[01:25] <Dotted> how do you push a local charm to launchpad? im getting "bzr: ERROR: Invalid url supplied to transport: "lp:~dotted1337/charms/precise/sendy/trunk": No such source package sendy." when doing bzr push
[01:31] <Dotted> figured it out bzr didnt know my lp username
[01:32] <davecheney> nice
[01:33] <freeflying> I ran into a bootstrap issue within a maas environment
[01:34] <freeflying> bootstrap has its node installed, but juju then can't connect to mongodb on that node
[01:36] <freeflying> 2013-09-21 16:59:18 ERROR juju.agent agent.go:448 failed to initialize state: no reachable servers
[01:36] <freeflying> 2013-09-21 16:59:18 ERROR juju supercommand.go:282 command failed: no reachable servers
[01:36] <freeflying> error: no reachable servers
[01:36] <davecheney> freeflying: what does /var/log/cloudinit-output.log on the bootstrap node say ?
[01:36] <davecheney> most likely it could not add the PPA
[01:36] <davecheney> and installed the wrong version of mongodb
[01:37] <freeflying> davecheney, which ppa shall it add?  I can try manually, see if its related to network(pproxy)
[01:38] <davecheney> freeflying: check the output of cloudinit-output.log
[01:38] <davecheney> all the details are there
[01:40] <davecheney> freeflying: are you using a proxy in your environment ?
[01:40] <davecheney> juj basically assumes it has a working defualt route
[01:40] <davecheney> if that isn't the case
[01:40] <davecheney> well, shit is going to break
[01:40] <freeflying> davecheney, yes
[01:41] <freeflying> davecheney, no direct internet for the maas environment, all traffic have to go via a proxy, which we set on maas server by using squid-deb-proxy
[01:41] <freeflying> davecheney, I can't add the proxy manually :)
[01:42] <davecheney> freeflying: yup, that is supported (and recommended)
[01:42] <davecheney> just finding the magic settings
[01:42] <freeflying> davecheney, thanks :)
[01:42] <davecheney> basically if the host you are bootstrapping from uses the proxy
[01:42] <davecheney> we sniff that and pass the same details down to the bootstrap node
[01:44] <davecheney> freeflying: try this
[01:44] <davecheney> apt-config dump | grep -e 'Acquire::[a-z]+::Proxy\s+"[^"]+"'
[01:44] <davecheney> if that returns anything in on the client
[01:44] <davecheney> then we'll pass those details via cloud init
[01:44] <davecheney> if the client and the maas instance are using differnt proxy config
[01:45] <davecheney> then youre probably stuck
[01:45] <davecheney> that isn't a configuration we anticipated
[01:45] <freeflying> davecheney, returns empty
[01:45] <davecheney> freeflying: short solution
[01:45] <davecheney> configure your workstation to use the apt proxy
[01:45] <freeflying> davecheney, what does my workstation mean?
[01:46] <davecheney> freeflying: where you typ ethe command 'juju bootstrap'
[01:46] <freeflying> davecheney, on maas server
[01:47] <davecheney> freeflying: that honestly isn't a configuration we anticipated
[01:49] <freeflying> davecheney, interesting thin is the node has no problem to access internet
[01:52] <davecheney> freeflying: which is 'the node' ?
[01:54] <freeflying> davecheney, the bootstraped one
[01:55] <davecheney> freeflying: i'm sorry, i am confused
[01:55] <davecheney> i thought you told me that the bootstrapped node did not have direct access to the internet
[01:55] <davecheney> 11:41 < freeflying> davecheney, no direct internet for the maas environment, all  traffic have to go via a proxy, which we set on maas server by  using squid-deb-proxy
[01:57] <freeflying> davecheney, in maas's preseed, the proxy has been set up, so bootstrap node has a default gw via that proxy
[01:59] <davecheney> freeflying: so, the host does _not_ have direct internet access ?
[01:59] <davecheney> is that correct ?
[02:00] <freeflying> davecheney, the host here means which?  maas server?
