[01:58] Has anyone run into issues w/ 13.10/Saucy commissioning image not preseeding installation (auto configuration for enlisting)? [02:02] Things looked real good on 12.04/Precise except was having issues w/ juju. So I went to 13.10, but now the commissioning image goes to manual install. [02:03] http://paste.ubuntu.com/6327603/ pxe boot from maas fails, is this my corner case? or worth a bug report? [02:04] I'm running a fresh installed saucy, installed maas from archive [02:04] in a vm [02:04] saucypreseed: use precise for commissioning if you can [02:04] nice name btw [02:04] freeflying: looking [02:05] freeflying: I have seen that on my network too - It happens when you mix 10/100 switches on a network [02:05] bigjools, thanks, hypervisor runs saucy too, vm connect ovs [02:06] freeflying: either way tftp packets are getting dropped so it times out [02:07] bigjools, tcpdum shows clients request sent to tftpd [02:07] are you mixing switches? [02:07] bigjools, is saucy commisioning broken? I've been hacking the initrd images to include my ethernet drivers... [02:08] saucypreseed: it's possible, I think LTS images are the ones that get the most testing [02:08] bigjools, no, both vms run on same hypervisor [02:08] did you try the fast installer? [02:08] freeflying: ok then I don't know, sorry [02:08] bigjools, thanks anyway [02:09] freeflying: sure thing. I'd try a ping flood and see how many packets it drops [02:09] bigjools, nice, thanks for the hint [02:09] bigjools, thanks. Good idea. I'll try to resurrect my precise commissioning initrd [02:11] *kicks self for having an iWARP card* [02:11] bigjools, ping from my laptop to the vm didn't see any packet lose [02:12] freeflying: well do it between your node and your cluster controller [02:12] since that's where you see problems [02:19] freeflying, does your ovs have spanning tree enabled? [02:19] saucypreseed, no, didn't configure it [02:20] You get a DHCP lease but TFTP doesn't start? [02:21] saucypreseed, did see tftpd, but it doesn't response upon client's request [02:21] there is a bug in the upstream tftp code IIRC, it needs bigger timeouts [02:21] Can you TFTP from another machine to the cluster controller? [02:22] saucypreseed, tried to get pxelinux.0, tells file doesn't exist [02:23] Ya I ran into that once [02:23] Is it on the file cluster controllers file system? [02:24] saucypreseed, cluster and regional are on the same machine, and sure tftp too [02:24] I think it in /var/lib/maas/tftp// or something like that [02:24] yep [02:25] and /var/lib/maas/tftp has uid:gid of maas, but the files/dirs under belong to root [02:27] did you see anyting in /var/log/maas/*.log? [02:28] I think when I had it there was an obvious error in /var/log/maas/celery*.log [02:28] something like run the download-pxe-files... [02:29] *embarrassing* [02:29] Good luck [02:29] saucypreseed, thanks, very helpful [02:30] gotta run thanks [07:00] bigjools, tested tftp between 2 vms, works fine, but maas-pserv still doesn't work === CyberJacob|Away is now known as CyberJacob [07:13] freeflying: timeouts? [07:13] and how did you test tftp exactly? [08:51] bigjools, i set up tftpd on another vm, and run tftp from within maas node, get a file from tftpd, works [09:25] freeflying: ok thanks [09:25] freeflying: please file a bug and add the pserv log [09:26] bigjools, on it :) [09:39] bigjools, https://bugs.launchpad.net/ubuntu/+source/maas/+bug/1246236 [09:39] Ubuntu bug 1246236 in maas (Ubuntu) "pxe boot from maas fails due to time out" [Undecided,New] [10:14] allenap: FYI ^^^ I think you looked at TFTP timeouts once? I can recreate it locally by hanging my 100M and 1G switches next to each other on my maas network [10:14] we talked about forking that tftp project didn't we? [10:15] anyway, good night [10:15] bigjools: We did, but I think it's okay how it is. Upstream does eventually land my patches. In the meantime we carry them in packaging. [10:15] bigjools: nn [11:16] allenap, hey [11:17] allenap, I'm seeing an odd situation