[00:02] 17:41:47 < cjwatson> evand: something like http://paste.ubuntu.com/60697/ maybe? [00:02] kirkland: ^ apply that to /lib/partman/fstab.d/hostboot [00:02] and see if it still hangs [00:03] before you hit partman in d-i, that is [00:03] gotta run out again, unfortunately. Back in a little bit. [00:41] evand: that pastebin patch didn't work as is [00:41] evand: i'm looking at it a bit more [00:46] cjwatson: ^ [00:47] how did it fail? [00:47] it certainly shouldn't have been in the same way [00:47] although I can believe that it is busted somehow, it was totally untested [00:47] cjwatson: i'm reading the patch, and I don't quite get it... [00:48] cjwatson: so this "loop" file shows up [00:49] yes, that indicates that the device is a loop-mount [00:49] it's set by autopartition-loop [00:50] cjwatson: hmm, okay, so the situation i'm hitting is: /boot on a raid1 of 2 disks, and / on a raid5 of 3 disks [00:51] cjwatson: that's sort of my setup on a few servers, seems reasonable [00:51] are you actually using wubi? [00:51] cjwatson: when doing so, installer "hangs", with several instances of 07root_on_raid [00:52] cjwatson: looking at 07root_on_raid's code, running "ext3" seems to hang [00:52] cjwatson: i've never run wubi in my life [00:52] i'm wonder if we've gotten our bugs crossed........ [00:53] cjwatson: i got the impression from evand and yourself that hostboot farkage might be at the root of this problem [00:55] if you get a hang on open_infifo, it's because something fucked up the parted_server communication [00:55] it could be basically anything that ran just before it [00:56] it's like a malloc failure, manifesting after the actual cause [00:56] we need to see the partman log again from after you applied that patch [00:56] syslog wouldn't hurt either, just in case some shell script died in the middle [00:57] cjwatson: http://pastebin.com/f7500f9d6 [00:57] cjwatson: http://pastebin.com/f70d3af56 [00:58] cjwatson: i inadvertentantly left off DEBCONF_DEBUG=developer [00:58] :-/ [01:01] http://pastebin.com/f70d3af56 breaks. try paste.ubuntu.com please? [01:01] "Fatal error: Allowed memory size of 8388608 bytes exhausted (tried to allocate 1154191 bytes) in /home/pastebin/lib/geshi/geshi.php on line 2474" [01:02] cjwatson: shite, okay [01:03] cjwatson: http://pastebin.ubuntu.com/60747/ [01:03] cjwatson: http://pastebin.ubuntu.com/60748/ [01:16] kirkland: hmm, still hostboot - you sure you applied that patch? [01:17] kirkland: and can you see what's happening in ps output? [01:17] Have there been any recent installation reports with grub failing to install ? [01:18] (alternate i386) [01:18] kirkland: I'm fading fast, going to have to come back to this in the morning [01:45] cjwatson: sorry, family obligations here on my end [01:45] cjwatson: let's revisit tomorrow [01:47] cjwatson: the hostboot: http://pastebin.ubuntu.com/60765/ [01:47] cjwatson: ps -ef > http://pastebin.ubuntu.com/60766/ [05:24] It appears I can't run partman under Ubiquity with --automatic and debugging, since it dies with an AssertionError [05:25] StevenK: it should work fine as that's the debug-ubiquity kernel command line argument. What's the assertion error? [05:26] assert self.extra_choice is not None [05:26] ubiquity/components/partman.py:422 [05:26] * evand raises an eyebrow [05:27] and this only happens with both --debug and --automatic enabled? [05:28] if that's the case I suspect there is some very bad interaction in debconf land. [05:28] So it seems. I can copy the debug log out of the way and then try with just --automatic if you wish [05:28] you shouldn't need to move the debug log, but give it a try with just automatic (on a fresh debconf db) [05:29] * StevenK has forgotten how to wipe debconf's database [05:30] StevenK, You don't want to wipe completely, or you lose the preseeding. Maybe reboot? [05:31] Then I lose the debug log [05:31] indeed, I don't know of any way to revert to the intial state short of keeping a copy of /var/cache/debconf/config.dat handy. [05:32] * StevenK scp's off the log [05:33] StevenK, You probably want /var/log/partman and /var/log/syslog to go with that. [05:51] evand: Right, it doesn't break with 'ubiquity --automatic' but it does ask the question [05:53] StevenK, Same initial setup, partman breaks with -d and --automatic isn't skipping a question (but you can't get to that question with debug)? [05:53] persia: Suprisingly, it did skip the username question [05:54] Which question is it asking when you say "it does ask the question"? [05:54] * StevenK reboots, and tries it from the icon [05:54] persia: How to partition the hard disk [05:54] hrm, perhaps it's a red herring. Do the logs show anything interesting? Perhaps