[02:01] <davecheney> freeflying: the bootstrap node
[02:01] <davecheney> freeflying: can you paste the /var/log/cloudinit-output.log from the bootstrap node
[02:01] <davecheney> that will make everything clear
[02:02] <freeflying> davecheney, hold one plz
[02:09] <freeflying> davecheney, sent you in msg
[02:10] <davecheney> kk
[02:10] <davecheney> is your squid-apt-proxy configyured to proxy ppa.launchpad.net
[02:10] <davecheney> by default it does not
[02:12] <davecheney> freeflying: does this work
[02:13] <davecheney> env https_proxy=$YOURPROXY curl https://launchpad.net/~juju/+archive/stable/+files/mongodb-dev_2.2.4-0ubuntu1~ubuntu12.04.1~juju1_i386.deb
[02:14] <davecheney> sorry add a -L to that cur;
[02:14] <davecheney> sorry add a -L to that curl
[02:14] <davecheney> basiscally trying to emulate what cloud-init is doing
[02:15] <freeflying> curl: (56) Received HTTP code 403 from proxy after CONNECT
[02:15] <davecheney> ok, that is why it isn't working
[02:15] <davecheney> by default squid-apt-proxy does not allow that
[02:16] <davecheney> from memory there are comments in the configuratoin file to allow it to talk to ppa sources
[02:17] <freeflying> you mean enable ssl? or have ppa.launchpad.net enable in squid-deb-proxy?
[02:17] <davecheney> http://askubuntu.com/questions/303150/apt-get-403-forbidden-but-accessible-in-the-browser
[02:17] <davecheney> sorry, not an awesome answer
[02:18] <davecheney> https://answers.launchpad.net/ubuntu/+source/squid-deb-proxy/+question/179075
[02:18] <davecheney> gives another hint where the config is
[02:40] <freeflying> davecheney, I have ppa.launchpad.net added to squid-deb-proxy's mirror-dstdomian.acl via maas, but still, it can cache ppa
[02:41] <davecheney> freeflying: can or cannot ?
[02:41] <freeflying> davecheney, sorry, can't
[02:42] <davecheney> freeflying: add some more, like launchpad.net and launchpadlibrarian.net
[05:15] <freeflying> davecheney, thanks, have it solved
[05:16] <davecheney> freeflying: cool
[05:16] <davecheney> what was the solution ?
[05:17] <freeflying> davecheney, add keyserver and launchpad to squid-deb-proxy's mirror-dstdomian
[05:17] <freeflying> davecheney, also modify maas's preeed to add the ppa in late_command with http/https_proxy
[05:18] <freeflying> davecheney, all in short, its a proxy issue :)
[05:18] <davecheney> freeflying: you do not need to modify the maas preseed
[05:18] <davecheney> we handle that in cloud init
[05:19] <davecheney> maas preseed and cloudinit have a strong overlap
[05:19] <freeflying> davecheney, seems it for http_proxy, but doesn't work for https_proxy
[05:20] <davecheney> freeflying: did you setup the apt-proxy as i showed above
[05:20] <davecheney> if that regex does not return anything
[05:20] <davecheney> then the correct value will not be supplied to cloud init when you bootstrap an environment
[05:20] <freeflying> davecheney, no, didn't set anything then
[10:50] <jamespage> raywang, should work as described in the README
[10:51] <raywang> jamespage, it claims supporting add relation to keystone, but it's missing hooks, dosaboy report that bug  https://bugs.launchpad.net/charms/+source/ceph-radosgw/+bug/1229645
[10:51] <_mup_> Bug #1229645: ceph-radosgw missing identity relation links <ceph-radosgw (Juju Charms Collection):In Progress by hopem> <https://launchpad.net/bugs/1229645>
[10:54] <jamespage> raywang, I've merged dosaboys branch
[10:57] <jaywink> hi all. wondering if anyone has a link or explanation to what is required from a cloud image to bootstrap with juju? Say I get a non-ec2 but ec2 API compatible provider to work up until bootstrap - will a default ubuntu server image do or does it require some custom image that for example Amazon has?
