[00:06] qhartman: yes, preseed adjustment is the way to go, for now at least. [00:07] allenap: I think there is a caching proxy in the way somewhere which doesn't have the new layout === CyberJacob is now known as CyberJacob|Away === 21WAAD3MF is now known as wallyworld [02:32] jtv: http://paste.ubuntu.com/7311567/ [02:33] So you're getting it too? I'm not. [02:33] Though let me try trunk. [02:33] Whoops, yes, lots of errors! [02:34] Eenteresting. [02:35] ah [02:35] apparently [02:35] functools.wraps expects the callable being wrapped to have a __module__ attribute [02:36] Hey. [02:36] I'm getting this in my branch. [02:36] Of which I ran the tests endlessly yesterday without these failures. [02:36] I am getting this in trunk [02:37] So... package update? [02:37] I expect a django update broke things [02:37] Yes, I'm getting it in trunk too. Thing is, I wasn't getting this yesterday, and it's not a code change. [02:37] of which there was just one [02:37] * bigjools looks at the django change [02:38] jtv: yep I reckon this is a regression in the security update [02:38] * bigjools files bug [02:39] Might I recommend a priority on the high side? [02:41] jtv: https://bugs.launchpad.net/ubuntu/+source/python-django/+bug/1311433 [02:41] Launchpad bug 1311433 in python-django (Ubuntu) "REGRESSION: AttributeError: 'functools.partial' object has no attribute '__module__' " [Undecided,New] [02:41] can you confirm it :) [02:42] I registered my being affected by it. [02:42] Hope that's what you mean. [02:42] change the status [02:42] well it does that automatically I think [02:44] I didn't think the Confirmed status did much, except maybe make triagers think the bug doesn't need attention. [02:44] Oh, it's Confirmed now. [02:48] jtv: this is very serious, maas no longer runs at all in trusty [02:56] jtv: ok we can fix this locally [02:56] jtv: https://code.djangoproject.com/ticket/22486 [02:56] Django bug 22486 in Core (URLs) "urlresolvers.reverse() security fix (1.5.6) breaks when a view is a partial" [Normal,New] [02:57] I'll file a maas bug [02:58] I wonder where we're doing the partial thing. [02:59] src/maasserver/views/combo.py? [03:00] get_combo_view is the only thing (outside of tests) that looks to me like it might be doing this. [03:02] jtv: can you do me a favour and see if this patch fixes things? https://github.com/prestontimmons/django/commit/b63ae5c60619a257ad57cf6043e71f681283e47b [03:02] Looking.. [03:02] I was just trying a patch of my own. [03:03] it's a patch to django though [03:03] security team will apply it if we confirm it works [03:03] Ah. Meanwhile I seem to have a working patch to maas. [03:04] jtv: ah! [03:04] bigjools: my test run is not done yet, but this simple maas fix seems to work around the problem at least for us: http://paste.ubuntu.com/7311704/ [03:04] jtv: land it so it unblocks the lander at least [03:04] I'll start a VM to try the django patch. [03:05] Aye-aye sir. [03:05] Got a maas bug number? [03:05] jtv: same number, I added a maas task [03:05] OK [03:20] jtv: any news? [03:21] mdeslaur: about what? [03:21] the bug? [03:21] I'm landing a workaround in maas. [03:21] jtv: yeah, the upstream patch [03:21] I'm building security regression fixes with the commit from the upstream bug [03:21] I'm still setting up for the attempt. [03:22] By the way, you wouldn't happen to know a way to get a plain diff from that github page? [03:22] jtv: add ".patch" to the end of the url [03:23] Ah, of course! Why didn't I figure that out myself! [03:23] [03:23] Thanks. :) [03:23] yeah, super-intuitive [03:23] I just uploaded a fixed package for trusty here: https://launchpad.net/~ubuntu-security-proposed/+archive/ppa/+packages [03:23] and will upload for the other releases in a few minutes [03:24] Are you saying I could just use that package instead of applying the diff? [03:24] jtv: yes [03:24] Great, thanks! [03:24] and I would like to know if it fixes maas before I push them out ASAP [03:25] Yes, of course... I'm setting up to establish that. [03:28] jtv: cool, thanks === wallyworld_ is now known as wallyworld === vladk|offline is now known as vladk [04:02] mdeslaur: for maas, the proposed update fixes the failures! [04:02] jtv: awesome, thanks for testing! [04:02] I'll release them right away [04:02] (Although of course independently from that, I already landed a workaround in maas) [04:02] Thanks. This has bitten us pretty hard, so glad to see it being resolved for everyone. [04:02] thanks mdeslaur [04:03] and thanks for testing jtv [04:03] Still seeing _some_ test failures with the new package, but let me see what they are first. [04:03] May be unrelated. [04:03] bigjools, jtv: thanks! [04:03] xargs: env: terminated by signal 9 [04:03] That's something we've seen before... probably unrelated. [04:03] I'll make sure we add maas to our django testing going forward [04:03] I'm investigating. [04:03] mdeslaur: \o/ [04:11] Found the cause of those other test errors: VM ran out of memory. Not an issue. [04:13] jtv: thanks === CyberJacob|Away is now known as CyberJacob [06:57] gmb: did you get to the bottom of the "AttributeError: 'functools.partial' object has no