[04:41] <viswesn> Why I am not able to see juju model-config updated value - http://pastebin.ubuntu.com/24358872/
[07:00] <kjackal> Good morning Juju world!
[07:25] <erik_lonroth> good morning.
[08:45] <joedborg> morning! o/
[08:46] <joedborg> Quick juju question - will 2.2 have the same behaviour as 2.1.2 in terms of not automatically bridging interfaces?
[09:47] <cnf> in a model yml file, can I define to: lxd, and not specifiy what machine it should put it on?
[10:25] <rick_h> cnf: you can for a machine in the bundle, but not an existing machine.
[10:35] <cnf> rick_h: right, that's what i meant
[10:35] <cnf> what is the correct syntax?
[11:01] <cnf> how do you delete subordinate units?
[12:30] <tvansteenburgh> cnf: juju remove-unit
[12:31] <anrah_> is there a reason why 2.0.X are not listed on juju-stable repository?
[12:31] <cnf> tvansteenburgh: not for subordinates
[12:32] <tvansteenburgh> cnf: that's news to me, why doesn't it work?
[12:33] <cnf> i don't know, it tells me remove-unit doesn't work with subordinates
[12:39] <tvansteenburgh> cnf: huh. i never realized that. but: https://jujucharms.com/docs/2.0/authors-subordinate-services#caveats
[12:40] <cnf> hmm
[13:57] <lazyPower> cnf: is this via the CLI or via the Gui?
[13:58] <lazyPower> cnf: subordinates are a special case where they presume to exist for the lifecycle of the related unit. I've had some talks with rick_h about this in the past, but I don't think we've settled on anything beyond we have differing opinions about subordinate units and their lifecycle.
[13:58] <lazyPower> and when i say related unit, i mean principal unit.
[14:02] <cnf> cli
[14:02] <cnf> i uh, never use the gui?
[14:02] <cnf> i don't understand it
[14:03] <rick_h> cnf: so there's an issue in that subordinates are colocated on the root machine and could do things like install PPAs, modify packages, etc. Such that figuring out what's the subordinates changes vs the host application is impossible.
[14:03] <rick_h> cnf: anything that claimed to "remove the subordinate" would be a lie you can't trust in your model. It's dirty at that point
[14:04] <rick_h> cnf: so yea, we go the route that to get a clean host application you need to redeploy it w/o the subordinate relation
[14:18] <cnf> rick_h: right
[14:18] <cnf> rick_h: so i did a silly, and installed 2 applications on 1 machine that _both_ had ntp as a subordinate
[14:18] <cnf> so they where fighting over who got to config it
[14:20] <cnf> rick_h: anyway, they are no longer on the same machine, but thanks for the explanation
[14:21] <rick_h> cnf: yea, sorry. I agree that it'd be cooler if we could make that cleaner. Honestly, I think things like snaps help in this world a little bit.
[14:21] <rick_h> cnf: but we'd rather be a bit more up front about a known possible spot than have folks chasing why something is failing in production for a non-obvious reason that there used to be a subordinate that didn't remove cleanly there.
[14:23] <cnf> rick_h: sure
[14:23] <cnf> rick_h: it's not always clear where the problem is coming from :P
[14:23] <rick_h> cnf: yea, it's a bit of magic managing I guess.
[14:31] <Trefex> hi guys, is there any information on what Ceph-Dash looks like? https://jujucharms.com/u/canonical-storage/ceph-dash/
[14:33] <rick_h> cholcombe: ping ^
[15:03] <cholcombe> Trefex: you're looking for screenshots?
[15:04] <Trefex> cholcombe: yeah for instance, i found a ceph-dash from Crapworks
[15:04] <cholcombe> Trefex: ah yes.  this one is a different than the crapworks dash
[15:04] <Trefex> cholcombe: but I have no idea what ceph-dash does, why I should pay for it (I guess it's behind canonical paywall) and what the rest of the architecture should look like
[15:04] <Trefex> eg telegraf, collector, influx, etc
[15:20] <cholcombe> Trefex: yeah that's an excellent question. i'll work on getting some nice screenshots prepared :)
[15:20] <Trefex> cholcombe: wohooo thanks :)
[16:38] <pranav> Hi Folks. Need help with the cinder charm. getting the following error in cinder-scheduler log
[16:38] <pranav> oslo_service.service RPCVersionCapError: Requested message version, 3.0 is incompatible.
[16:38] <pranav> any pointers
[16:38] <pranav> couldn't find anything on google
[16:50] <bdx> pranav: ask in #openstack-charms
[16:51] <pranav> Already did. Still waiting there. Thanks
[17:50] <Budgie^Smore> o/ juju world
[18:53] <rick_h> anyone know a twitter handle for tengu/merlijn's folk?
[19:01] <SimonKLB> is it possible to change the profile image of a team in jujucharms.com ?
[19:12] <urulama> SimonKLB: the profile is actually a gravatar defined by an email. we could add an option that one can set that for the team, but we there no such functionality atm
[19:40] <cnf> kwmonroe: so setting no_proxy didn't help, btw
[22:42] <smgoller> Hey all, I'm trying to bootstrap a juju controller on our local openstack cloud, and it's not clear to me what the answer is to the question "Enter the API endpoint url for the cloud"
[22:42] <smgoller> The docs mention port 35357, which I can't find any reference to in the openstack documentation
[22:43] <hml> smgoller: use your keystone endpoint
[22:44] <hml> smgoller: i believe 35357 is the port for authenication v3, 5000 is for v2
[22:44] <smgoller> ahhh
[22:45] <hml> smgoller: the same endpoint can be used for both questions related to endpoints in add-cloud
[22:45] <smgoller> failed to bootstrap model: no image metadata found
[22:45] <smgoller> is it expecting an image with a particular name to exist?
[22:53] <smgoller> figured it out
[22:53] <smgoller> thanks!