[11:10] <davecheney> jaywink: the first thing you need is a juju provider
[11:11] <davecheney> we have providers for ec2, openstack (including private openstack clouds), maas and azure
[11:11] <davecheney> if your virtual environment looks *identical* to one of those
[11:11] <davecheney> it will work
[11:11] <davecheney> if not, it won't
[11:23] <jaywink> davecheney, by provider you mean the actual cloud image? I've got up to the actual RunInstance call on GreenQloud which has an EC2 compatible API - they have a precise server image available
[11:24] <davecheney> jaywink: we call a provider a piece of software designed to translate juju's abstract view of the world onto a real provider of virtual (or physical) machines
[11:24] <davecheney> jaywink: the problem you'll find with something that smells like EC2 is the AMI instance numbers won't match
[11:24] <davecheney> so, you say to juju 'precise/amd64 1 core, 1.7gb of ram please'
[11:24] <davecheney> and we have to find the AMI that matches that
[11:25] <jaywink> davecheney, I set it as POC to the code itself :P
[11:25] <davecheney> these values are fixed for ec2, and vary even per refion
[11:25] <davecheney> jaywink: i think IRC is a poor medium to this
[11:25] <davecheney> i suggest continuing this discussion on juju-dev ML
[11:26] <davecheney> jaywink: what you want to do is possible
[11:26] <davecheney> but will take some tweaking
[11:26] <davecheney> especially with the ec2 provider as it has never been tested against ec2 workalikes
[11:27] <jaywink> davecheney, don't mind tweaking :) and GQ is very interested too to get things working, even if as POC with non-stable code mods to begin with. I'll subscribe to the list and throw in the questions - thanks! :)
[11:28] <davecheney> jaywink: i'm sure it can be done with a bit of hacking
[11:28] <davecheney> the image/AMI selection will be the trickiest bit
[11:30] <davecheney> jaywink: writing a new provider, copying the ec2 provider would be the simplest solution
[11:30] <davecheney> again, the best way to approach this is via the development list
[11:31] <jaywink> davecheney, I just used 'euca-describe-images' against GQ and took the most suitable one - GQ doesn't have that many :) But just worried while waiting for GQ to fix one thing that whether their server image will work or does Juju require some custom Ubuntu builds.
[11:33] <davecheney> jaywink: we have a product
[11:33] <davecheney> well
[11:33] <davecheney> not really a product
[11:33] <davecheney> it's just a shittonne of json
[11:33] <davecheney> which allows you to describe the images available in a cloud
[11:33] <davecheney> this will probably be the best solution to solving the image selection issue
[11:33] <davecheney> and glue together what juju expects with what GQ provides
[11:36] <jaywink> ok cool, sounds good. just joining the list, will throw something in there. Thanks!
[11:36] <davecheney> jaywink: kk
[13:10] <sinzui> adeuring, Bug #1229708
[13:10] <_mup_> Bug #1229708: UnavailableSradsException from cstats <elasticsearch> <charmworld:Triaged> <https://launchpad.net/bugs/1229708>
[13:10]  * adeuring is looking
[14:59] <jamespage> adam_g, some of my ceph redux was fud
[14:59] <jamespage> https://code.launchpad.net/~james-page/charm-helpers/fixup-ceph-pool-creation/+merge/187260
[14:59] <jamespage> I've already merged those fixes into the openstack-charmers branch and re-syced cinder and glance
[15:06] <wedgwood> how do I recover a local environment after a reboot?
[15:07] <wedgwood> (reboot of the host system)
[15:33] <sinzui> adeuring, how goes the elasticsearch issue on staging?
[15:34] <adeuring> sinzui: fixed, but in a hackish way: I changed the ES clustername and rebuilt the index
[15:34] <sinzui> \o/
[15:34] <adeuring> sinzui: where do we store the configs of charmwold?
