[01:04] <mpae> hello
[01:04] <mpae> is anyone using juju with openstack?
[01:04] <mpae> by that I mean using openstack as the environment...
[01:06] <hloeung> mpae: yes, I am
[01:07] <mpae> hloeung: question, if you don't mind...  I'm getting a "could not find AWS_ACCESS_KEY_ID" error when I try to bootstrap, where did you find this?
[01:08] <mpae> or do you already have amazon credentials...
[01:09] <hloeung> mpae: from the sound of it, it sounds like you're using the Go version of juju?
[01:09] <mpae> hloeung: actually I'm not...  I had that problem even when trying to use maas with the Go version
[01:09] <mpae> so I switched to the non Go version early on
[01:10] <mpae> err the python version :)
[01:10] <hloeung> mpae: hmm interesting. What does 'juju --version' return?
[01:11] <mpae> juju 0.7
[01:11] <jose> hey marcoceppi, did you get to take a look at my charm/
[01:11] <jose> ?
[01:11] <mpae> hloeung: I'm also getting a warning about OpenStack identity service not using secure transport, I don't know if that's significant to this
[01:12] <hloeung> mpae: hmm, so looking at one of the environments we've deployed using the Python version. AWS_ACCESS_KEY_ID isn't set
[01:13] <mpae> hloeung: juju -v bootstrap shows that it's using "auth-mode 'userpass'"
[01:14] <mpae> hloeung: I followed the instructions provided here https://juju.ubuntu.com/docs/config-openstack.html to configure the environments.yaml
[01:14] <mpae> hloeung: is there a different guide I should be following for the python version?
[01:16] <hloeung> mpae: I'd try remove auth-mode
[01:16] <mpae> hloeung: but I don't have that anywhere in my environments.yaml
[01:17] <marcoceppi> jose: not yet
[01:17] <mpae> hloeung: do you have the same options specified in your environment as what's specified in https://juju.ubuntu.com/docs/config-openstack.html ?
[01:17] <jose> ok :)
[01:17] <mpae> hloeung: and are you using https or http for the keystone url?
[01:18] <hloeung> mpae: here's an example of my juju environment - http://pastebin.ubuntu.com/5857107/
[01:18] <hloeung> mpae: HTTPS for keystone
[01:19] <mpae> hloeung: thanks for that, I see you're using openstack type rather than openstack_s3
[01:19] <mpae> hloeung: you also don't have the keystone url, is that a copy paste ommission?
[01:21] <hloeung> mpae: the environment variable OS_AUTH_URL sets my keystone URL
[01:21] <mpae> hloeung: I see, I have that set as well, so I guess I don't need it
[01:22] <mpae> hloeung: I mean I don't need it in environments.yaml
[01:22] <hloeung> mpae: right
[01:22] <hloeung> mpae: let me know how you go
[01:22] <mpae> well, different error now :)
[01:23] <mpae> hloeung: different error now but progress :)  I'm going to try digging in to this a bit, thank you
[01:24] <hloeung> mpae: heh cool :)
[01:30] <mpae> hloeung: what version of openstack are you using?  the examples on the juju website say use openstack_s3 for and openstack deployment
[01:36] <mpae> this would be so much easier of the juju credentials download in horizon was working :'(
[01:39] <mpae> dang...  all I had to do is download the ec2 credentials and add the ec2rc.sh to my profile
[01:39] <mpae> now it no longer complains about AWS credentials
[01:39] <mpae> and I can use openstack_s3 provider type just fine
[02:07] <_mup_> Bug #1199205 was filed: juju boostrap does not work if keystone url does not have trailing / <juju:New> <https://launchpad.net/bugs/1199205>
[02:10]  * davecheney facepalm
[06:28] <AskUbuntu> simple MAAS setup in ubuntu 13.04 and virtual box | http://askubuntu.com/q/318032
[08:06] <melmoth> I need a way to get a list of all units peers. I can get info about new one in peer-relation-joined all right.
[08:06] <melmoth> but i need this information anytime, so i have to store those peers somewhere.
[08:06] <melmoth> how can i "lock" this somewhere for both reading and writing so i m sure i change thing atomically there ?
[08:07] <melmoth> if i dont use any locking mechanism i loose data because when i add n unit, i end up with n hooks changing the data there without tacking care of each others.
[08:07] <melmoth> is there a way to get either a write _and_ read lock in python ? (i only manage to make a write lock with fcntl but i dont undertsdant how to block reading too)
[08:08] <melmoth> or a thread like semaphore within a hook  code ?
[08:26] <henninge> melmoth: You have the "relation-list" inside a relation hook. That will give you the names of all the peers (minus the one the hook is running on).
[08:27] <melmoth> henninge, only when this hook is fired up by a "relation hook"
[08:27] <melmoth> and i need the data at another stage, so i need to store it when i have access to it.
[08:27] <henninge> AFAIK, yes.
[08:27] <melmoth> getting the data is not a problem, it s storing it in a safe way.
[08:28] <henninge> You can store the data in the charm directory.
[08:28] <henninge> I think that is ok to do.
[08:28] <henninge> or some other place.
[08:29] <henninge> Maybe some other place is better. ;-)
[08:42] <melmoth> henninge, the problem is concurent acces. When i have n unit added at the same time, i need a way to synchronise access to wherever the data is store
[08:43] <melmoth> i think i ll juts use a simple lock file. if it s there, let s wait. if its not, lets put it and do the job
[08:50] <henninge> melmoth: All hooks on one machine run in one thread.
[08:50] <henninge> So I don't think there is concurrency.
[08:50] <henninge> (I assume they run in one thread, actually)
[08:51] <melmoth> there is.
[08:51] <henninge> ok
[08:51] <melmoth> yesterday i added like 5 unit, saw in the log that all the unit where viewed, but in the resulting file i wrote, some were missing
[08:52] <melmoth> the only way i can explain that is a race condition between the moment the file was read by one hook, and the moment it was written again wit the new data
[08:52] <melmoth> i bet between those two moments,  an other unit did the same stuff and i lost 1 info
[12:24] <jamespage> mgz, how are we with regards 1.11.2 release of juju-core? I'd like to get something uploaded to saucy and I just resolved the golang 1.1.1 issue in the interim
[12:25] <mgz> jamespage: I'll poke dave, I don't feel we're blocked on anything now
[20:20] <marcoceppi> quiet today
[21:05] <henninge> exit
[21:05] <henninge> ;-)
[22:03] <arosales> any one have any ideas on https://lists.ubuntu.com/archives/juju/2013-July/002657.html
[22:34] <jose> marcoceppi: hey, wht did you mean by 'change configuration'?
[22:46] <jose> also, do I need to design my own icon?
[23:42] <marcoceppi> jose: if a user runs `juju set postfix hostname="something else"`
[23:43] <marcoceppi> right now nothing happens, this triggers the hooks/config-changed hook in the charm though
[23:43] <marcoceppi> Basically you'd just put the bulk of the hooks/install in to hooks/config-changed
[23:44] <marcoceppi> jose: not your own, but you should try to find the logo for postfix and put it on the charm icon template
[23:44] <sarnold> haha, http://www.postfix.org/mysza.gif
[23:45] <marcoceppi> Yup :) that should make for an interesting charm icon
[23:46] <sarnold> perhaps a touch less professional looking than exim or sendmail
[23:46] <sarnold> ah but still better than qmail :)  http://www.qmail.org/Q.6.01.Logo.lg.jpg
[23:46] <marcoceppi> Well, not much we can do a out that
[23:46] <marcoceppi> about*