[05:24] anybody here familiar with initramfs? So i was trying install 10.04 on a netbook and it got hung on 95% of the installation forever so like an impatient kid I yanked it out of the usb port.how do i get the usb flash to install? [05:29] anybody familiar with initramfs? === chuck_ is now known as Chux [05:48] found a bug on quantal aternate i386 https://bugs.launchpad.net/ubuntu/+source/debian-installer/+bug/1040427 [05:48] Ubuntu bug 1040427 in debian-installer "oem/lvm installation of quantal alternate i386 iso failed : Unable to locate package python-couchdb , subunit , python-subunit , python-junitxml" [Undecided,New] [06:16] slangasek, ^ [06:17] if you're alert enough, can you look at it? otherwise maybe cjwatson can look? [08:07] i am trying to do a net install on some servers, and it seems like since yesterday, dhcp configuration is failing [08:07] is anyone aware of this [08:34] cjwatson: I have a problem with LTS-LTS upgrade that skaet asked me to do yesterday night [08:34] cjwatson: https://bugs.launchpad.net/ubuntu/+bug/1040494 [08:34] Ubuntu bug 1040494 in ubuntu "Unable to upgrade from Lucid to Precise 12.04.1" [Critical,New] [08:35] cjwatson: I have the sneaky feeling that my 16 GB USB may be the culprit and I need to reinstall lucid and try with a 2GB one [08:35] cjwatson: but before that I wonder if you need any further information [08:37] I'm buried in a stack of quantal livefs build failures - can somebody else look? [08:38] You need to at least extract some useful logs from that, I think - the screenshots are pretty generic [08:38] cjwatson: ok, I was trying not to connect to internet because of the test case I was running, now that it has aborted I will run apport to colect more logs [08:52] cjwatson: added all the dist-upgrade folder for reference [11:44] * mpt wonders where xnox is [11:44] mpt: Sleeping:) [11:44] LOL [11:54] Anyway xnox's here, mpt:) [11:54] * xnox irc's been broken. [11:54] The summons worked! [11:54] need to read backlog. [11:54] what's up? [11:55] xnox, how is your installer work going? Do you need any design clarifications or other help? [11:55] xnox: Why isn't there a Wubi channel in IRC, btw.. [11:55] mpt: yeap. there is the question of shortening "fair password" to just "fair" that I've emailed you about. [11:56] mpt: and the encryption settings, that I didn't email you about. one moment. [11:58] Sorry, xnox, I forgot about the passwords mail. [11:58] mpt: that's ok. it was a bit out of the blue =) [11:59] mpt: also, have you seen the "gnome accounts" dialog? They put the password strength "progress bar" & Label between the two passwords. [11:59] and the progress bar has like 4 sections with colors, and then just short strings. [11:59] smartboyhw: We don't need a separate channel for every separate bit of software. [12:00] xnox, replied. [12:00] mpt: cool. Let me get you the settings. [12:00] xnox, no, I haven't seen that one [12:01] mpt: encryption settings that is. Yet, it's not sent yet. [12:01] I meant the Gnome Accounts dialog [12:07] oh. ok. [12:10] mpt: https://picasaweb.google.com/lh/photo/WNInNZFmc08H2xuZ4DJ5pcjnfn85PTV3DlClUSiESgU?feat=directlink [12:10] mpt: it has much stronger, e.g. stuff that ubiquity considers "strong password" might be just "fair password" in the accounts dialog. [12:42] xnox, wow, that's hideous. [12:43] mpt: is that your official review of Quantal? =) [12:43] https://bugs.launchpad.net/ubuntu/+source/debian-installer/+bug/1040627 [12:43] Ubuntu bug 1040627 in debian-installer "DHCP autoconfiguration fails on servers since 20120822 image" [Undecided,New] [12:43] One password field shorter than the other two, an outline that (in total) is thicker than the actual meter color, and a button disguised as a link. [12:43] Oh, and missing colons for all those labels. [12:43] mpt: it might be our theming issue (probably not) [12:49] ev: will you review geonames merge and/or deploy it? unless deployment goes via RT.... [12:49] xnox: a pols, I haven't gotten to that review yet today [12:49] deployment goes via RT [12:50] ev: ok. cool. =) [12:50] brendand: I'll see if I can reproduce it in kvm [13:13] ev: i just ponder if it is relevant to deploy it soon, e.g. for 12.04.1 release ;-) =) [13:13] xnox: I suspect not. It's been this way for quite a while [13:13] and 12.04.1 is being released today [13:13] it's already too late [13:13] fair enough =))))) [13:36] brendand: Hm. Works in KVM, at least. [13:36] did you guys have time to look at bug 