives/juju-dev/2013-April/000921.html
[10:38] <bbcmicrocomputer> raywang: seems to be same problem
[10:38] <bbcmicrocomputer> raywang: perhaps try changing regions?
[10:38] <raywang> bbcmicrocomputer, i see, thank for the findings :)
[10:40] <raywang> bbcmicrocomputer, actually, i managed to bootstrap, but got 'no instance found', will try another region as you suggested
[10:47] <raywang> hit another error "error: provider storage is not writable"...
[10:54] <bbcmicrocomputer> raywang: do you need to clear out or change control-bucket?
[10:55] <raywang> bbcmicrocomputer, yes, i clear out control-bucket, secgroup, and then switch to new region (us-east-1), and that's what I got after that
[10:58] <ehg> woohoo upgrade worked
[10:59] <ehg> does anyone know how to get rid of machines that failed come up?
[10:59] <ehg> e.g.
[10:59] <ehg> "58":
[10:59] <ehg>     agent-state-info: '(error: use of closed network connection)'
[10:59] <ehg> the service won't get destroyed, and i can't even terminate that machine
[11:00] <ehg> i'm willing to edit whatever DB it's stored on in the bootstrap mode - mongo?
[11:00] <ehg> although when i tried to do that last time i couldn't connect because of SSL issues, enterprise mongo client blahness
[11:03] <bbcmicrocomputer> raywang: tried using a different control-bucket id, not sure if Juju locks the S3 bucket to a particular region... didn't think it did
[11:04] <raywang> bbcmicrocomputer, ok, i would like to try it :)
[11:05] <pavel> hello
[11:05] <pavel> guys, does 'expose' work for you right now?
[11:05] <pavel> I run juju expose and don't see any message in the log
[11:06] <pavel> and of course it doesn't get exposed, though in the juju status id shows 'exposed: true'
[11:11] <raywang> bbcmicrocomputer, it's interesting, I change control-bucket id, and it works, but I got another error, http://pastebin.ubuntu.com/5924870/, and i comment out the region, and re-run bootstrap, now it seems works
[11:11] <bbcmicrocomputer> pavel: is this using aws?
[11:11] <pavel> bbcmicrocomputer, yes
[11:15] <pavel> bbcmicrocomputer, now I'm going to rebuild my env and try again
[11:15] <drj11> has the config.yaml file format changed between 0.10 and 0.11 (the config file we specify with --config when we deply?)
[11:16] <pavel> drj11, I'm not sure about exact version, but at some moment there was difference to include or not service name
[11:16] <pavel> e.g mysql: {}, or just {}
[11:17] <pavel> in 1.11.4 you have to include service name
[11:17] <pavel> in 1.10 you should not do this
[11:18] <drj11> thanks pavel. that seems like quite a big change to me.
[11:18] <drj11> stuff seems to be working now.
[11:20] <drj11> althought it's very bad news for us, because we use new service names whimsically. We need to change our YAML files for every new service name
[11:20] <pavel> drj11, np, I was confused by this too
[11:21] <drj11> pavel: it's very handy to have someone experience the same problem
[11:22] <pavel> drj11, I didn't thought about service names in that way, this means you can't deploy two different services with the same config
[11:23] <drj11> indeed you cannot. which is the kind of thing we do all the time (for debugging for example)
[11:24] <pavel> idea here is that if you want to deploy another service perhaps it has different config
[11:24] <pavel> otherwise why you need this
[11:26] <drj11> pavel: all sorts of reasons. totally plausible that difference services use the same connection IP and password to access a shared database, or shared redis
[11:26] <drj11> s/difference/different/
[11:28] <pavel> drj11, yes, it's weird that there is now such option
[11:30] <pavel> bbcmicrocomputer, after destroying and bootstrapping expose works again
[11:30] <bbcmicrocomputer> pavel: hmm..not good
[11:31] <pavel> bbcmicrocomputer, yes, totally weird
[11:34] <pavel> 'juju add-unit mysql --to 1' and 'juju add-unit mysql --to lxc:1'
[11:34] <pavel> does I get it right that in first case it will try to deploy mysql in the global environment on machine 1?
[11:41] <bbcmicrocomputer> pavel: yes
[11:41] <pavel> bbcmicrocomputer, thanks
[12:09] <pavel> Since containers already work, are there any plans for Digital ocean support?
[13:55] <marcoceppi> pavel: There's an "SSH" provider coming to juju-core which will add support to hosts like Digital Ocean
[13:55] <pavel> marcoceppi, cool
[14:07] <Tantor> Hello. I'm trying to get juju working on ubuntu 12.04
[14:08] <Tantor> However when I do fw juju deploy wordpress it keeps bugging me with an Invalid SSH key error
[14:36] <arosales> Tantor, which version are you running?
[14:36] <Tantor> the one from the ubuntu ppa
[14:36] <marcoceppi> Tantor: do you have an ssh key generated?
[14:36] <Tantor> Yes
[14:36] <Tantor> Without it MAAS won't run
[14:37] <marcoceppi> Tantor: please pastebin the errors you're getting
[14:37] <Tantor> ok
[14:37] <Tantor> hold on
[14:44] <arosales> bbcmicrocomputer, fyi looks like your up on the review queue
[14:44] <bbcmicrocomputer> arosales: yep :)
[14:44] <arosales> the review queue also looks like it could use some love
[14:46] <jseutter> Is there a pattern for how are upgrades to a service done using a charm?  Any examples?