partman bailed out? [05:54] evand: I doubt it bailed, it says it ready to install [05:55] hrm [05:55] sda1 -> ext3, sda5 -> swap [05:55] Looks like guided to me [05:57] StevenK, Did it perhaps bail the time it didn't work? [05:57] evand: Shall I quit, capture logs, and then try with --debug --automatic again? [05:59] sure, can I see the logs? syslog, debug and partman, that is. [06:01] I have the earlier debug ones, too [06:06] ok [06:09] evand: http://people.ubuntu.com/~stevenk/ubi/ [06:09] * StevenK stares at his laptop. [06:15] nothing is jumping out at me, but I'm still looking over the logs [06:16] There's a *big* difference in the partman logs there, although the initial conditions look similar. Did partman give you a useful error of any sort with --debug --automatic? [06:17] An assertion error, which is in the logs [06:17] evand: In the --debug --automatic logs, I hit the "I don't care, do it again" button, so the error is probably duplicated [06:18] ah, ok [06:18] can you pastebin your preseed file? [06:20] evand: http://paste.ubuntu.com/60852/ [06:21] Oct 22 04:13:31 debconf (filter): <-- INPUT critical partman-auto/init_automatically_partition [06:21] debconf (developer): <-- METAGET partman-auto/init_automatically_partition Type [06:21] debconf (developer): --> 1 select [06:21] debconf (developer): <-- FGET partman-auto/init_automatically_partition seen [06:21] debconf (developer): --> 0 true [06:21] Oct 22 04:13:31 debconf (filter): --> 30 question skipped [06:21] I wonder why that's happening again, given that Colin fixed this (bug 182004) [06:21] Launchpad bug 182004 in ubiquity "partitioner fails if partially preseeded due to seen flag madness" [Medium,Fix released] https://launchpad.net/bugs/182004 [06:22] might also be a red herring [06:22] * evand digs further [06:34] and curiously I cannot reproduce it here, given your preseed file [06:34] hrm [07:24] I'll pick back up with this in the morning. === cjwatson_ is now known as cjwatson [10:33] guys you know on encrypted lvm install where the system boots and asks you for a password for the drive... Is there anyway to make that font clearer? It's quite fuzzy especially on the s's [10:57] superm1 I cannot replicate the crash on vista, will try tonight using xp in vm [10:58] davmor2 do you happen to have xp? [10:59] testing across the board at the moment xivulon although I will be doing some wubi testing in about an hour [11:01] cool, the issue reported by superm1 is that when you use wubi in xp attached to a network but without internet connection, then it crashes at the reboot screen [11:04] xivulon: on what cd version [11:04] I guess using latest ubuntu desktop daily [11:05] okay np's I'll have a look shortly :) [11:05] but in theory it should not matter, since the issue is in windows, and wubi is still rev 513 [11:16] davmor2: font> looks like it's example-theme/helvB10.bdf in the usplash package [11:17] (or possibly bogl/helvB10.bdf, but they're the same) [11:18] cjwatson: if you get chance have a look if you can. The s's are worse hit [11:18] not for intrepid I think [11:18] but do file a bug [11:19] kirkland: 227848 has a release-notes task, but ISTR you were working on fixing this? [11:19] I thought it wouldn't be in for intrepid. I'll bug it after [11:19] (for intrepid) [12:14] xivulon: seems okay here on kubuntu. I'll try with ubuntu in a bit though incase it is specific [12:17] xivulon: is his copy of xp up-to-date? [12:21] davmor2: not sure, superm1 ^ [12:22] you have to be connected to the router but the router must not be connected to the web (or at least so I understood) [12:27] xivulon: ah hangon then I was connected to my switch but not router [12:30] I'll try it like that on the next one shouldn't be too long [12:37] is there some way to check the syntax of a debian preseed file? [12:37] it just says "no!" withtout any hints [12:37] debconf-set-selections -c [12:38] but you may not actually have a syntactic problem [12:40] cjwatson: Thanks for you pointer about /var/{run,lock} yesterday. http://bazaar.launchpad.net/~ubuntu-virt/vmbuilder/trunk/revision/258 is clearly the better solution. [12:49] soren: looks good [13:13] xivulon: happens on xp and vista it hangs on reboot not of xp/vista but of ubutnu [13:13] ubuntu even === davmor2 is now known as davmor2_lunch === davmor2_lunch is now known as davmor2 [14:07] davmor2 that is then a different issue [14:08] it usually depends on /etc/init.d/sendsigs or /etc/init.d/umountfs [14:10] you have no hangs within windows though, correct? [14:15] hmm sendsigs and umountfs look good to me [14:16] xivulon: no hang in windows no [14:26] davmor2, it should be possible to narrow down the issue by going