[00:43] roaksoax, are you around ? [00:48] roaksoax, there are two MP there. i'm really ok with you chosing either one. === freeflying_away is now known as freeflying === freeflying is now known as freeflying_away === freeflying_away is now known as freeflying === julian__ is now known as julianwa [03:29] smoser: sorry wasn [03:29] feeling so good [03:29] smoser: i'll take a look tomorrow === CyberJacob|Away is now known as CyberJacob [07:29] hey, is anyone here willing to help out a noobie? [08:31] ping rvba [08:32] racedo: Hi (otp right now). [08:32] rvba: oh sorry [08:32] rvba: quick one when you can [08:33] rvba: when splitting maas region controller and maas cluster controller into different servers, the juju client will never talk directly to the maas cluster controller, right? [08:43] racedo: correct [08:43] rvba: many thanks :) [08:43] Welcome :) [08:44] rvba: and we can have just one juju bootstrap node for all the maas cluster controllers? [08:45] rvba: or should we commission one per cluster controller? [08:45] racedo: yes, as long as all the nodes managed by all the cluster controllers can talk to each other. [08:45] rvba: ok, so multiple juju bootstrap nodes is kind of pointless then [08:45] Yes. [08:45] cool, thanks again rvba === CyberJacob is now known as CyberJacob|Away === freeflying is now known as freeflying_away [12:10] Migration from generic OpenStack to Ubuntu Openstack | http://askubuntu.com/q/353127 === freeflying_away is now known as freeflying [13:15] rvba howdy [13:15] Hi roaksoax [13:15] rvba: hold on those ladt two mps [13:15] last* [13:15] roaksoax: don't worry, I don't intend to land them now. [13:15] rvba: ok ;) ill review them in a bit [13:16] roaksoax: they are mostly to get us started on testing the new script. [13:16] rvba: ok but im releasing maas again today [13:17] Okay. [13:48] rvba: ok, commented on them [13:48] rvba: other than that if they can be merged today [13:48] rvba: it would be great [13:50] roaksoax: Thanks for the review. We will need more testing before we land it; in particular, we want to make sure the old config gets migrated ok. [13:50] rvba: ok [13:51] rvba: though it would have been simple to keep the old config style as it is with import_pxe_files [13:51] y [13:51] roaksoax: we will rewrite maas-import-pxe-files in Python too. We thought it was a better idea to unify the configs. [13:52] rvba: ok [14:37] rvba, if import-pxe-files support specify mirror, might be more convenient [14:40] freeflying: true, we're still working on maas-import-ephemerals and import-pxe-files is our next target but you're right, we need to be able to configure where simplestreams will fetch its information. === freeflying is now known as freeflying_away [16:26] is the web interface the only way to manually enlist nodes? [17:00] roaksoax: ping [17:38] kurt_: pong [17:39] stokachu: if you have a live image on the node, you could install maas-enlist [17:41] roaksoax: ok ill try that [17:43] roaksoax: something I was thinking through with MAAS/juju - there is no way to guarantee a particular node with particular resources is used for a enlistment/commission/juju deployment - is that right? [17:44] if you don't have control over WOL, you are in bad shapee [17:44] shape [17:45] actually that doesn't even matter from juju's perspective it seems [17:46] kurt_: ideally you need control over ipmi, but enlistment/commuissioning are steps that have to be done in order for the machines to be available to juju [17:47] kurt_: now, you would need to add tags to the noeds to identify them based on constraints [17:47] and that way, when you juju deploy something you can gfarantee that a particular node with mparticular hw detauls is assigned to juju by specifying the constraints [17:47] tags? [17:48] first I'v heard of this [17:48] kurt_: http://maas.ubuntu.com/docs/tags.html [17:49] thanks I'll have a look - that integrates with juju? === matsubara is now known as matsubara-lunch [17:50] ahhh…I seee [17:50] * kurt_ lightbulb lights up [17:53] stokachu, you can enlist nodes with the cli too [17:56] smoser: i was able to get it to work with maas-enlist, is there another preferred way? [17:56] stokachu: maas-nelist when manually [17:56] stokachu: it is much easier, and that's what it is actually used when running the