[14:46] <arosales> hazmat, jamespage, negronjl ^ if you can find any spare cycles for the review queue
[14:46] <arosales> bbcmicrocomputer, :-)
[14:56] <marcoceppi> jseutter: there are two ways to upgrade a running service in a charm. One is via configuration (ie: `juju set <service> version=1.1.0`) or via upgrade-charm hook which is run when `juju upgrade-charm <service>` is executed
[14:57] <marcoceppi> the former is considered a "best practice" when it comes to writing a charm
[14:57] <jseutter> marcoceppi: when would you choose one over the other?
[14:58] <marcoceppi> jseutter: there are very few times when you wouldn't have a "version" configuration option
[14:58] <marcoceppi> For the majority of charms, that's likely the best way to go
[14:59] <jseutter> marcoceppi: do you know of some charms that do this?  I'd like to have a look
[15:04] <hazmat> arosales, hit 3 fwiw
[15:04] <arosales> hazmat, much appreciated. thank you sir :-)
[15:11] <weblife> Tanto I was disconnected for a bit but have youmade a ssh key?
[15:12] <weblife> or has someone helped you
[15:42] <jose> guys, just so you know the postfix charm's been there for 3 weeks but its age changed when hazmat posted a comment :)
[15:43] <hazmat> hi jose
[15:43] <hazmat> so what's the use scenario for the charm?
[15:43] <jose> hey, hazmat, have a min?
[15:43] <jose> I wanted to talk about that
[15:44] <hazmat> jose, me too, i'm in a meeting for the next 15, but free after that
[15:44] <jose> I can wait for sure :)
[16:01] <hazmat> jose, free
[16:01] <hazmat> jose, chat or g+?
[16:01] <jose> hazmat: which one would you prefer?
[16:01] <hazmat> jose, g+ sounds good to me, /me sends priv link
[16:02] <hazmat> actually public. in case anyone else is interested in postfix charm.. jose.. https://plus.google.com/hangouts/_/01fbf06fe36cb7b88119171e3f86c98e3b56b421?hl=en
[17:22] <weblife> @marcoceppi  I ave node-app patched to install from source but  the application isn't spinning up at startup. Try and submit the patch when I figure out why and fix it.   Just wanted to let you know since you said conntact you about it.  I didn't forget you just got tied up with my wedding plans.
[17:22] <weblife> I have
[17:23] <weblife> I'm in a typo mood today ;:x
[17:24] <marcoceppi> weblife: no problem, take your time with it
[18:31] <Tantor> I got juju running. But when I do juju status. I see my 2 maas nodes as instance 0 and 1, but also instances 2-9 as pending. I don't have any other nodes installed though. Are those instances normal?
[18:31] <Tantor> Or should I remove them?
[18:33] <weblife> @Tantor I would think it depends on the charms your using and what you did to deploy those instances
[18:34] <weblife> I will be right back.
[18:37] <weblife> Its because the script is based on chris lea's or apt-get install specific instructs.  I am somewhat new to doing system processes so I need to look into what the differences would be between doing a apt-get install vs. make install.
[18:38] <Tantor> !bin
[18:38] <weblife> sorry that was at @marcoceppi
[18:39] <Tantor> My problem is this: http://pastebin.com/WVdnvHed
[18:39] <Tantor> There are two instances, the instances 2-9 already disappeared
[18:39] <Tantor> There are 2 machines, srv7 and srv8
[18:39] <Tantor> However the service juju-gui shows as a public address srv9
[18:39] <Tantor> There is no srv9
[18:40] <Tantor> So I think that I messed up my config
[18:40] <Tantor> How can I remove this srv9?
[18:43] <Tantor> Tried destroy-environment too, but that doesn't help
[18:43] <weblife> @Tantor its probably because you spun service up and destroyed prior I am thinking.  These wont disappear for some time and you can't remove them.  When looking in you ec2 console make sure your looking at running services.  Those others are probably eliminated already with juju destroy-environment
[18:46] <Tantor> Ok, I did another destroy-environment now. So you think I should wait a while and then do a bootstrap again?
[18:47] <Tantor> I'm not using ec2 btw, but Ubuntu Maas (openstack)
[18:47] <weblife> Maybe someone can confirm juju-gui spins up two service machines.  I don't know for sure myself.  Anyone know that could answer Tantor?
[18:48] <weblife> @Tantor Sorry,  dont think it matters.  Probly the same in both.
[18:48] <weblife> brb
[18:49] <Tantor> According to the ubuntu instructions I need at least 2 free nodes to use juju. As soon as I do juju bootstrap. Automatically juju will use these two nodes
[18:57] <dpb1> does lxc on juju-core still honor the data-dir setting, or is there something else I should be using now?
[19:01] <dpb1> Got it: root-dir  (was data-dir)
[23:25] <weblife> @marcoceppi I just realized I need to be a member of charmers to submit my patch after nearly going crazy in a quest to find out why it would not let me(first time using launchpad).  Could you approve my membership,  I already made the request.