=== kadams54 is now known as kadams54-away === natefinch-afk is now known as natefinch === brandon is now known as web [05:32] Does Juju have an environment API I can play with? for example, a command that will list all the deployed services, but just service names (not status) === zz_CyberJacob is now known as CyberJacob [06:23] jose: Yes. I don't know where it is documented, but I do know there are python bindings (jujuclient on pypi) [06:24] stub: oh, awesome. I'll check later - need to rest a bit now. thanks a bunch! === CyberJacob is now known as zz_CyberJacob === Spads_ is now known as Spads [10:28] gnuoy, pxc seems unhappy - non-leaders are trying to migrate stuff from local storage [10:28] debugging that now [11:17] Hey Liam [13:18] hmm, marcoceppi, tvansteenburgh, aisrael - when yinz get in I have a question about deployer and some bundles, i've managed to build a bundle that only works against deployer in pip - not whats in the ppa. I'm curious about the skew here and what i can do to resolve this problem. [13:35] lazyPower: the ppa is well ahead of pypi right now. probably means there's a regression in a recent commit [13:36] ok, is the head of development for deployer still in launchpad? [13:36] lazyPower: wait, the juju/stable ppa, or other? [13:36] lazyPower: yes [13:36] tvansteenburgh: i have both enabled on my system, aiui deployer only lives in the juju/stable ppa [13:38] lazyPower: i think there are daily builds too. if you're on stable i'd expect that to match what's on pypi but it sounds like that may not be the case right now [13:39] ok, I'll dig a bit deeper and see if i can find whats going on. Thanks for the info tvansteenburgh [13:40] lazyPower: np, good luck [13:44] aisrael: 1463117 hit fix-committed lastnight \o/ [13:46] lazyPower: excellent! [14:55] tvansteenburgh, do juju-deployer packages follow the juju/devel juju/proposed and juju/stable ppa flow? [14:56] beisner: not sure. dpb, you know? [14:56] dpb1 ^ [14:56] tvansteenburgh, i ask because uosci uses deployer from those ppa sources, and we can exercise proposed packages ahead of landing in stable if so. [14:56] beisner: no, nothing that complex [14:57] beisner: we build nightlys from trunk in two ppas [14:57] i'd have to manually kick a job, but would be happy to do that in the interest of validating juju-deployer changes against existing implementations [14:57] beisner: then every once in a while, marco or someone puts them into juju/stable, then every once in a while they are released into the distro in universe. [14:58] ah i see. [14:58] beisner: two ppas: [14:58] https://code.launchpad.net/~ahasenack/+recipe/python-jujuclient-daily [14:58] https://launchpad.net/~ahasenack/+archive/ubuntu/juju-deployer-daily [14:59] beisner: personally, I run from those two ppas [15:00] dpb1, ok thanks for the info. and thanks for being the perpetual deployer gate ;-) [15:01] beisner: lol :) === kadams54 is now known as kadams54-away === kadams54-away is now known as kadams54 [15:47] marcoceppi: hey, is there a way to use amulet to test actions? not documented afaics === redelmann is now known as rede|eateateatea [16:27] jose: you can, but its not using amulet syntax sugar [16:28] jose: you use subprocess to call the juju action do stanza [16:29] lazyPower: huh, good idea. I'm gonna check that and see what I can do on my end. thanks! === natefinch is now known as natefinch-afk === rede|eateateatea is now known as redelmann === scuttle|afk is now known as scuttlemonkey === liam_ is now known as Guest41765 === kadams54 is now known as kadams54-away === kadams54-away is now known as kadams54 === kadams54 is now known as kadams54-away === JoshStrobl is now known as JoshStrobl-Codin === JoshStrobl-Codin is now known as JoshStrobl-Busy [19:29] hi === natefinch-afk is now known as natefinch [19:48] jcsackett: marcoceppi arosales lazyPower anyone else, we're bringing up new jujucharms.com env and data is loading. Data might be loading (like stats) for some time so if you see any funkiness around that (we've not switched dns yet) that's why and should settle out by tomorrow [19:48] ack, thanks rick_h_ [19:49] rick_h_: think you probably meant to ping jcastro there, not me? ^ [19:49] jcsackett: ah yea, sorry jcastro ^ [20:17] rick_h_, thanks for the fyi [20:17] noted [20:27] Hi is there any official solution for