to a terminal and running the scripts in /etc/rc6.d [14:27] xivulon: we're bhind with testing but I might be able to tomorrow [14:28] ok, will have a go myself tonight, do you see anything on script? I/O errors? [14:28] s/script/screen [14:34] xivulon: nothing just a flashing underscore [14:35] was this in vm or real hardware? [14:36] real [14:37] thx will have a look [15:27] davmor2, xivulon this is xp sp3 that i did it on [15:39] superm1 as mentioned on vista worked will, will give it another go in xp/vm, did you use a vm by the way? [15:39] xivulon, yeah that's the only way i have to test it is vm [15:41] cjwatson: well, if it just stops, how can I know what is the problem? [15:41] if its a missing package, something syntax error, or something else [15:56] holst: you need to look in the syslog [15:56] holst: put it on a pastebin and I'll be happy to have a look at it [15:58] cjwatson: hey, i'm working on a writeup on encrypted private for james_w ... I've not installed from a desktop intrepid iso in a long time ... did setting up an encrypted private directory ever make it into the desktop installer? [15:59] kirkland: I swear I've answered this once a day for the last week :) [15:59] kirkland: no, it didn't [15:59] cjwatson: :-) not from me! [16:11] cjwatson: ah, syslog. I will try that [18:32] cjwatson: we're getting a hang at 5% on installing ubuntu dvd via ubiquity [18:33] IIRC, that's formatting. If you run top and sort by CPU, do you see mkfs chugging along? [18:33] How long has it been hanging for? [18:34] long enough for the screensaver to kick in so minimum of 10minutes [18:35] can you pastebin `ps auxf`? [18:36] evand: I'll re run in a second I was running cd check to make sure it was alright [18:36] ok [18:37] evand: re-running now no errors found [18:43] evand: heno has just said that his vm install just started up again after 15-20 minutes or so [18:44] started up again? Can you elaborate on what you mean by that [18:44] install continued [18:45] ah [18:45] evand: right I'm at 5% again and my hd is going like mad at the minute so I'll give it till the hd stops and and then paste the info is that okay? [18:47] sure thing [18:49] evand: http://paste.ubuntu.com/61106/ [18:53] hrm [18:54] it's just sat there no hd light going no dvd light going [18:55] evand: I'm off for my lugmeeting soon but I'll leave it to run and see if it has installed when I get back [18:55] ok [18:55] davmor2: when you're about to go can you run strace -p 14115 -o ubiquity.log , let it run for a bit and then pastebin [18:56] might be a bit large for pastebin [18:57] I'll tag it to http://www.davmor2.co.uk/...... Will 4 minutes be enough? [18:58] oh apparently operation is no permited [19:00] got it [19:00] need to sudo it [19:00] sudo strace [19:00] less than a minute will be fine [19:00] indeed [19:01] err about a minute, that is [19:02] hrm it seems to be empty should that be the case? [19:03] no [19:04] is this still the same install as when you pastebin'ed? [19:04] yeap at the same place still too [19:05] double check that 14115 is the process ID of python /usr/share/ubiquity/install.py [19:05] 14115 /usr/bin/python /usr/share/ubiquity/install.py [19:06] re-running strace but I really need to go in a minute [19:06] indeed [19:06] don't let me keep you [19:06] we can continue this at some other time [19:07] oh it just shot up to 15% [19:09] evand: http://www.davmor2.co.uk/ubiquity.log [19:09] got to go now bye [19:10] cheers [19:10] 32% and still going === cjwatson_ is now known as cjwatson [21:23] Why does "do-release-upgrade -d" want 88.1M free space on /boot ? [21:23] I only have 12M on my /boot now (8.04)! [21:23] you want mvo for that, who isn't here; try #ubuntu-devel [21:24] OK... [21:24] either clean up some old kernels or have a less inconveniently small /boot, though :-) [21:25] Do you think 100M /boot is to small!? [21:25] if you only have 12M left, you need the first option (clean up some old kernels) [21:25] kernels consume about 12M apiece [21:25] (when you count initrd etc.) [21:27] I meant that my files on /boot (8.04) take 12M [21:27] oh, I see [21:27] its calculation does sound wrong then [21:28] Is mvo a user or does it mean anything else? [21:33] ClaesBas: mvo is Michael Vogt, the author of do-release-upgrade [21:33] ClaesBas: alternatively, just file a bug: https://bugs.launchpad.net/ubuntu/+source/update-manager/+filebug [21:34] The hanging DVD situation appears to be the fault of generate_blacklist(), which makes sense as that's one area where we'd see a difference between the CDs and DVDs. [21:34] oh, so it's just slow? that would figure [21:34] is it at the dependency calculation stage, or the stage where it does