[00:36] slangasek: ok thats going to make kees happy (sarcasm) [00:37] he's welcome to fix it ;) [00:49] zul: PIE exposes it, but it's a sourceful bug; there's no prototype for cache_path() in source/include/proto.h [00:58] zul: fix committed to the Debian svn tree; I can do an upload to unstable so you can pick it up via MoM? [00:58] slangasek: that would be great [01:17] hm... is it just me, or are the .ini files linked to cisco vpn settings in jaunty? [01:20] heh in /usr/share/mime/packages/freedesktop.org.xml [01:25] stgraber: Ping! [02:15] zul: 2:3.3.0-4 uploaded to unstable, so should be available in MoM tomorrow [02:20] slangasek: I believe I have a patch to fix kipi-plugins FTBFS with Qt4.5. Is it OK if I upload it or would you rather I hold if until after the Alpha? [02:21] ScottK: how big of a change is the upstream diff from rc1 to rc2? [02:21] Not sure. [02:21] ok [02:22] well, amd64 is already on rc2, so if you don't know of any reasons why it'd be disruptive, go for it [02:22] The patch is add one def and change 3 lines. [02:23] OK. Assuming it test builds and stuff OK, I'll upload it. [03:13] ArneGoetje, pitti: could we please not have langpack exports the Tuesday of a milestone? [03:15] slangasek, gah i didn't realize that you flipped the switch for freezing for alpha builds already. i just uploaded a small diff for libsmbios [03:16] i had a hal upload ready to go with it, but i'll hold back [03:16] superm1: Tuesday of a milestone, should be pretty predictable? :) [03:17] superm1: so this is hal depending on smbios-utils, rather than using libsmbios directly? [03:17] slangasek, the actually it's going to be hal not needing smbios-utils [03:17] slangasek, it links directly to libsmbios_c [03:18] superm1: ah - I'd be happy with that being uploaded now if it's been tested, since it also kills off an NBS [03:18] slangasek, yeah that's what i've been up to today, working out the kinks on it. [03:19] I've not removed an NBS in a while, can I push the button? [03:19] StevenK: once it's no longer a dependency of hal? :) [03:20] slangasek, however for feature parity it might still make sense to leave smbios-utils as a recommends to ubuntu-desktop (but still removing libsmbios-bin from hal's recommends) [03:20] that can be sorted out after alpha though [03:21] slangasek or StevenK: Modulo one powerpc FTBFS and one package I'm waiting for the slow archs to build, I think we're ready to remove boost (1.34). Often when I file removal bugs the archive admin in question will find a residual rdepends on some obscure arch. Any suggestions on how I can dig those up? [03:21] slangasek: you mean langpack uploads, right? [03:21] Speaking of removals... [03:21] ArneGoetje: yes [03:22] slangasek: sorry, forgot to disable the crontab entry. [03:22] ArneGoetje: ah - didn't realize there was a cronjob; should I add it to my checklist to prod you about that? :) [03:23] slangasek: would be nice, yes. ;) [03:24] superm1: libsmbios-bin was only ever pulled in by hal, not treated as a top-level component of the desktop; seeding smbios-utils ought to be discussed separately then, I think [03:24] slangasek: Did you mean to give-back d-i on ia64 as well? [03:24] StevenK: no [03:24] slangasek: So ia64 will remain busted? [03:24] slangasek, okay sounds good [03:24] slangasek: langpack uploads to jaunty happen on Tuesdays and Saturdays [03:24] Not that I care, just curious. [03:24] StevenK: it won't have an up-to-date d-i for alpha-5, yes [03:25] ScottK: there's a 'checkrdepends' script we use, which basically just runs grep-dctrl over the mirror... [03:26] OK. I'll try that. [03:26] Thanks. [03:26] I can post a copy somewhere if you like, but it's a trivial shell script (and not at this very moment, I'm on the way out to grab dinner as soon as I wrap up my current stack) [03:28] I just had an idea, probably a duplicate, I might post it to idea storm. What if we have an option in the windows cd program that adds cd-booting to the Boot.ini? [03:28] I just had someone claim ubuntu didn't work, turned out their bios wasn't set to boot from cdrom and they thought that meant ubuntu had failed because it went directly into