[15:35] <adeuring> (we don't want to run juju set elasticsearch cluster-name=new_name manually during deployment ;)
[15:35] <sinzui> adeuring, I don't think that is hackish? I added the config but  but never used it
[15:36] <adeuring> sinzui: well, I just declared a new cluster name, but that should be automatically used when staging is freshly deployed
[15:37] <sinzui> adeuring, your are right about the deploy...and I warned webops that running 2 ES stacks in prodstack will require them to have different cluster names. I suck for not writing it down
[15:37] <sinzui> adeuring, staging was deployed using the pseudo script in the orangestack branch
[15:38] <adeuring> sinzui: can you give me the URL?
[15:38] <sinzui> We want to update ./deploy script in it
[15:38] <sinzui> adeuring, you don't already have a branch?
[15:39] <sinzui> ahh, I am using the old name. adeuring look in lp:~ce-orange-squad/charmworld/staging-tools
[15:39] <adeuring> sinzui: could be -- can't find it at least... Maybe I'm just confused
[15:39] <adeuring> sinzui: ah, sure...
[15:40] <adeuring> sinzui: I'll update it
[15:40] <sinzui> adeuring, lets update the Charmworld page for webops to include the cluster name with each deploy.
[15:40] <sinzui> we might also want to update the deployer file. I need to find that branch
[15:41] <adeuring> sinzui: yep. I'll also write an internal email that we might step on each other's toes by not setting the cluser name. Perhaps my change affects otherES instances...
[15:42] <sinzui> adeuring, thank you!
[18:22] <hatch> when deploying the gui using local provider on 1.14.1-precise-amd64 the instance-state of the GUI machine is 'missing' - is this intended or an issue of some sort
[18:42] <hatch> ahh found the docs - that's normal
[19:11] <jcastro> hey sinzui you're doing release now?
[19:11] <jcastro> I mean you're doing the juju releases these days?
[19:12] <sinzui> yes
[19:12] <jcastro> are we defaulting to the local provider yet?
[19:13] <sinzui> jcastro, Sorry I don't understand.
[19:13] <jcastro> ok so say I do a clean install
[19:13] <jcastro> and do `juju init`
[19:14] <jcastro> do we write out local as the default environment yet?
[19:14] <sinzui> May be to 1.15.0 (if we fixed the bug)
[19:14] <sinzui> I can check
[19:16] <jcastro> I know it's a filed bug
[19:16] <jcastro> darned if I can't find it now though
[19:17] <sinzui> jcastro, https://launchpad.net/juju-core/+milestone/1.16.0
[19:17] <sinzui> ^ targeted to next week
[19:18] <jcastro> ok
[19:18] <sinzui> jcastro, I need to reset my juju-core. It thinks it is 1.14.1 from the release
[19:21] <jcastro> sinzui: https://bugs.launchpad.net/juju-core/+bug/1229903
[19:21] <_mup_> Bug #1229903: Default to local provider <juju-core:New> <https://launchpad.net/bugs/1229903>
[19:21] <sinzui> jcastro, 1.15.0 will use amazon as the default provider
[19:21] <jcastro> sinzui: I'd like to nominate it for 1.16 but not like, step on any release toes by just assigning it in LP
[19:22] <sinzui> jcastro, I (as a heavy juju lxc user) endorse your nomination. Are you arguing that Saucy (1.16.0) needs this out of the box setting for a great 30 minute experience?
[19:22] <jcastro> right!
[19:23] <jcastro> we had agreed verbally at the sprint, along with thumper
[19:23] <jcastro> I just realized today that we hadn't followed up and actually done that yet
[19:24] <sinzui> jcastro, I accept your argument and have targeted the bug. I will ask for forgiveness if I must
[19:24] <sinzui> I'll talk to the leads
[19:24] <jcastro> yes, cowboy the world, I like how you think. :p
[21:49] <weblife> afternoon juju mack daddy's
[21:49] <jcastro> hi!