dpkg -L everything? [21:34] indeed, it seems we set out to reduce installation time and created a monster in the process :) [21:35] not sure, I'll probe further [21:35] what we have done at least is reduce the required disk space [21:35] ah, very true [21:35] we could back it all out; I'm insanely cautious about any optimisation on that code right now since we've already created a release notes entry requirement ... [21:36] has anyone been paying attention to ensure that it isn't removing anything it shouldn't remove any more? [21:37] I haven't seen any new complaints, but I haven't run a test to make sure that two installs with and without the blacklist generates the same files. [21:37] I can do that tonight though [21:39] you could also check the logged files [21:41] actually, indeed [21:41] watching that now [21:42] kirkland: can you confirm whether bug 227848 should be release-noted or fixed for intrepid? In the former case, could you provide some sample text? [21:42] Launchpad bug 227848 in open-iscsi "boot order wrong for iscsi" [Critical,In progress] https://launchpad.net/bugs/227848 [21:43] cjwatson: release noted; we don't have time/resources for a proper fix for intrepid [21:43] cjwatson: how long is such release note text, generally? [21:44] cjwatson: 1 paragraph? 5 sentences or so? [21:44] kirkland: http://wiki.ubuntu.com/IntrepidReleaseNotes [21:44] cjwatson: thx [21:44] usually a paragraph or two is fine [21:48] cjwatson: actually, I stand corrected .... [21:49] cjwatson: i don't think we really have anything to release-note [21:49] cjwatson: we don't really have a viable workaround for intrepid [21:49] "it's broken"? :-) [21:49] cjwatson: i think the priority should be lowered from Critical to High [21:49] please do that [21:49] cjwatson: oh, just a release not to note the bug? [21:49] sometimes that is worthwhile [21:50] cjwatson: okay [21:50] not currently supported, even though it looks like it should be, or whatever [21:54] cjwatson: i'm working with EtienneG on it.... [22:22] debian-installer: cjwatson * r982 ubuntu/ (build/Makefile debian/changelog): [22:22] debian-installer: Take /usr/share/localechooser/languagelist.data.gz into account when [22:22] debian-installer: building reduced font (LP: #287668). [22:26] cjwatson: About the /boot size problem, it looks like do-release-upgrade dowloads the file intrepid.tar.gz which does some checks in DistUpgradeController.py ... That checks calculate from what it thinks is installed, NOT what is physically on the partition! [22:26] seriously, I'm not the expert here, it's probably best not to talk to me about it :) [22:26] I had to "remove" all old kernels from apt.... [22:26] err. you removed old kernels using rm rather than by removing the package? [22:27] be sure to say that in your bug report; that's unusual practice [22:27] I agree that it ought to cope with that [22:29] I'm going to try fill in some bug report about now... [22:51] evand, given a DVD was finally functional today, i grabbed it and dropped it onto a recovery partition. has a bunch of stuff changed with partman preseeding since the last dvd? http://paste.ubuntu.com/61230/ [22:52] Oddly, in the current -mid image, I am shown step 5 iff I don't run in debug mode. How could I best troubleshoot this? [22:53] along with this preseed: http://paste.ubuntu.com/61231/ [22:54] superm1: could do with syslog and partman too [22:55] I think I know what it is [22:55] * evand digs [22:56] https://lists.ubuntu.com/archives/intrepid-changes/2008-October/008520.html [22:56] superm1: preseed partman/filter_mounted to false [22:56] * evand steps out for a few [22:56] evand, okay i'll try [22:57] sorry about that. I had intended to send a notice of new preseed options the ubuntu-installer ML [22:57] I'm going to try to get in the habit of doing so with each new ubiquity release [22:58] so this is supplemental to the existing ubiquity/partman-skip-unmount boolean true then? [23:04] well that at least got it further, but it's just hanging at "1%: Creating ext3 file system for / in partition #3 of SCSI1 (0,0,0) (sda) ..." now, the python process with install.py as an argument is pegging the CPU too. hum [23:06] superm1: read up a bit, we're on it :) [23:07] evand, ah okay. need any other data points for that too then? [23:07] oh half that conversation is in #ubuntu-release [23:07] back in a bit [23:08] cjwatson: iscsi text sent your way via email [23:08] kirkland: can you put it in the bug, please? [23:09] cjwatson: yup [23:10] cjwatson: done: https://bugs.edge.launchpad.net/ubuntu/+source/open-iscsi/+bug/227848 [23:10] Launchpad bug 227848 in open-iscsi "boot order wrong for iscsi" [High,In progress] [23:10] thanks