[06:15] hello, I have maas installed,and there is only one dhcp & dns server which is maas itself in the network, but machines in the same network can not be booted via pxe, it said no proxyDNCP server found. anyone has an idea? [06:20] execuse me , but anyone knows which dhcp service/port maas provide? [06:21] raywang: maas just relies on the system's ISC DHCP and bind server [06:21] standard ports [06:21] and it auto-configures them if you tick the right UI boxes [06:21] bigjools, the client (machine) said, No DHCP or proxyDHCP found. [06:22] proxy? [06:22] is it on the same lan segment? [06:22] bigjools, so for the check, what port i need to check if they are listening on maas? [06:22] bigjools, yes, [06:22] bigjools, there is only one switch, which connect maas and client (two machines) [06:23] try dhcping to see if it finds the server [06:23] bigjools, i use maas from precise repos btw [06:23] ah [06:23] precise is a bit broken, we're SRUing a big change soon [06:23] bigjools, how do i dhcping to see? :) [06:23] that's the name of a package/command line util [06:24] bigjools, but the problem is on client side, it can not be pxe booted via dhcp, it said "No dhcp found" [06:25] bigjools, so how do I dhcping the server? you mean run dhcping on maas server? [06:25] yes [06:25] then try another machine on the same lan [06:26] you might need to restart cobbler [06:26] reconfigure it and restart I mean [06:30] ihashacks, cobbbler comes from maas-provision, dpkg-reconfigure maas-provision? [06:30] bigjools, cobbbler comes from maas-provision, dpkg-reconfigure maas-provision? [06:30] sorry ihashacks :) [06:30] raywang: no, "cobbler sync" IIRC [06:30] it makes sure its config is flushed [06:31] thankfully cobbler is going the way of the dodo soon [06:31] bigjools, is there any wiki to introduce how to use new maas? [06:31] raywang: Docs on maas.ubuntu.com [06:32] step-by-step instructions [06:32] thanks bigjools [06:32] wselcome [06:32] and welcome [06:34] bigjools, dont' work, i have run "dpkg-reconfigure maas-provision, cobbler sync, restart the maas-pserv, maas-tx* services and dnsmasq service", the client still can not be pxe booted.which use to work. :( [06:35] bigjools, so i wonder how to check the dhcp/tftp/ services on maas is working? [06:37] is the dhcp daemon running? [06:38] bigjools, i think so, dnsmasq is running [06:38] something wrong with your booting node, I expect [06:38] can anything pxe boot? [06:40] bigjools, for now, it can't, it use to work [06:40] what did you change? [06:42] bigjools, just shutdown the maas server and juju bootstrap server, move the machines to another room, booted them up, and they are fails [06:43] bigjools, run juju status return nothing, so i want to "destroy-environment", and try to bootstrap again. but when run "juju destroy-enviroment" always time out, [06:43] I expect you have a network problem [06:44] bigjools, run juju boostrap, says I need a environment which is there, so I start to look at the maas UI [06:44] bigjools, and in maas UI, but bootstrap server is still "allocated to root" [06:44] so destroy-environment didn't release the bootstrap server, [06:45] I want to remove the node from maas db via "from maasserver.models import *" [06:45] but it says 'DJANGO_SETTINGS* " is not set. [06:46] raywang: I have no idea what you broke [06:47] I would blow it away and reinstall [06:47] the precise release is buggy as hell [06:47] bigjools, ok, thanks a lot anyway, at least you help me to sort of my mind. :) [06:48] the biggest problem is that cobbler gets out of sync with maas [06:48] bigjools, so by reinstalling the maas, what version do you recommend? [06:48] use the one from the stable PPA [06:48] it is the version that gets SRUed soon (or thereabouts) [06:48] bigjools, one the from stable ppa use cobbler right now, right? [06:49] no [06:49] it's about a thousand revisions newer [06:51] bigjools, but I just install the maas, which is 0.1+bzr482+dfsg-0ubuntu1, from precise/main [06:51] bigjools, this one use cobbler [06:51] raywang: that's not a PPA package [06:51] you want ppa:maas-maintainers/stable [06:51] bigjools, ah, ok, get it [06:52] bigjools, do you have the timetable when the maas from stable PPA be SRUed to precise-update? [06:53] raywang: end of this week if you believe roaksoax :) [06:53] bigjools, ok, i'm really looking forward to it. :) [06:53] bigjools, thanks a lot [06:53] you and me both! [06:54] :) [07:00] raywang: enter the python shell from 'maas shell' [07:00] raywang: that will avoid the DJANGO_SETTINGS_ is not set error [07:01] trevorj, ok, any instrodution to remove a "allocated to root" node form maas? :) [07:01] raywang: otherwise you can set it yourself to maas.settings [07:01] raywang: yeah, do this [07:01] raywang: maas shell [07:01] raywang: in there, run 'from maasserver.models import Node' [07:01] raywang: n = Node.objects.get(hostname__startswith='beginningofhostname') [07:02] raywang: n.release() [07:02] raywang: that will unallocate it [07:02] raywang: if you want to then delete it do n.delete() [07:02] raywang: it's standard django ORM [07:03] raywang: the quantal maas using the experimental repo is much nicer, btw [07:03] raywang: but it's a pain to get running, I must say ;) [07:04] trevorj, thanks, it's very helpful, i will try it later [07:04] raywang: np [07:05] raywang: install ipython before you run maas shell, it'll make it alot nicer [07:06] trevorj, the one in stable PPA is quantal maas or not? [07:06] quantal has a newer version included [07:07] maas in maas stable ppa is quantal ? [07:07] I don't know to be honest [07:07] ok [07:07] I'm using the testing ppa [07:07] brave man [07:07] I don't recommend using that unless you've had experience with Django and such though [07:07] bigjools: I needed to get quantal nodes deployed [07:08] trevorj, ok, so maas in stable ppa wold be much safer :) [07:08] raywang: definitely [07:08] raywang: it doesn't use cobbler and such [07:09] raywang: the newer versions I mean [07:09] raywang: so it's a bit less finicky [07:09] It appears they wrote a tftp server in twisted python as well, fantastic stuff [07:10] I'm still going through the code, it's very well done [07:11] * bigjools feels the love [07:12] bigjools: you a dev? [07:12] allegedly I am the lead, yes :) [07:12] Wow, nice. [07:12] bigjools: I've been following it for about a year [07:12] glad you like it [07:12] bigjools: very good work, especially the recent stuff! [07:12] bigjools: makes my Django work look rather unprofessional I must say [07:13] well you can thank allenap, jtv and rvba for most of it [07:13] Huh? What did we do now? [07:13] lol [07:13] jtv: we're getting thanked for maas! [07:13] * jtv checks calendar [07:13] Sorry, just being silly. I do that sometimes. :) [07:13] the mayan apocalypse was last year [07:14] ahaha [07:14] Ah yes. Slept right through it. [07:14] Anyway, trevorj, stand-up comedy is our day job but we do some coding on the side to make ends meet. Glad you like it. :) [07:14] If I submit some commissioning stuff for HP iLO configuration in the user data, does it have a chance to be included? [07:14] Are you guys interested in vendor specific stuff like that? [07:14] sure [07:14] roaksoax is doing a fix for that actually [07:15] but if it's generic we'll take anything, or at least put it in docs [07:15] awesome [07:15] what's your fix? [07:15] I've got it working in python script form, but so far it sets the boot order [07:15] and sets the server name to the hostname [07:16] among with other things just for us internally [07:16] I did use a python library however, I'm not sure what the policies for that are? I used python-hpilo as I hate writing HPONCFG XML files. [07:17] check with roaksoax and he'll guide you, most of that stuff is done by him [07:17] really? wow. [07:17] Love reinventing wheels. [07:18] will do though. [07:18] well I mean the ilo config was originally coded by him :) [07:18] Ah, ok, cool [07:18] it has bugs [07:18] and ilo seems a bit, errr, funky [07:18] for how little code it is it works though [07:18] yeah iLO is always funky [07:18] not in a good way [07:18] there's some weird stuff where it programs the wrong slot when enlisting [07:19] the wrong blade slot? [07:23] Somehow