enlistment process [17:56] cool that worked like a champ [17:57] stokachu, are you using real nodes ? === CyberJacob|Away is now known as CyberJacob [17:57] smoser: virtualbox atm [17:57] virtualbox. [17:57] for real? [17:57] are you running this on windows! [17:58] haha [17:58] smoser: I have been very successful with VMWare workstation too [17:58] im trying to automate a maas+10 node setup [17:58] and Fusion on Mac OSX [17:58] I'm in the middle of blogging the experience [17:58] stokachu, http://bazaar.launchpad.net/~virtual-maasers/charms/precise/virtual-maas/trunk/files [17:58] see README-nojuju.txt [17:58] that is basically what you're after [17:59] and you dont have to use virtualbox [17:59] ok ill try that way next [17:59] what does virtual-maas do differently? [17:59] uses kvm [18:00] in what way? I've got a hypervisor [18:01] you manage all your nodes and maas server with libvirt [18:01] regulär maas has that built in too, does it not? [18:01] maas has support for "power" control via virsh [18:01] virtual-maas just uses that. [18:02] and http://bazaar.launchpad.net/~virtual-maasers/charms/precise/virtual-maas/trunk/files [18:02] has tools to create nodes and insert them into maas. [18:02] and remove them. [18:02] ok because virsh is a part of libvirt [18:02] kurt_: this helps you automagically deploy maas+nodes [18:02] virsh is a part of libvirt, yes. it wouldn't be hard to extend maas's power control to cover virtualbox. [18:03] ok, I just figured out how to do this all manually on my own, so I'm curious about the benefit [18:03] and if you have some reason that you want virtualbox (like you want to use this on a mac) then that might make sense to re-invent things. [18:03] but what is there in 'virtual-maas' is very close to "just works". [18:03] as a "all in one maas" with virtual nodes. [18:04] for some reason I remember reading virtual-maas not going to be updated too [18:04] kurt_: i posted a blog article on setting up maas within lxc [18:04] stokachu: I'm working on blogging my results too [18:04] kurt_: http://astokes.org/running-maas-vagrant/ [18:04] I'll have a look [18:05] kurt_, sothere are some parts of virtual-maas that are hacky. [18:05] its basically all in one. [18:05] it doesn't survive reboot [18:05] those htings can be fixed for sure [18:05] I just found out about the tagging thing, which will make life exponentially better [18:06] I've managed to get MAAS/Juju and Openstack fully up and running [18:06] on VMWare Fusion and Workstation that is [18:07] anyways - you have now piqued my curiosity [18:07] to see how much work I did for NOTHING :D [18:07] lol [18:08] smoser: so is virtual-maas officially supported on precise? [18:09] "officially", no [18:09] it will probably work. [18:09] its really a development thing. [18:09] right [18:10] but there is a lot of generally usable things there. [18:10] ie, scripts to add a node to libvirt with 2 disks and X memory and then shove that node into maas [18:10] and then (generically) power that machien on and install it with release X [18:11] the only hacky things really are in getting maas up and going and listening on the maasbr0 network and the networking involved. [18:11] ok. [18:11] so has anyone run maas in virtualbox on windows? :) [18:11] workstation :) [18:11] kurt_, really. give it a try. if you have a system with 8G memory, its fairly easy. [18:11] kurt_: did you use vagrant or just manually through vmware? [18:11] slow, but easy. [18:12] and you can play. it is a very easy way to get a juju going [18:12] manually [18:12] as I said, it works quite well [18:12] kurt_: ok, ive got a pretty automated thing ive setup if you dont mind testing on windows [18:12] im not done yet but hopefully tomorrow [18:12] it would use virtualbox though [18:13] stokachu: thing = a package, scripts? [18:14] kurt_: yea using vagrant+virtualbox [18:14] but automated [18:14] just curious if it is portable across mac/windows [18:15] if it means I have to take down my workstation install, I'm afraid its going to have to wait - I'm knee deep in blogging [18:16] it wont touch your vmware stuff [18:16] in any case just let me know when/if youre interested [18:16] ok. I'm playing with ceph next too [18:17] kurt_: nice, dont forget to write your findings i'd be interested in looking at that too === matsubara-lunch is now known as matsubara [18:23] will do ;) [19:14] is there a way to alter the preseed file or do i need to edit it manually [19:15] http://maas.ubuntu.com/docs/development/preseeds.html i see this page [19:18] im trying to force the enlist server to be on a specific network device [19:18] seems to default to the first device no matter what [19:20] whats the best way to override http://paste.ubuntu.com/6189451/ [19:20] thats from maas_local_settings.py [19:30] stokachu: can't you go in to the interface and specify which device its doing dhcp on? [19:30] interface = web gui [19:30] lemme check [19:31] eth1192.168.50.255/24Manage DHCP and DNS [19:31] right [19:31] so its using eth1 [19:31] correct [19:31] but the default maas url is set to 10.0.2.15 [19:31] which is eth0 [19:31] you need to dpkg-reconfigure [19:31] set it to the IP of the maas url you want [19:32] in my dhcpd.conf? [19:32] no [19:32] hang on [19:33] dpkg-reconfigure maas-region-controller lets me change the maas url [19:33] right [19:33] that's what I meant [19:33] ah ok, how do i do this programmatically though [19:33] you mean from the maas-cli? [19:34] sure or any other avenue [19:34] * kurt_ crickets [19:34] lol [19:35] hmm if only i could pass an ip to dpkg-reconfigure [19:35] during setup [19:35] there has to be a way [19:35] did you fully research the maas-cli options? [19:35] ill look through it again and see if im missing something [19:35] hmm…well that wouldn't work too well [19:36] because the authentication is tied in to the url for the maas-cli [19:37] i found this http://paste.ubuntu.com/6189525/ [19:37] that's a roaksoax or bigjools question [19:38] not sure if enlistment_domain would work or not [19:38] why do you need to change the maas url once or twice? [19:39] so when i install maas onto a server with eth0 and eth1 i need mass to only listen and respond on eth1 [19:39] for whatever reason eth0's ip is being picked for the maas default url [19:39] you could hack IP forwarding to make it not matter :) [19:40] true [19:40] which is what I've done as I recall [19:40] ive edited the maas_local_settings.py for now which seems to provide a proper preeseed [19:40] but I've always just gone back in and reconfigured as you put in above [19:41] ok [19:41] I need IP forwarding on my virtual network anyways [19:41] that's the only way my maas nodes are going to get access to the internet [19:41] so its always just worked for me [19:47] stokachu: btw this is the best guide on easy IP forwarding in case you didn't have an easy source [19:47] http://askubuntu.com/questions/95199/two-network-cards-and-ip-forwarding [19:48] thanks [20:26] i think cloud-tools archive died [20:41] The following packages have unmet dependencies: maas : Depends: maas-region-controller but it is not going to be installed [20:42] smoser: ^ was there a recent push to archive [20:47] guess that should be roaksoax ^ [21:04] should this work? curl http://169.254.169.254/2009-04-04/meta-data/ (from a maas node) ? [21:08] bah ithink the archive is broken :( [21:16] Anyone seen this error with juju/maas from the UCA: agent-state-info: '(error: could not access file ''tools/juju-1.14.1-precise-i386.tgz'': [21:16] gomaasapi: got error back from server: 401 UNAUTHORIZED (Nonce already used: [21:16] 33410643))' [21:16] machines 1-18 went fine, on deploying this one, I got this error. === CyberJacob is now known as CyberJacob|Away === matsubara is now known as matsubara-afk [23:21] is anyone else using the cloud-tools archive? [23:21] it seems to be busted [23:48] roaksoax: ^ [23:51] stokachu: why is that? [23:51] stokachu: what are you experiencing? [23:52] 5 [23:52] 4 [23:52] 3 [23:52] 2 [23:52] 1 [23:52] * roaksoax dead [23:52] :) [23:53] roaksoax: sec [23:54] roaksoax: http://paste.ubuntu.com/6190310/ [23:55] stokachu: are you using saucy? [23:55] this is on precise using the cloud-tools repo [23:55] this seems because tierh package has a higher version and it is trying to install a lower version [23:55] stokachu: weird, I'll test later [23:55] gotta go [23:55] ok [23:56] deb http://ubuntu-cloud.archive.canonical.com/ubuntu precise-updates/cloud-tools main [23:56] deb-src http://ubuntu-cloud.archive.canonical.com/ubuntu precise-updates/cloud-tools main [23:56] thats my sources.list