windows again. [03:29] slangasek: I would appreciate the script when you have a moment. [03:46] Is it possible to remove multiple lines of a file using either grep or sed regular expressions? [03:47] TheMuso: grep -EV '^(line 1|line 2)$' [03:47] StevenK: Yeah I knew about the -v switch, but thanks for the rest of the syntax, much appreciated. === nhandler_ is now known as nhandler [03:56] StevenK: hrm thats not quite what I am after. I need to remove a sequence of lines, as in three lines one after naother after another. I thought grep understood \n, and I thought sed did as well, but \n to indicate newlines is not working... [03:57] TheMuso: grep -v -A 3 'line' ? [03:59] StevenK: That would work if there weren't the same strings elsewhere in the file. I am trying to strip some XML from /usr/share/gconf/defaults/05_panel-default-setup.entries to remove the fast user switch applet from the top panel, to enable logout menus for accessibility profiles. [03:59] TheMuso: Perl? [03:59] StevenK: This will be running in an initramfs/casper script with sh. [04:00] I hate you [04:00] So I don't think perl will be available, unless I chroot, which is possible. [04:00] Ummmm. sed can do it. [04:00] Just not sure my sed-foo bends that far [04:01] | sed -e '/line/,3d' ? [04:02] I'll try it. My regexp skills in general aren't that great either. :) [04:04] hrm o that just gets rid of one line. I'll have a play and see what else I may be able to come up with. [04:04] Or maybe just modify the required settings with gconf itself, i.e gconftool-2. Hrm. [04:08] TheMuso: have you tried pcregrep ?:) [04:08] savvas: Is it in a default install on aan Ubuntu live CD? [04:09] I don't think so :\ [04:14] here are some handy sed one-liners: http://sed.sourceforge.net/sed1line.txt [04:15] or awk might be easier to handle, never tried awk though [05:01] TheMuso: if still looking for a multi-line delete, way I've done in the past is with awk: awk '{ if (f==2) {f=0} else if (f==1) {f=2} else if ($0 ~/line/) {f=1} else print $0 }' [05:01] I was actually doing multi line edits based off the 1st line, so it made sense that way. ymmv.... [05:01] The first line is not appropriate here, because the string in that line exists elsewhere in the fie. [05:02] bummer [05:03] tho... you could still do with awk; pull the "previous matching line into a var", and print that out if the next line isn't what you're expecting kind of thing. ugly as sin tho.. [05:17] slangasek: I also have a fix for the konq-plugins FTBFS. May I upload that or should I hold it? [05:20] ScottK: go ahead [05:20] ScottK: and, script: http://pastebin.ubuntu.com/122673/ [05:21] Thanks. [05:33] slangasek: OK, so it turns out it's also accidentally a native package: https://launchpad.net/ubuntu/+source/konq-plugins/4:4.2.0a-0ubuntu1/ [05:33] Want me to fix that too or just leave it for now? [05:33] hem; leave it for now, I guess [05:33] OK. [05:37] hrm ok, I can set what I want via gconf. [05:37] I'll have to see if it works however, but fingers crossed. [05:37] TheMuso: Oh good. Then I don't have to cry at what you're adding to casper :-P [05:37] StevenK: Nor do I./ [05:42] good morning [06:30] StevenK: there's an NBS package with no reverse-deps if you want it ;) [06:30] I'm so there! [06:31] slangasek: Binned, thanks [06:34] slangasek: Should I be dealing with the libmysqlclient-dev rdepends? What's the situation in Debian, if you know? [06:34] dunno [06:36] Last I heard in Debian there was no appetite for making mysql 5.0 and 5.1 co-installable. [06:37] ScottK: You sleep with your laptop? [06:37] Wait, don't answer that. [06:37] Heh. No, I just procrastinate a lot, including going to bed. [06:37] ScottK: Okay, so shall I fix the ~ 26 build-deps to 5.0? [06:39] I have no idea. I've been avoiding knowing any of the details of the mysql changes. [06:44] ScottK, slangasek: I'll pick on a few source packages out of the list tomorrow and see what they're doing. If we can get away with a few syncs, I'll sort it out. [07:19] Good morning [07:19] bryce: yes, I'm following the upstream bug tracker [07:19] mornign pitti [07:20] bryce: of course that "black screen" bug hit me again right