some of my maas created logins were not working, is that related? [07:23] Only a couple of them, 2/9 [07:23] the user slot in the bmc [07:23] I looped through the nodes and reprogrammed them using python-hpilo to what they should be and after that it's been golden [07:25] trevorj: https://code.launchpad.net/~andreserl/maas/ipmi_usercreation_ilo_versions_trunk/+merge/148579 [07:26] bigjools: nice, looking through now [07:27] Ah, that's what you mean by slots, and that's why I see 'User3' all over the user data [07:27] it all makes sense [07:30] yup [07:31] Are there any plans for using MaaS to boot from network disks or even into ram? [07:32] no, the only major item upcoming is the fast installer [07:32] image based? [07:32] yes [07:32] fantastic. [07:32] I know - can't wait [07:32] would make my testing life 1000x easier [07:33] every time I deploy a node I wonder why it isn't done via images, so I'm glad that's on the radar ;) [07:33] d-i was the easiest way at the time [07:33] because there was no actual work to do it :) [07:33] yeah, I can see why [07:34] I used to do preseeds myself a few years ago for such things, it gets a bit hairy after a while imo [07:34] I was playing with debirf and then started modifying ltsp to to handle pxe booted kvm hosts a year ago [07:35] I like the idea of booting from an RO image, but it's even more hairy [07:36] MaaS has definitely simplified the PXE install scenario for Ubuntu [07:37] Will the images be downloaded preconfigured or will they be generated on the spot? [07:37] not sure yet [07:38] Would be nice to be able to easily customize them if they have build tools of some sort [07:39] Oh, you know, you're the guy to ask about my avahi issue I suppose [07:39] My DNS domain is a child domain of .local, so avahi takes precedence of revolves [07:39] I disabled avahi-daemon in the generic preseed to work around this [07:40] but I'm afraid of a package pulling it in, so my options are to customize nsswitch.conf or find/make a way for avahi-daemon to not resolve a certain domain and it's children [07:41] Unless you know of something I'm perhaps missing? [07:44] I think you need to ask on the server channel about that [07:45] I have, I'll ask again tomorrow though [07:47] It mainly causes an issue with juju and less so with maas, I forgot I wasn't in the juju channel [09:41] hi there. a folsom openstack installation does not let the admin user log in the dahsboard. [09:41] the keystone charm has set the admin user password correctly, nova commands works all right [09:42] when trying to log with the admin horizon, the only funny log is something that mention "%SERVICE_TENANT_NAME%" (defautl setting in some config file) [09:42] http://pastebin.ubuntu.com/5569969/ (see line 442 and 455) [09:43] there are some file wich have still the defualt %SERVICE user and pass set, but this is also the case on a deployment where things works. [09:43] any idea what to do, where to look, well, how to investigate this ? === kentb-out is now known as kentb === matsubara is now known as matsubara-lunch === kentb is now known as kentb-afk [16:06] rvba: you are not in today right? === matsubara-lunch is now known as matsubara [19:29] Hi! I have a question about disaster recovery. [19:30] let's say I have a node with a failed motherboard, but the disk is fine. [19:30] I'd like to swap the mobo and boot it up again, but I know MAAS will not recognize it because the MAC address changed. [19:31] what do I need to do to update that mac address? [19:31] as far as I can tell I just need to update the maasserver_macaddress and maasserver_dhcplease tables in Postgres? [19:31] et1337: edit the node in question and add an additional mac address and remove the old one [19:32] you can do that from the webui for sure, and I'm guessing you can do that through the maas cli [19:32] oh dang, awesome. I can't believe I dove into postgres before the ui. hah. [19:33] I see it now. sweet, thanks. [19:33] cool [19:35] gotta love easy fixes [19:38] indeed === kentb-afk is now known as kentb === matsubara is now known as matsubara-afk === kentb is now known as kentb-out