1040494? [13:36] Launchpad bug 1040494 in update-manager "Unable to perform cdromupgrade without network from Lucid to Precise 12.04.1 AMD64 : E:Could not perform immediate configuration on 'python-minimal'" [Critical,New] https://launchpad.net/bugs/1040494 [13:37] it's impacting 12.04.1 [13:37] Well, it's not really anything to do with the installer - you want somebody with time to dig into upgrade bits [13:37] cjwatson, yes - i can't reproduce anywhere else. if i try mondays image though, it works [13:37] cjwatson: in which channel? [13:37] Not sure, sorry. You could ask on -release [13:37] cjwatson: ack [13:37] gema_, talk to mvo if he's around [13:38] mvo is mostly consumer apps these days. [13:38] cjwatson, you're right [13:38] Though in former days that would certainly have been the right answer. [13:38] cjwatson, i wonder who's looking after update-manager now? [13:39] Foundations in general [13:39] At some point I think we hope to get a bit more headcount to dedicate more specifically to it [13:44] brendand: Ah, I see it on real hardware - my guess would be regression due to biosdevname [13:44] Sigh [13:44] oh phew [13:44] cjwatson, sorry. i'm just relieved it's not something stupid in our lab setup [13:48] Yeah, you'd just only have seen it on some machines because biosdevname only does anything for some systems [13:50] brendand: Do you by any chance happen to have old syslogs from installs on the same machine lying around? It'd be slightly helpful to compare with what I have here [13:51] cjwatson, most recent image i can use is alpha2 [13:51] That's fine [13:51] cjwatson, ok i will get one [13:51] Much appreciated, thanks [13:52] Wait, I'm an idiot, the ethernet cable isn't plugged in here [13:52] :) [13:52] Works better if you etc. [13:53] Let me try again ... === cyphermox_ is now known as cyphermox [13:56] brendand: OK, false alarm - it works fine here :-( [13:56] With biosdevname renaming the interface [13:56] oh noez [13:56] So I think I'm going to need some way to get at your logs - do you have any kind of physical access or is this all remote? [13:56] yes there is kvm access [13:57] Hm. Do you have any way to get a USB stick plugged in? [13:57] Failing that, if you can reproduce the failure, Alt-F2, nano -v /var/log/syslog, and successively screenshot the log? [13:58] cjwatson, see Remove KVM section here: https://wiki.canonical.com/InformationInfrastructure/IS/ServerEnablement [13:58] Assuming Alt-F2 works [13:58] cjwatson, Remote KVM that is [13:58] Yeah, um, in theory I think I have that set up but I'm leaving for holiday in ~3/4 hours [13:59] And the proxy stuff takes me an hour to figure out every time I do it :) [13:59] (Which is about once a year) [14:01] cjwatson, let me see how fast the usb stick option can be achieved [14:01] I dunno, let me see if foxyproxy is less painful than it used to be [14:06] brendand: OK, what machine am I looking at here? [14:07] If this thing will condescend to unhang my browser [14:08] cjwatson, it's called wildpeach [14:08] cjwatson, i just need to release it [14:09] I'm pretty sure it's hung my browser [14:09] The Raritan window has had a spinner going for several minutes now [14:09] And java is using >95% CPU [14:09] Which I realise is java's natural state, but [14:10] So if you might be able to take a screenshot of the last few pages of syslog, that might well be faster than whatever undefined thing I need to do to fix this [14:12] Yeah, killed my browser [14:22] cjwatson, ok. this is the end of the log: http://ubuntuone.com/0xi8O4BZ84Dqx9D6xloKEM [14:22] cjwatson, let me know if i need to keep going further back [14:22] I think it would be better to do this without DEBCONF_DEBUG=developer [14:23] That introduces a ton of noise which is useful for some bugs but in this case I suspect it won't be [14:23] cjwatson, how am i going to disable that? [14:23] As it is you'd have to go back quite a bit before you found something useful [14:23] You must be *enabling* it [14:23] It's on the kernel command line [14:23] cjwatson, ok. maybe i need to look in the preseed [14:23] And it's not on by default [14:23] It's not preseedable - boot parameter only [14:24] Try your pxelinux config [14:26] yeah DEBCONF_DEBUG=5 [14:27] Right, more or less equivalent. Drop that for the moment? [14:27] cjwatson, we may be about to get the log manually. if not i'll rerun without DEBCONF_DEBUG [14:27] cjwatson, or with a different number? [14:27] Just without [14:27] Then I guess I'll probably