after asking you about whether I should fix it :) [07:20] hey highvoltage [07:21] asac: every time I open firefox I get a "new addons added" with 4 langpacks; known already, or want a bug? [07:22] heya pitti! [07:22] * pitti hugs Hobbsee and throws a gummybear [07:23] :D [07:23] * StevenK waves to pitti [07:41] hey juanje, hey rcmorano! [07:42] hiya man! [07:42] :] [07:42] how's life in Spain? :) [07:42] dholbach: hi! :-) Good morning [07:43] it's ok, weather starts to get perfect :D [07:43] dholbach: still cold and sleepy, but in few hours it sure will be nice [07:43] nice, you guys are lucky :-) [07:43] ready to start working [07:43] dholbach: how is up there? [07:44] 2°C, grey :) [07:44] hi, crazy holbach [07:44] dholbach: ouch.... [07:45] Mithrandir: crazy Tollef :) [07:45] 2°C is boring, all the snow starts melting. [07:45] dholbach: ummmm.... are you planning to go to the Gran Canaria Desktop Summit? is in my town... ;-) [07:46] Mithrandir: wasn't my idea [07:46] juanje: no, unfortunately not [07:46] :-/ [07:46] ouch :/ [07:50] trying to pin down a bug, but testing requires that i edit /etc/udev/rules.d/70-persistent-net.rules ...do i have to restart udev or something after this for it to go into effect? [07:55] maco: maybe with "udevadm control --reload_rules" [07:56] ok. im probably about to go offline then ;) [07:56] maco: but sometimes you have to reboot the machine to rebuild a clean set of devices [07:56] hehe [07:57] i guess [07:57] ok i guess it does want me to reboot [08:00] Keybuk: ok, the kpartx rules get stuck in 'ENV{DM_PART}=="?*", SYMLINK+="disk/by-id/$env{DM_TYPE}-$env{DM_NAME}-part$env{DM_PART}"'. that rule is matched only after running 'kpartx -a ...' manually [08:02] Keybuk: and DM_PART should come from kpartx_id, hmm.. === thegodfather is now known as fabbione [08:29] Tonio_: kdesudo is failing to install in the kubuntu livefs builds, the removal of the transition package is wrong [08:30] Tonio_: this needs to be fixed so we can build ISOs for alpha-5 [08:39] bryce (or bryce_ ) I'v confirmed in bug #325394 that upstream patch from Alex Deucher (freedesktop) is fixing the bug. Please fixe the package when you can :) [08:39] Launchpad bug 325394 in xserver-xorg-driver-ati "[RV770 HD 4850] ATI Radeon HD 4850 not supported - white screen and system freeze" [Unknown,Fix released] https://launchpad.net/bugs/325394 [08:39] slangasek: how disruptive would uploads be today? ie should I respect the freeze or not ;-) [08:40] seb128: very disruptive [08:40] ok, the oem patches for notebook will wait then [08:40] slangasek: thanks [08:41] slangasek: we have a direct dep on the transitional package ? [08:41] slangasek: that's bad... [08:41] Tonio_: no, I mean the removal of the transitional package was done wrong [08:41] slangasek: ah ? [08:41] Tonio_: http://people.ubuntu.com/~ubuntu-archive/livefs-build-logs/jaunty/kubuntu/20090225/livecd-20090225-amd64.out [08:42] Tonio_: also, I vaguely mean "wrong" as in "shouldn't have been uploaded two days before a milestone release"... [08:43] slangasek: ouch... indeed, the package had crap postinst file and it looks like I missed the correct diversion... [08:43] fixing... [08:43] slangasek: and yeah, sorry for the upload that late... [08:43] s/kdesudo-kde4/kdesudo/ in the preinst should fix that part [08:44] but I haven't thought through what actually happens with this on package upgrades given that kdesudo-kde4 owned the diversion previously [08:44] slangasek: there is a postrm that should drop the diversion and install the new one [08:45] Tonio_: the confidence with which you assert this makes me suspect you don't have much experience with diversions ;-) [08:46] slangasek: :) no I mean, I upgraded many machines and it still works here, so I suspect that part is fixed [08:46] slangasek: I perfectly know that diversions can have weird artefacts at some points [08:46] Tonio_: did you manually uninstall the transitional package? [08:46] Tonio_: An upgrade isn't the same. [08:47] Tonio_: I think the new kdesudo needs to Conflict: kdesudo-kde4, so that the transitional package (and its diversion) is removed before kdesudo is unpacked [08:47] otherwise