[02:21]  * hazmat looks wwitzel3's state services branch
[09:19] <schegi> jamespage?
[09:48] <jamespage> schegi, hello!
[09:51] <schegi> hi got two question. you mentioned that you fixed somthing related to the missing zap with ceph. i updated your network-slpit charms but they are still up to date. have you commited?
[09:52] <schegi> second. i got problems to deploy the checked-out network-split version of cinder and cinder-ceph. he does not recongnize them as charm if i want to deploy them from local repo. is there something missing in the charm??
[09:56] <schegi> And i have some strange behavious related to the hacluster charms. when i deploy it like it is described in https://wiki.ubuntu.com/ServerTeam/OpenStackHA or change to go for the percona-cluster charm (doent matter). the deployment works fine very service is up and running, no hook fails, BUT if i log in to one of the machines and check the corosync/pacemaker cluster by crm_mon or crm status it seems to me that the single services are r
[09:57] <schegi> but it looks like they are not connected. i am no corosync pacemaker pro. but i know from an manual deployment taht all nodes in the cluster should appear in the output of crm status and they didnt.
[09:59] <jamespage> schegi, rev 84 in ceph contains the proposed fix
[10:00] <jamespage> schegi, you have multicast enabled on your network right?
[10:00] <jamespage> (re the hacluster issue)
[10:01] <jamespage> schegi, cinder and cinder-ceph should be OK - do you get a specific error message?
[10:02] <schegi> currently rebootstrapping but if i do something like juju deploy --repository /usr/share/custom-charms/ --to 0 --config ~/juju-configs/cinder.yaml local:cinder i get some charm not found message
[10:02] <schegi> doning the same with ceph works fine
[10:09] <schegi> ok first problem was my fault, just using bzr to rarely. tried update but had to pull. :) (still using too much svn)
[10:10] <schegi> jamespage, here is the error message i get ERROR charm not found in "/usr/share/custom-charms/": local:trusty/cinder
[10:11] <schegi> trying to do juju deploy --repository /usr/share/custom-charms/ --to 0 --config ~/juju-configs/cinder.yaml local:trusty/cinder or juju deploy --repository /usr/share/custom-charms/ --to 0 --config ~/juju-configs/cinder.yaml local:cinder
[10:11] <jamespage> schegi, dumb question but it is under a trusty subfolder right?
[10:12] <schegi> all the charms are there and for all the others it works perfectly
[10:12] <schegi> could do the same line just replacing cinder with ceoh and it deploys ceph.
[10:12] <schegi> away for 20 mins brb
[10:16] <g0d_51gm4> hello guys, my situation is the following: i've an vMaaS environment with 3 vm node and a juju environment. I've deployed juju-gui on vm node 1 and openstack on the other 2 nodes. now I'd like to add other 3 vm nodes and use them to deploy hadoop master plus a slave and cassandra but i've a doubt is it necessary to create another juju environment to dedicate to that? thanks..
[10:22] <g0d_51gm4> the error is the following http://paste.ubuntu.com/7813669/ while if i specify the node obtain that error http://paste.ubuntu.com/7813674/
[10:33] <g0d_51gm4> is there  anyone can support me to resolve that? thanks
[10:43] <schegi> jamespage, back
[10:44] <jamespage> schegi, great
[10:51] <schegi> jamespage, ok checked it twice. path is correct but still not able to deploy cinder from local repo
[11:04] <jamespage> schegi, testing myself now
[12:39] <g0d_51gm4> anyone can help me?
[13:37] <g0d_51gm4> please is there someone who can help me to resolve it?please
[13:38] <Sh3rl0ck> Hello..Has anyone observed ceph.conf being configured wrongly after deploying juju ceph charm?
[13:40] <Sh3rl0ck> The keyring values in the ceph.conf file seems to be incorrectly populated as it contains "$cluster.$name.keyring rather than actual values
[14:00] <ctlaugh> jamespage - I'm having an issue with the cinder charm on a single-drive system.  Would you be able to help?
