[01:17] <roaksoax> bigjools: we'll have to test that with machine with md raid arrays
[01:18] <bigjools> roaksoax: fair enough.
[01:18] <roaksoax> bigjools: and I don't know whether we wanna do that
[01:19] <roaksoax> bigjools: btw.. is there any ETA on the cluster url thing for the preseeds?
[01:19] <bigjools> roaksoax: this week
[01:20] <roaksoax> bigjools: ack! thanks
[01:29] <roaksoax> bigjools: btw.. are we going to start releasing trunk for precise?
[01:29] <bigjools> roaksoax: no!
[01:29] <roaksoax> bigjools: one of the things that came up last week was the ability to have custom kernel command lines
[01:30] <bigjools> I know
[01:30] <roaksoax> as they do some costumization some times
[01:30] <roaksoax> ack
[01:30] <bigjools> people want their cake and eat it - a stable LTS but at the same time all the shiny new features
[01:31] <bigjools> from my PoV I am happy to release trunk to precise but you'd have to work hard with the techboard, no?
[01:31] <roaksoax> bigjools: oh I meant PPA
[01:31] <roaksoax> not actually SRU it
[01:31] <bigjools> ok
[01:31] <bigjools> we can do a backports PPA
[01:32] <roaksoax> yeah that's what I meant
[01:32] <bigjools> no problems with that :)
[01:32] <bigjools> roaksoax: do you know anything about this? https://bugs.launchpad.net/maas/+bug/1087183
[01:33] <roaksoax> bigjools: i know the context, but I think the removal of it didn't really fix the issue james was having
[01:33] <roaksoax> you'll need to check with him and scott just to be sure
[01:33] <bigjools> ok I'll leave it incomplete, thanks
[01:34] <bigjools> hey, you have nothing better to do on a Sunday night? :)
[01:34] <roaksoax> bigjools: ok so this wewk I'll make sure all the MAAS dependencies are in precise, and then hopefully upload maas by thursday, wtih 2 fixes I want to fix. 1. ipmi stuff, 2. fix a packaging bug on upgrade
[01:34] <bigjools> woohoo
[01:34] <bigjools> roaksoax: what about testing with Django?
[01:35] <roaksoax> bigjools: heh.. was doing homework, then learning how to dj, and now preparing my notes from last week to send to you tomorrow
[01:35] <roaksoax> bigjools: I think we are good with that. ScottK did say he was fine with it if the techboard approved it.But i don't know who can test it apart from the maas team
[01:35] <roaksoax> bigjools: so i'll check on that this week too
[01:36] <bigjools> roaksoax: well we need to run up some other Django apps with the patched Django
[01:37] <roaksoax> yeah, bigjools horizon (openstack) could be a good example
[01:37] <bigjools> ok
[01:39] <roaksoax> bigjools: anyways, I'm off, have a good day!
[01:40] <roaksoax> oh btw.. thanks again for the help last week!
[01:40] <bigjools> roaksoax: pleasure, have a good evening
[01:40] <roaksoax> thanks!
[10:55] <Wout> I'm having an issue with maas where my node names are not consitent with the ip address
[10:57] <Wout> for example: 192-168-30-55.master = 192.168.30.50 not 192.168.30.55
[10:58] <Wout> nslookup on maas master doesn't use local server
[10:59] <bigjools> you might be falling foul of a problem in the dhcp server
[10:59] <bigjools> where it assigns different IPs to the same machine
[10:59] <bigjools> depending on whether it's requesting IP from the BIOS or from the OS
[11:01] <Wout> The OS
[11:04] <Wout> what does maas use for dhcp?
[11:04] <Wout> sorry DNS
[11:04] <Wout> what does it use for dns
[11:04] <bigjools> bind
[11:06] <Wout> does maas create a mac <-> reserve somewhere?
[11:06] <bigjools> what do you mean by reserve?
[11:07] <Wout> I can see dns name in /etc/bind/maas/zone.30.168.192.in-addr.arpa
[11:08] <Wout> it say's ip 55 in range 192.168.30 is DNS name 192-168-30-55.master
[11:08] <bigjools> maas scans the leases file that the DHCP server maintains, and keeps track of IP vs MAC in its database
[11:08] <Wout> where does maas link 192.168.30.55 to a mac address
[11:09] <bigjools> you probably have this issue with the dhcp server: https://bugs.launchpad.net/maas/+bug/1069570
[11:09] <Wout> I'l look at that
[11:10] <bigjools> someone is looking at a fix on the dhcp server
[11:13] <Wout> wow looking at the bu report...
[11:13] <Wout> this s*cks
[11:13] <Wout> I'm implementing this now
[14:59] <roaksoax> bigjools: btw we didn't experience that bug above
[15:55] <Wout> Anyone in heer been able to de pxe maas with uefi?
[15:56] <Wout> MAAS doesn't create the boot files foor uefi systems
[15:56] <Wout> I need to enlist a uefi only system
[15:56] <Wout> How would I do that?
[15:57] <Wout> Anyone there?
[16:04] <Wout> Hello??