you have two packages trying to divert the same file at the same time [08:47] slangasek: nope, but probably it wasn't installed for me [08:47] slangasek: well maybe we should just reintroduce the transitional package and fix this after the milestone release, no ? [08:48] * Mithrandir ponders upgrading to jaunty. [08:48] Tonio_: actually... yes, because ubiquity-frontend-kde depends on kdesudo-kde4 [08:48] Tonio_: so best to revert for now [08:48] slangasek: okay, so I'll switch back and fix this after the milestone release [08:48] Tonio_: ok, thanks [08:49] slangasek: sorry for the issue... [08:49] superm1: great work on hal+smbios! [08:52] In intrepid lsof sometimes displayed information like "(path dev=253,0, inode=475136061)" at the end of the lines - my interpretation was that it was stuff in memory, that is different on the disk now - I found this very useful after doing upgrades, because I knew which services/programs I had to restart to benefit from changes that were done - in Jaunty this information seems to be gone - can anybody tell me why or where it went? :-) [08:54] dholbach: I don't remember ever seeing anything like that; but grepping for 'DEL' should give you that information [08:54] slangasek: FANTASTICO [08:54] gracias :) [08:56] * DktrKranz hugs dholbach for his "italiano perfetto" :) [08:56] DktrKranz: it was spanish! :) [08:56] * dholbach hugs DktrKranz back :) [08:57] slangasek: uploading kdesudo now [08:57] thanks [08:58] pochu: it was not french ;) === mvo__ is now known as mvo [08:59] slangasek: are you okay with uploading more intrusive bug fixes to packages not on the CD? (sl-modem in my case) [08:59] pitti: in general, yes; hopefully nothing that's going to take buildd time away from the langpacks, though [09:03] slangasek: I can easily lower its build score [09:22] hi [09:22] i am using 8.10 on GPU i915 Intel. i got the problem when run glxgears, like "Failed to initialized GEM. Falling back to classic". Now OpenGL doesnt run anymore. Anybody knows this problem? [09:22] it didnt happen with me before, so perhaps the recent update of xorg is the culprit?? [09:24] please soebody help? [09:25] have you already checked launchpad for a bug? [09:27] hi [09:27] zinzin: ^ [09:28] maco: is there anything about that? [09:29] maco: how can i find the information on the "launchpad"? sorry i am not aware of that [09:29] er...i was asking if you'd checked launchpad already and were trying to find a dev who knew about not-on-the-bugtracker (or reported upstream) issues [09:29] http://bugs.launchpad.net/ubuntu/+bugs <-- the home of ubuntu bugs [09:30] zinzin: this question should have gone into #ubuntu first AFAIK [09:30] or since you're asking about intel graphics, more specifically http://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bugs is likely to narrow it down [09:32] zinzin: scizzo- is right about #ubuntu though, or #ubuntu-bugs [09:34] ok, i will try with ubuntu-bugs first. thanks [09:41] bryce: what's missing for xorg-ctrl-alt-backspace to be set to implemented; some documentation to write? [09:41] pitti: yeah [09:41] bryce: and x-testing-infrastructure ? [09:43] pitti: mostly I just need to finish up the historical driver page for that === zaafouri is now known as zaafouri` [10:00] pitti: i guess we need a bug. was that first triggered by a langpack update? [10:02] pitti: do you close ffox properly sometimes? or just forceful exit through log-out/shutdown? [10:03] asac: has there been some change that says its ok for networkmanager to try to control an interface that is configured in /etc/network/interfaces? [10:03] maco: no. only if you run the plugin in managed mode [10:04] hrm? how would i find out if it's doing that? [10:04] i just started using the interfaces file last week when i had trouble with wpa and wanted to do it manually...have since discovered that nm is still trying to manage even when ive got interfaces setup === mt is now known as mat_t [10:05] maco: /etc/NetworkManager/nm-system-settings.conf [10:06] look if its managed=true [10:06] asac: main plugins=ifupdown,keyfile ; ifupdown managed=false [10:06] asac: most of the time just logout (i. e