need the last half-dozen screens or so, based on that size [14:57] cjwatson, can you get this tar file? http://ubuntuone.com/0rKDRNATcizUzg7pUiDpD4 [14:58] cjwatson, it's everything that's not preceeded with 'kernel:' [15:00] Yeah [15:00] Hmm, the interface is basically completely busted [15:00] What does 'ip link show' say? (I just need the interface names, not all the other gubbins) [15:01] Also what sort of general model of machine is this? [15:01] but it's not physically busted, because the previous days image works [15:01] and alpha2 definitely works [15:01] Right, busted at the software level [15:02] cjwatson, it's an ibm server with onboard intel ethernet [15:02] This is the first day with biosdevname turned on, so it's not totally unexpected [15:02] cjwatson, i've tried on several servers [15:02] some with broadcom, some with intel [15:02] What interfaces does 'ip link show' list? [15:04] cjwatson, let me run it [15:05] usb, p5p1, p7p1 [15:05] and lo [15:05] but i know what that is [15:05] usb0 to be exact [15:06] so my guess now is i need to change our preseed to work on p5p1 [15:08] Oh, eth0 is preseeded somewhere here? [15:08] Can I see the preseed file? [15:09] The biosdevname scheme is em for embedded NICs and pp for PCI slots. [15:10] You can also put biosdevname=0 on the kernel command line to disable this; but that will make your tests less like out-of-the-box installs. [15:11] ok. we are going to have to change tons of tests :( our fault really [15:12] It's possible you don't need to hardcode the interface at all [15:13] As of 11.10, d-i checks for BOOTIF= on the command line, which pxelinux passes if you put 'IPAPPEND 2' in your pxelinux configuration [15:13] So you can use that to tell the installer to use whichever interface you PXE-booted from [15:13] I should send mail to u-d about all this though [15:17] cjwatson, we haven't seen any problems with the desktop installer. is this not on in ubiquity? [15:20] cjwatson, how much longer until holiday time? [15:21] cjwatson, i'm just getting these file to you [15:23] It's not on in ubiquity [15:23] About two and a bit hours [15:24] ev: filed RT with links to deployment instructions. I wonder if it will raise eyebrows 1.5 years since last deployment ;-) [15:25] brendand: I posted advice to ubuntu-devel@ and a link to the bug [15:26] xnox: if you don't get a response on it, ask slangasek to raise the priority [15:26] cjwatson, http://ubuntuone.com/6bBypXrSGmHI64PB0RcS4I (preseed) [15:26] ev: ok. [15:27] cjwatson, http://ubuntuone.com/029p5d8qIAdQrSg3eGgvFz (pxelinux.cfg) [15:27] cjwatson, looks like we hardcoded eth0 in both places - hurray [15:28] brendand: Right. My advice to ubuntu-devel should be good then, although you'll need to adjust preseed/late_command too. Either use the biosdevname-generated name there, or use something like interface="$(debconf-get netcfg/choose_interface)"; interface="${interface%%:*}" to fetch the selected interface name [15:29] Aside from that, "IPAPPEND 2" is probably the preferable approach since it involves least configuration. [15:30] cjwatson, whatever is the most future proof is my preference [15:31] Right. [15:33] cjwatson, i think the IPAPPEND thing sounds like what we want [15:36] Yeah, I think it's what most pxelinux users want [15:37] The options in that mail are basically in order of preference [15:37] cjwatson, so *will* this change in ubiquity eventually? [15:39] I'm not sure [15:39] It has rather less obvious benefit on desktops to justify the disruption [15:39] OTOH there is benefit in similar behaviour [15:39] So I'm not sure :) [16:31] xnox: My diff between ubiquity 2.11.{22,23} shows gui/gtk/centrefix.{py,ui}, which I think indicate that your local copy is unclean [16:31] cjwatson: woops. [16:31] cjwatson: yeah that was just a quick local prototype debugging py-gi gtk [16:31] xnox: I'll go ahead with the upload anyway, since those files are unreferenced [16:32] true. [16:32] But thought you might like to know [16:32] sorry about that, my bet. [16:32] Ah.... it's not in the branch, but it is in the tarballs [16:32] Right [16:33] * xnox dpkg native should only take stuff committed into VCS and ignore the rest [16:33] similar to setuptools manifest === jibel__ is now known as jibel [19:58] Trying to budget lines of code, similar to launchpad [19:58] 3 files changed, 957 insertions(+), 989 deletions(-) [19:59] this _adds_ crypto UI to manual partitioning with aid from refactoring. [19:59] less is more =)