[23:17] <bigjools> smoser: can you respond to my comment here please? https://bugs.launchpad.net/maas/+bug/1116331
[23:18] <bigjools> Oh I see you commented on the MP
[23:19] <bigjools> BTW if you open up the expander at the bottom of the MP page you can set it to WIP immediately so no emails get sent containing bogus diffs
[23:32] <bigjools> how can I help mwhudson?
[23:32] <mwhudson> bigjools: hello
[23:32] <mwhudson> bigjools: i have a task which i've just realized sounds very much like maas
[23:33] <mwhudson> or rather, one small part of maas
[23:33] <mwhudson> so i wonder if can reuse it or at least see how it does it
[23:33] <mwhudson> bigjools: basically i think i want a super dumb netinstaller
[23:33] <bigjools> this is pretty much what maas does, yes
[23:33] <mwhudson> that can be configured to say "when the device with mac address $foo boots, dump this disk image onto it"
[23:34] <bigjools> maas is in control of the boot
[23:34] <bigjools> it powers up the machine
[23:34] <bigjools> so if you can deal with that, maas will work nicely
[23:35] <bigjools> it'll even put your ssh key on there
[23:35] <mwhudson> this is for lava, automated testing stuff
[23:35] <mwhudson> i'
[23:35] <mwhudson> for now i'll be doing stuff over serial console to control the node after it boots the desired image i think
[23:36] <bigjools> so maas serves up different boot images depending on what state it thinks the node is in
[23:36] <bigjools> therefore unless maas thinks the node is booting, it won't do much
[23:36] <bigjools> maas also uses cloud-init
[23:36] <mwhudson> right
[23:36] <mwhudson> i think perhaps i only want a very small part of maas
[23:37] <bigjools> it would be hard to use maas in bits
[23:37] <mwhudson> there is always c-c c-v :)
[23:37] <bigjools> all you need then is a tftp server, a dhcp server, and some boot images
[23:37] <mwhudson> probably
[23:38] <mwhudson> but i don't want the default netinstaller initrd i think
[23:38] <bigjools> and frig the kernel options so it starts a net install
[23:38] <mwhudson> i just want something that splats a tarball into a partition
[23:38] <bigjools> we use iscsi
[23:38] <mwhudson> none of this packaging stuff
[23:38] <bigjools> ah
[23:38] <bigjools> maas doesn't do that - yet.  should be ready RSN
[23:39] <mwhudson> oh ok
[23:39] <bigjools> it currently runs d-i
[23:39] <mwhudson> ah
[23:39] <bigjools> but work is afoot to splat an image
[23:39] <mwhudson> so you don't have anything like an "ami for maas"?
[23:39] <bigjools> no
[23:39] <mwhudson> ok
[23:39] <mwhudson> ffs
[23:39] <bigjools> not yet
[23:39] <bigjools> give it a week :)
[23:39] <mwhudson> why am i always at the bleeding edge of everything :)
[23:40] <mwhudson> bigjools: what does maas use for dhcp/tftp etc?
[23:40] <bigjools> internal TFTP implementation, but ISC DHCP
[23:40] <mwhudson> ok
[23:41] <bigjools> bleeding edge = on the curve man
[23:44] <mwhudson> bigjools: is pserv the tftp bit?
[23:45] <bigjools> mwhudson: yes, twisted process
[23:45] <mwhudson> cool
[23:47] <mwhudson> bigjools: and all this is doing the stuff that cobbler used to do?
[23:48] <bigjools> mwhudson: kinda.
[23:48] <roaksoax> bigjools: https://lists.ubuntu.com/archives/ubuntu-devel-announce/2013-February/001012.html
[23:48] <bigjools> roaksoax: I saw that, good stuff
[23:49] <roaksoax> bigjools: ok cool, so what you think about the GeneircIPAddressField?
[23:50] <bigjools> roaksoax: yes we could pull that in.  I don't like doing that though.
[23:51] <mwhudson> bigjools: generally maas uses the regular initrds that the distro provides?
[23:51] <roaksoax> mwhudson: it does
[23:51] <bigjools> mwhudson: yes
[23:51] <mwhudson> ok
[23:51]  * mwhudson has found class TFTPBackend(FilesystemSynchronousBackend): now
[23:52] <roaksoax> bigjools: so what do you think we should do?
[23:53] <bigjools> mwhudson: the tftp server is tightly integrated to maas as it serves up different things depending on the state of the node
[23:53] <mwhudson> yeah
[23:53] <bigjools> roaksoax: do we have a choice?
[23:54] <bigjools> roaksoax: can we carry it as a quilt patch?
[23:54] <roaksoax> bigjools: yes
[23:54] <roaksoax> bigjools: that's what *should* be done
[23:54] <bigjools> roaksoax: probably easiest
[23:55] <bigjools> since it's easiest to remove
[23:55] <roaksoax> indeed
[23:55] <bigjools> and it's distro-specific, not upstream issue
[23:56] <bigjools> mwhudson: you could serve up an ephemeral image to a booting node and use a cloud-init script to splat a tarball
[23:56] <roaksoax> bigjools: yeah, so ;et's do that. That's the easiest, to ship it as a patch
[23:56] <bigjools> roaksoax: sweet
[23:56] <bigjools> mwhudson: then you can just boot something and maas will install it
[23:57] <bigjools> all highly unsupported :)