kill) [10:07] asac: yes, very probably triggered by langpack update === Lure_ is now known as Lure [10:09] maco: have you rebooted in between? [10:10] asac: yes, many times [10:10] it also turns out that if i let NM connect, itll rewrite my /etc/hosts into broken-ness and then i cant launch anything in X, so ive been killing NM before logging in, then using ifup manually [10:12] maco: is nm-system-settings process running at all? [10:12] nope [10:12] thats the problem then [10:12] but i did a sudo service NetworkManager stop already [10:13] that doesnt matter [10:13] because if i let it run, i cant use graphical apps [10:13] yes. thats understood [10:13] maco: if nm-system-settings isnt running thats the problem [10:14] if you think it has to do with stopping NM check before you kill it after reboot [10:15] ok [10:16] will try that then [10:16] thanks [10:16] maco: what do you have in /var/crash? [10:16] probably a LOT [10:16] yeah, a lot [10:17] nm-system-settings is in there, if that's what you're asking [10:17] maco: yes. [10:17] maco: please submit that crash file [10:18] through apport [10:18] ok [10:18] does apport succeed at submitting now? [10:19] asac: confirmed bug 289469 and added theory/screenshot [10:19] Launchpad bug 289469 in firefox "Firefox opens "new add-ons have been installed" dialog window at startup" [Undecided,Confirmed] https://launchpad.net/bugs/289469 [10:20] it wont let me because i havent updated python [10:20] * asac looks for theory ;) [10:21] pitti: there are extensions* files in your profile ... is the timestamp up to date? [10:24] asac: https://bugs.edge.launchpad.net/ubuntu/+source/firefox/+bug/289469/comments/3 [10:24] Ubuntu bug 289469 in firefox "Firefox opens "new add-ons have been installed" dialog window at startup" [Undecided,Confirmed] [10:25] pitti: also check whether the langpacks that are shown to be new are in the extensions.ini file [10:26] asac: they aren't really new, they just got updated; but they all are [10:27] asac: bug updated [10:27] ok thanks [10:28] asac: I guess eventually it's seb128's fault :) (gnome-session) [10:29] pitti: heh ;) ... i am not sure. i have to check what is supposed to happen when on global extensions update [10:29] asac: trying to blame yet another bug on GNOME I see ;-) [10:30] * pitti hugs seb128 [10:30] * pitti kicks gnome-"Terminator ZAP!"-session [10:30] seb128: hey, i even tried to be politically ;) [10:31] * seb128 hugs pitti === shiyee is now known as abemad [10:50] pitti: do we currently have problems with langpack-o-matic? [10:50] pitti: i wanted to see what happens when installing language-pack-fi [10:50] ls -l /usr/lib/firefox-addons/extensions/langpack-fi\@firefox-3.0.ubuntu.com/ [10:50] total 0 [10:50] also there is nothing about langpack-fi in the -base package [10:51] interesting that Mirv didnt complain yet ;) [10:51] asac: there is a bug about it [10:53] slangasek: ubiquity-frontend-kde Depends: kdesudo | kdesudo-kde4, FWIW, not just kdesudo-kde4 [10:55] TheMuso: you probably want to look up address ranges in sed; for instance you can do sed -e '/line/,+2d' which deletes anything matching /line/ and the two lines after it [10:55] asac: I filed the bug already :) === thunderstruck is now known as gnomefreak [10:58] Mirv: :-D [11:02] cjwatson, did you remove the keymaps as well for ixp4xx ? [11:02] (the changelog entry only talks about console-setup) [11:05] ah, you did :) [11:06] yes [11:06] I meant console-setup the whole pile, not console-setup-udeb the package [11:06] btw, i still see the uuid prob here [11:06] could it be that partman doesnt disconnect the device as soon as swap is in use ? [11:07] so that udev cant pick up the new uuids [11:07] thats my only explanation for that behavior ... [11:07] Audio is borked on daily iso that is meant to be a5 candidate playback shows Null output (PulseAudio Mixer) [11:09] seems partman calls swapon first and then goes on with formatting [11:09] Who knows about usplash here? === mpt__ is now known as mpt [11:09] ogra: what does "disconnect the device" mean? [11:10] ogra: vol_id goes and physically looks in the contents of the block device - it doesn't matter whether