[15:28] <jcastro> Hey everyone, so for a demo I want to launch a bundle into multiple environments
[15:28] <jcastro> I know I can do
[15:28] <jcastro> juju deploy blah
[15:28] <jcastro> juju switch
[15:28] <jcastro> juju deploy blah again
[15:28] <jcastro> Is there a way I can do `juju deploy this bundle into these environments` all at once? Not serially.
[15:31] <rick_h__> jcastro: I've not tried it but quickstart takes a -e flag
[15:32] <rick_h__> jcastro: so you could in theory run a bunch of quickstart commands backgrounded each with a diff -e?
[15:32] <jcastro> yeah, but then I am concerned that they will step on each other
[15:32] <rick_h__> jcastro: yea, that's the untested part
[15:33] <jcastro> hmm, actually, as long as I don't switch while they are launching ....
[15:33] <jcastro> that _should_ work
[15:33] <rick_h__> but I think it would work since once the script runs it's got all the config loaded.
[15:33] <jcastro> ok I will try that
[15:44] <Sh3rl0ck> Hello..has anyone used the ceph charm for ceph backend to cinder? Do we need to add a relation between Nova compute and Ceph?
[16:02] <jcastro> rick_h__, so that totally worked!
[16:04] <rick_h__> jcastro: woot!
[16:50] <Sh3rl0ck> Hello..has anyone used the ceph charm for ceph backend to cinder? Do we need to add a relation between Nova compute and Ceph?
[16:50] <jcastro> hey sinzui
[16:50] <jcastro> I am having some issues with juju on joyent
[16:51] <jcastro> it bootstraps, and deploys the gui
[16:51] <sinzui> yeah you see that too
[16:51] <jcastro> but subsequent deploys are waiting on the machine
[16:51] <sinzui> jcastro, Exactly what CI sees
[16:51] <jcastro> but oddly enough, juju just returns "pending" and continues on with life
[16:51] <jcastro> oh whew! so not crazy!
[16:52] <sinzui> jcastro, I manually deleted all the machines in our joyent account this morning to fix a test that was failing
[16:52] <sarnold> Sh3rl0ck: note also http://manage.jujucharms.com/charms/precise/cinder-ceph
[16:52] <jcastro> sinzui, I have one more issue with hp cloud
[16:52] <jcastro> do you have a network config to put into their horizon console to make juju work?
[16:53] <sinzui> jcastro, This is an old problem that happens, but Juju might be partially to blame. I often see stopped machines instead of deleted machines. My api calls also fail to delete, so maybe joyent is at fault
[16:54] <sinzui> jcastro, creating one network is enough to make juju work in new Hp. New accounts get a default network, but migrated accounts may need to add one, just accept the recommended settings
[16:54] <jcastro> yeah I tried that but no joy, I'll give it another shot
[16:54] <jcastro> good to know I wasn't going crazy wrt joyent though
[16:55] <sinzui> jcastro, I wrote this a few weeks ago. HP has been lovely since then. http://curtis.hovey.name/2014/06/12/migrating-juju-to-hp-clouds-horizon/
[16:55] <Sh3rl0ck> sarnold: Thanks for link. I actually tried out cinder-ceph charm as well but seems like its required only if you want multi-backend support for Cinder
[16:55]  * sinzui reads hp network config
[16:55] <jcastro> thanks, I'll try that
[16:55] <jcastro> <-- lunch, bbl
[16:55] <Sh3rl0ck> But not sure ceph charm needs to be connected to nova-compute
[16:56] <Sh3rl0ck> Also, does anyone know if there is a way to add netapp backend to Cinder? Is there a separate charm for this?
[16:56] <sinzui> jcastro, CI has "default  default 10.0.0.0/24  No	ACTIVE	UP"
[16:57] <sinzui> jcastro, charm-testing is similar, the only difference is "default" is a string that tells me Antonio named it
[17:13] <sebas5384> if an agent of a machine is down
[17:13] <sebas5384> what can I do?