where maas-dhcp is giving out the same IP address to two different macs [11:17] one ilo and the other a actual server - any ideas? [11:17] jamespage: Weird! I'm otp right now, but I'll help out as soon as I'm done. [11:17] allenap, ta [12:04] jamespage: Can you paste a log of that? [12:04] allenap, dhcp leases file? [12:06] jamespage: Yeah, that might be useful. [12:07] allenap, http://paste.ubuntu.com/6329619/ [12:07] the offending IP is 10.246.8.10 [12:16] jamespage: Is the IP address actually being used, or did you just notice this in the leases file? [12:16] allenap, no its being used [12:16] that was one instance [12:17] the entry at the top of the file is the server, the others are the ilo [12:17] and not the same server/ilo if that makes sense [12:34] jamespage: This /looks/ like a bug in isc-dhcp-server, however unlikely that may seem... which makes me fairly sure that I'm wrong. [12:35] jamespage: I'm going to dig some more after lunch. Is this blocking you or someone else? [12:36] jamespage: Also, can you file a bug report for this? [12:40] allenap, its not blocking because I ditched the leased file and made everything re-request DHCP leases [12:40] allenap, I can after lunch [12:40] * jamespage is hungry [12:40] hungry for bugs! [13:16] jamespage: Coolio :) [13:16] mgz: Are you coming to cow-ork tomorrow? [13:17] allenap: yeah, will also bring your shizzle [13:17] mgz: Ta muchly :) === Spideyman is now known as Spideyman_afk === Spideyman is now known as Spideyman_afk [14:28] allenap: have a moment? === sfeole` is now known as sfeole [14:47] tych0: Hiya. I have to go out briefly, but I'll be back. Perhaps rvba can help? [14:48] allenap: perhaps, danwest told me you had some ideas about how to do the seamicro implementation [14:48] i'm here now and i saw the box in all its glory earlier, so i can fiddle with it for the next few days [14:57] roaksoax, 'apt-get install maas-region-controller' wants to install isc-dhcp-server. do you know why? [14:57] would'nt / shouldn't that be a cluster controller dependency? [14:59] hm... maas-region-controller recommends maas-dns. maas-dns depends on maas-dhcp. [14:59] smoser: i think someone made maas-dhcp dependency of maas-dns [14:59] but that generally doesnt sound right. [14:59] its a recommends. [15:00] well, dns is a recommends. [15:00] and dhcp is a depends of dns [15:00] which doesn't seem right. [15:00] smoser: yeah ibrealized that too late to fix it in saucy === Spideyman is now known as Spideyman_afk === Spideyman_afk is now known as Spideyman === Spideyman is now known as Spideyman_afk === Spideyman_afk is now known as Spideyman === Spideyman is now known as Spideyman_afk === Spideyman_afk is now known as Spideyman === Spideyman is now known as Spideyman_afk === Spideyman_afk is now known as Spideyman === Spideyman is now known as Spideyman_afk === mwhudson- is now known as mwhudson [17:52] Is it a known bug that when running MAAS from a branch (make run), that the regional controller can't talk to the rest of the cluster? [17:52] * NCommander smacked the run script to get around it, but I'm concerned there's something funkly with my environment [18:22] Hey, does anyone know the method that the preseed file gets down to the commissioning image? [18:22] I've been trying to get MaaS going on Saucy but it keeps going into manual config rather than preseeding [18:24] I looked thru the init scripts in initrd, but failed to find any other case than /preseed.cfg (which of course is not present) === Spideyman_afk is now known as Spideyman [20:10] I have a question about tftp [20:11] I know it is a build-in from MAAS. But is it configured with cobbler? I am using 13.10! [20:18] Is there somebody! [20:19] list [20:19] help [20:19] quit [20:19] exit === Spideyman is now known as Spideyman_afk === Spideyman_afk is now known as Spideyman [21:24] Does the commissioning kernel need to have iscsi target statically built into it to mount the ephemeral image? [22:54] saucypreseed: no, maas sets the kernel command line