it's mounted or not [11:11] cjwatson, well, i have definately another failed install here where the uuid's are the old ones [11:12] ogra: I'm not disputing that, merely disputing your suggestion for a cause :-) [11:12] i was expecting something like blockdev rereadpt [11:12] no, that gets the kernel to reread its in-memory copy of the partition table, but vol_id doesn't look at that [11:12] though i'm not sure udev recats on that and changes the uuids [11:12] and FWIW libparted does the equivalent of that internally *anyway* [11:13] hmm [11:13] *reacts [11:13] vol_id does something much more like open("/dev/sda1", O_RDONLY) and then direct reads [11:13] it does not depend on the udev database, even [11:14] right, which means vol_id has the new uuid info, but udev doesnt get triggered to change the device node [11:14] sadly lowmem mode gets in my way a bit, having only half a log isnt really helping :( [11:14] so at which point do you observe udev having the wrong uuids? [11:15] just after partitioning, or after the first reboot, or when? [11:15] well, i get the error only if update-initramfs runs ... [11:15] (I assume that by "udev having the wrong uuids" you mean something like "symlinks in /dev/disk/by-uuid/ are wrong" [11:15] ) [11:15] but i.e. the fstab in /target has the new uuids [11:15] right [11:16] exactly what is going wrong? [11:16] the symlinks show uuids that arent existent ... given that i use the same USB disk all the time they are very likely from my last install [11:16] ~ # ls -lh /dev/disk/by-uuid/ [11:16] lrwxrwxrwx 1 root root 10 Jan 1 00:08 4069d9b2-b663-4633-815b-c09f32761279 -> ../../sda5 [11:16] lrwxrwxrwx 1 root root 10 Jan 1 00:08 794f4f3f-e988-45dc-83d1-f0f0a201914b -> ../../sda1 [11:16] lrwxrwxrwx 1 root root 10 Jan 1 00:08 7e895cfc-4d4c-4104-a9be-f9c06641e63b -> ../../sda2 [11:17] thats what i have right now in d-i [11:17] vol_id -u /dev/sda1 [11:17] ~ # grep boot /target/etc/fstab [11:17] # /boot was on /dev/sda1 during installation [11:17] UUID=7e1ff899-da3c-4ec8-9206-2219331bbdd9 /boot ext2 defaults 0 2 [11:17] ~ # vol_id -u /dev/sda1 [11:17] 7e1ff899-da3c-4ec8-9206-2219331bbdd9 [11:18] ok [11:18] vol_id, fstab and the like are right, udev isnt [11:19] update-initramfs looks for the new uuid and doesnt find it [11:19] thast the point where the install ends [11:20] so I guess the thing that would be interesting would be to get 'udevadm monitor' output across partitioning [11:20] this is more Scott's area than mine, of course, but anyway [11:20] hmm, tricky [11:20] you don't have a second console? [11:20] i'm not sure the system will survive a second ssh connection [11:21] 30MB ... without swap its very fragile [11:21] you could start it in the background from a shell earlier on [11:21] (and redirect to a file obviously) [11:21] * ogra tries if he can run a shell on serial console while d-i runs inder ssh without hitting OOM [11:36] yay, i wouldnt have expected the system to survive that, but i'm in base-install :) [11:37] Keybuk, so can we have a look at my uuid prob ? [11:38] i have a log of udevadm monitor across the d-i partitioning process [11:39] cjwatson, heh, as i suspected, /dev/sda1 has exactly the uuid i pasted above, even after partitioning, so udev re-uses the old one [11:44] cjwatson, when do you plan to upload d-i *ubuntu20 so i can actually test with a real d-i build instead of my hand knitted one ... seems the kernels have all build now [11:44] * ogra would like to have some testing time before A5 [11:45] ogra: ... last night? [11:45] has there been another kernel upload since? [11:45] I think Steve would be unhappy with me if I uploaded d-i again ... [11:45] yeah [11:46] well, he was aware that rtg uploaded a new kernel for ipx4xx [11:46] but that didn't bump ABI? [11:47] pitti, no, but ixp4xx needs a firmware image rolled by d-i [11:47] ah [11:47] and that needs to include the new kernel [11:54] slangasek, ^^^^ were you aware that the new ixp4xx kernel needs a no-change d-i upload to roll the firmware for ixp4xx when you approved the upload yesterday ? [12:12] ogra: could you put the udevadm monitor log somewhere please? [12:13] will do [12:13] if we're