High availablity OpenStack with multi-controller on juju charm store? Can someone point me towards it. Thanks [20:54] Has anyone here used juju to deploy a full open stack system on a single node (for test purposes) === natefinch is now known as natefinch-afk [21:03] Destreyf: closest i've seen is illustrated here: http://marcoceppi.com/2014/06/deploying-openstack-with-just-two-machines/ [21:04] Destreyf: but note: the article is a year old, and uses 2 machines [21:04] lazyPower: thanks for the link, i hadn't seen this one yet [21:04] basically i want to setup a cluster with 3 nodes [21:04] i know that's terrible to want [21:04] :P [21:05] but its only going to serve about 3-4VM's [21:05] so its very little usage [21:05] do you need the overhead of openstack for that? [21:05] juju supports deploy --to kvm:# [21:05] well [21:05] we need HA [21:05] currently using proxmox [21:05] 3 nodes will not give you a HA setup [21:06] 12 nodes minimum for a proper openstack, even more for HA [21:06] even when stacking services on a single machine? [21:06] if you unplug the machine, your stacked services all tank [21:07] i'm assuming you're talking having 2 noiva-computes as "HA" and everything else (cinder, glance, keystone, et-al) running off your bootstrap node [21:07] The goal was to actually setup 3 servers, with server 1 being teh bootstrap [21:08] but also having the rest on it [21:08] and possibly telling juju to be multi-node bootstrap [21:08] although i haven't actualyl researched that [21:08] i dont think thats going to work, but i haven't tried it [21:09] :P [21:09] thanks for your input [21:09] i appreciate it [21:09] just trying to find the most effective solution for workable ha [21:09] on a 3-4 node setup === kadams54 is now known as kadams54-away [22:01] ping hazmat [22:20] Destreyf: it's possible, I've been meaning to write a new blog post on it [22:20] marcoceppi: that's awesome to hear [22:20] i'm actually following your guide [22:20] using juju and bootstrap node [22:20] to test with single machine [22:20] any caveats to note? [22:23] marcoceppi: i bound juju-gui to port 8081 by doing juju set juju-gui port="8081" [22:23] so that's the only hiccup i could see so-far [22:24] Destreyf: you'll want to use LXC for almost everyhting [22:24] and you can't really co-locate ceph and nova-compute [22:25] oh? [22:25] basically, deploy nova-compute to the one machine, then everything else as LXC containers [22:25] ceph and compute can't co-locate [22:25] so ceph inside of lxc would work though? [22:25] Destreyf: last I checked, that may have changed. Ceph can't run in a LXC container [22:25] oh [22:25] okay [22:25] thanks [22:26] Destreyf: if i have time this weekend I'll give it a go again and blog about it [22:26] nuetron is also a bit sticky [22:26] if you skip neutron, it's a lot easier [22:26] otherwise you'll have to create a custom KVM container with two nics [22:26] Yeah i figured neutron would be a pain, but without neutron can't make it HA really can you? [22:27] also i had neutron mostly working on my last test run, but juju kept getting stuck installing OpenStack-Dashboard [22:27] once everythign else was running outside of containers [22:27] just used --to 0 [22:27] on everything [22:27] Destreyf: nova-compute is --to 0 [22:27] everyhting else is basically --to lxc:0 [22:27] that will give you containerisation of resources and prevent conflicts [22:28] nice, well letting LXC run and install teh rest of the mess right now :P [22:28] Destreyf: you could also use VMAAS [22:28] once i have this much done, i'll image the machine and play around with getting ceph into the mix [22:28] where you create a bunch of KVM containers on that one machine and use MAAS to manage the virtualization [22:29] marcoceppi: the end goal is to have a 3-4 node openstack setup, we're using a proxmox system right now, but we have too many performance issues to keep using it. [22:29] we might also just go with 3 proxmox and 3 ceph nodes instead [22:29] if we can't get this working [22:30] just would like to reduce the number of nodes to a more power managable number [22:30] :P [22:30] backup battery only lasts 5 minutes currently [22:30] with our 7 nodes [22:32] mysql came up [22:32] :D [23:29] marcoceppi: So my openstack-dashboard doesn't show a public ip other than the ip assigned by lxc