attribute '__module__'" failure? [06:58] (Hi gmb, btw :)) [06:58] jtv: ^ [06:59] I'm asking because I'm seeing that locally now… and I saw that the python-django package has been updated… [07:03] rvba: I did [07:03] django fail [07:03] Yeah, I just saw the fixes. [07:03] update again [07:03] What a mess. [07:04] yes :) [07:13] Bbiam === CyberJacob is now known as CyberJacob|Away === vladk is now known as vladk|offline [09:24] Who's up for what I hope will be a simple review? https://code.launchpad.net/~jtv/maas/env-context-manager/+merge/216840 [09:25] jtv: I'll take it [09:25] Merci beaucoup. [09:25] Pas de problème. === vladk|offline is now known as vladk [10:38] I'm having some strange mirroring problems. [10:38] http://archive.ubuntu.com/ubuntu/ is fine [10:38] http://ftp.se.debian.org/ubuntu/ does not work [10:39] The installer claims the archive is broken, and syslogs says 403 forbidden when accessing http://ftp.se.debian.org/ubuntu/ [10:39] http://ftp.se.debian.org/ubuntu/dists/trusty/Release [10:39] but it works fine outside the installer. [10:40] Trying to access it from the interrupted installation shell, but the kvm/keymapping does not allow me typing slashes... trying to find some workaround [11:08] Could it be a problem with a proxy somewhere inbetween? [11:15] dont think there are any. However, when I dump traffic on the gw/router, I see e.g. icmp packets to the mirror when I ping, but no web requests. [11:25] Seems it's not a mirror problem. Same error with e.g. ftp.uk.debian.org/ubuntu and others. [11:25] *digging for proxies* [13:26] hola ! now that trusty is out, i guess i can just apt-get install maas and juju-core and all the tools without having to add any ppa or are ppa still relevant ? === ming is now known as Guest7729 [14:57] melmoth: i'm not using any ppa's and my MAAS cluster works a treat. (i'm using kvm virtualization, ymmv if on bare metal) [14:59] anyone have any tips for debugging the pxelinux.0 that comes with MAAS? Cobblers works fine, but with MAAS it barfs on some iSCSI stuff [14:59] i think it may be using the wrong interface [14:59] lazyPower, nodes will be kvm vm too , it s just for playing :-) [14:59] actually, even the maas box is a kvm vm. [15:00] ah, i went the other way around with that and kept the region controller on bare metal so i could reach the aas nodes from outside my network. [15:00] bridging into the VM was a bit of a heady process [15:03] im install maas-cluster-controller on a brand new trusty install and the apache restart is bailing due to missing crochet module [15:04] magicrobotmonkey: looks like a packaging bug. Please report a bug with all the details and we will get back to you ASAP. [15:04] ok ty [15:24] so, I'm setting up a multi-homed MAAS cluster, and the routes and DNS on the nodes that come up are all sorts of wrong. [15:24] I'm wanting to treat the MAAS-side as a management network, and as such isn't bridged / routed onto the main network. [15:24] but the MAAS config tool doesn't want to let me make the DHCP changes that I think need to be made to make this work. [15:25] since the external interfaces (which are internet connected) are managed by a different DHCP [15:25] so, what's the right way to do this sort of setup? [15:25] I'm sure I could hand-jam it, but that seems counter to the point of MAAS. [15:26] or should I have my MAAS controller setup to bridge between the networks and have the nodes see the outside world that way? [16:03] qhartman: i setup a bridge device on the region controller - but my maas nodes are kvm machines and not bare metal. [16:03] its kind of dependent on your network topology [16:05] right [16:05] yeah, I've setup my controller to do forwarding now as well. We'll see if this actually fixes the Juju problem. [16:05] I'm nuking the other machine I was working on and starting with a fresh node [16:07] is there a strategy for importing pxe files from a host that can't reach the internet? [16:07] the region controller can [16:52] magicrobotmonkey: you should only need to import the pxe files on your region controller. I'm fairly certain that hands off the files to the zone controllers. [16:53] i could be wrong though - needs citation [16:53] hmm it doesnt seem like the docs say that [16:53] but it turns out i have some network issues to resolve first anyways === vladk is now known as vladk|offline === roadmr is now known as roadmr_afk [17:49] how do you login to a maas-cluster-controller from the cli? === CyberJacob|Away is now known as CyberJacob === vladk|offline is now known as vladk === roadmr_afk is now known as roadmr === matsubara is now known as matsubara-afk === roadmr is now known as roadmr_afk === roadmr_afk is now known as roadmr === matsubara-afk is now known as matsubara === vladk is now known as vladk|offline [21:16] jhobbs: Hi there. Have you had a look at the “HWE streams” thing? [21:22] allenap: a little bit, I wrote a doc on some of it https://docs.google.com/a/canonical.com/document/d/1CJlIu2wapKJgDl-MXmvK_RseqBHeWJQ6S-gJzn0bh34/edit#heading=h.eeybgo5224oo [21:24] jhobbs: Nifty, I’ll read that before the meeting. === CyberJacob is now known as CyberJacob|Away