[17:14] <sebas5384> agent-state: down
[17:14] <sebas5384> agent-state-info: (started)
[17:28] <sinzui> sebas5384, There are a few things to do after you ssh into the machine with the down agent
[17:28] <sinzui> sebas5384, get a listing of the procs: this will show both the machine and the unit agents ps ax | grep jujud
[17:29] <sinzui> sebas5384, sudo kill -HUP <pid> will restart a stalled agent
[17:30] <sebas5384> sinzui: there's only the 0 machine running
[17:31] <sinzui> okay, that lead to starting the unit.  jujud are under upstart, but uniquely named. We need to learn the name: ls /etc/init/jujud-*
[17:32] <sinzui> sebas5384, you can run something like this: sudo start jujud-unit-arm64-slave-0
[17:33] <sebas5384> hmmm sinzui yeah! I did a restart of the agent
[17:33] <sebas5384> because the jujud init isn't there
[17:34] <sinzui> oh
[17:34] <sinzui> sebas5384, that implies the service didn't complete installation
[17:35] <sebas5384> ls /etc/init | grep juju
[17:35] <sebas5384> juju-agent-devop-local.conf
[17:35] <sebas5384> juju-db-devop-local.conf
[17:35] <sinzui> but the status you reported says it was there
[17:36] <sebas5384> hmmm so it should be there
[17:36] <sebas5384> this one is really tricky
[17:36] <sinzui> sebas5384, right those are the machine and db procs that comprise the state server commonly on a machine 0. DId you deploy a service
[17:37] <sebas5384> sinzui: i'm using the cloud-installer
[17:38] <sebas5384> that tries to deploy a bunch of services of openstack into one nested container into a kvm
[17:38] <sinzui> sebas5384, is the first time on that machine that use used it? lxc needs to build a template first, and that is slow. After the first time, it gets fast
[17:38] <sinzui> ah kvm, not template
[17:38] <sebas5384> yeah
[17:38] <sebas5384> hehe
[17:40] <sinzui> sebas5384, I don't have much experience with cloud installer. the agent you want to start/restart will be in the container, kvm or lxc. your local host only gets the state-server in this case
[17:41] <sebas5384> sinzui: right
[17:41] <sinzui> sebas5384, in my example, I sshed to the machine the agent was down in, then restarted the proc (2 weeks ago actually, that example was from my histor)
[17:42] <sebas5384> but because i can't ssh into the machine
[17:42] <sebas5384> so i'm a bit into a limbo
[17:42] <sebas5384> hehe
[17:42] <sinzui> sebas5384, "juju ssh" will work regardless of the state of the agent, and that is a wrapper for real ssh
[17:43] <sebas5384> ok, but is like the machine isn't started
[17:43] <sebas5384> but when i do juju status, its saying that it is
[17:44] <sebas5384> other thing i went to the virsh console to see the kvm machines
[17:44] <sinzui> sebas5384, well that is a different matter. kvm provisioning is slow or something else has gone amiss.
[17:44] <sebas5384> yeah, I think something else happen here
[17:46] <sinzui> stokachu, do you have any cloud-installer experience that might help sebas5384
[17:47] <stokachu> sinzui, yea im working with him to figure out whats happening
[17:49] <sebas5384> thanks sinzui and stokachu !! :)
[17:49] <sebas5384> so i restarted the machine by the virsh console
[17:49] <sebas5384> and it seems to worked
[17:49] <sebas5384> now im into the machine
[17:50] <sebas5384> holly molly!!
[17:50] <sebas5384> its all installing now
[20:29] <shiv> trying to deploy openstack HA on physical servers...anyone see issues while deploying on physical servers....Mysql hook failing with cinder...this is the error: shared-db-relation-changed 2014-07-18 20:22:28.637 28128 CRITICAL cinder [-] OperationalError: (OperationalError) (1130, "Host '4-4-1-95.cisco.com' is not allowed to connect to this MySQL server") None None
[20:30] <shiv> here is the stack trace
[20:30] <shiv> http://pastebin.com/DRKfi4d1
[20:30] <shiv> 4.4.1.95 is the VIP for the for the cinder