very lucky it's something trivial like something missing from the udeb [12:13] if we're unlucky, the kernel isn't sending any uevents when partitions change [12:15] http://people.ubuntu.com/~ogra/arm/udevadm.log [12:16] it doesnt look like its recieving anything after the initial "add" events [12:19] an strace of parted_server wouldn't hurt, but that's going to be difficult in your environment [12:19] (to confirm that it's doing the right BLKPG ioctls) [12:19] well, i can try to swapon manually before the partitioning starts, not sure if that explodes in my face though [12:20] hmm, i can probably try with a second disk [12:21] * ogra fiddles a bit [12:22] second disk would reduce the probability of interfering with the observation [12:27] query asac did you get my direct message? first time I tried from gwibber [12:29] highvoltage: we can continue here if you want. i dont know where i would see direct messages (is that different from @asac)? === giskard is now known as sldkjfnqwe === sldkjfnqwe is now known as giskard [12:55] ogra: udev won't change the /dev/disk/by-uuid symlink tree unless something tells it to [12:55] Keybuk, yeah, which according to http://people.ubuntu.com/~ogra/arm/udevadm.log doesnt happen [12:56] sda is a usb disk? [12:56] yep [12:56] UEVENT[470.626540] add /devices/pci0000:00/0000:00:01.2/usb1/1-2/1-2:1.0/host0/target0:0:0/0:0:0:0/block/sda (block) [12:56] UEVENT[470.628185] add /devices/pci0000:00/0000:00:01.2/usb1/1-2/1-2:1.0/host0/target0:0:0/0:0:0:0/block/sda/sda1 (block) [12:56] UEVENT[470.629824] add /devices/pci0000:00/0000:00:01.2/usb1/1-2/1-2:1.0/host0/target0:0:0/0:0:0:0/block/sda/sda2 (block) [12:56] UEVENT[470.651764] add /devices/pci0000:00/0000:00:01.2/usb1/1-2/1-2:1.0/host0/target0:0:0/0:0:0:0/block/sda/sda3 (block) [12:56] UEVENT[470.653431] add /devices/pci0000:00/0000:00:01.2/usb1/1-2/1-2:1.0/host0/target0:0:0/0:0:0:0/block/sda/sda5 (block) [12:56] right, it gets added once [12:56] you said this was a udevadm *across* the partitioning [12:56] thats on boot [12:56] well, from boot until base-install [12:57] now look at ~657 [12:57] there are similar adds later too [12:57] the devices are removed and added again [12:57] oh, right [12:57] and again at 659 [12:57] and again at 660 [12:57] thats puzzling [12:57] can you do this log again [12:57] but use "-e" to udevadm monitor [12:58] why dont they get new uuids [12:58] sure [12:58] so, thinking about it, isn't the sequence of events as follows: [12:58] 1) parted_server tells libparted to commit [12:58] 2) libparted tells the kernel to remove existing partitions [12:58] 3) libparted tells the kernel to add new partitions [12:58] let me redo that setup (i was just experimenting wint an additional 2G swap disk to circumvent OOM) [12:58] *with [12:58] 4) something writes new filesystem bits into the new partitions [12:59] 2) and 3) correspond to the remove/add uevents [12:59] however, we don't have the right UUIDs until 4) [12:59] right [12:59] so what is supposed to deal with telling udev about 4)? [12:59] udev 138-2 can do that itself [12:59] earlier versions - you'd have to run udevadm trigger --subsystem-match=block --action=change [13:00] which indeed d-i did [13:00] Keybuk, 137 cant ? [13:00] my current version of d-i still has 137 [13:00] well, it did udevadm trigger --subsystem-nomatch=sound actually, which of course has other problems but ... [13:00] since i have to wait for a new initrd.gz first [13:00] ogra: the current version in the archive is built against 138-2 [13:01] * ogra pulls that [13:01] $ wget -q -O- http://ports.ubuntu.com/ubuntu-ports/dists/jaunty/main/installer-armel/current/images/MANIFEST.udebs | grep -m1 udev-udeb [13:01] udev-udeb 138-2 armel [13:01] cjwatson: that's probably the third set of adds then [13:02] Keybuk: so we should work out how to get rid of this trigger ... [13:02] simply deleting it runs into the problem that d-i loads extra modules at run-time, so udev needs to try some things again [13:03] http://pluto.htu.tuwien.ac.at/How-To_Live-CD_verändern <-- If i create a usplash for the Live CD with this tutorial on Ubuntu 8.10 it will not displayed while booting the CD [13:03]