/srv/irclogs.ubuntu.com/2008/10/06/#ubuntu-installer.txt

=== cjwatson_ is now known as cjwatson
davmor2evand: Final cracked the mail issue in vista I'm now retrying m-a and will bug on what doesn't transfer after :)10:21
CIA-50debian-installer: cjwatson * r970 ubuntu/debian/changelog: releasing version 20080522ubuntu1711:18
davmor2evand: bug 27900011:50
=== davmor2 is now known as davmor2_away
=== davmor2_away is now known as davmor2
evandthanks davmor214:02
superm1xivulon, i build a live disk with that network manager patch, and there were no issues with wubi 50914:35
superm1xivulon, i put some more thought into doing a combined fe/be though - I dont believe you can preseed a bind mount can you?14:35
xivulonhi superm1, what do you mean by "preseeding a bind mount"?14:54
xivulonon my side I can preseed anything I want... It is up to partman(-auto-loop)/d-i/ubiquity to deal with what I preseed...14:56
xivulonAh you mean preseeding the windows drive as a place to use for mythtv backend storage?14:56
superm1xivulon, yeah that's what i mean14:57
xivulonThat will always be mounted as /host so I do not see any need to bindmount it, you just need to create and use directories therein14:57
superm1well the thing is that all of the myth* packages expect to find it in /var/lib/mythtv14:58
xivulonbe aware that write speed to ntfs is worse though in 8.10 (10X worse) since there is no caching anymore14:58
xivulonwrite speed is ok for ntfs inside of ntfs14:58
xivulonsince then ext3 cache is in place14:58
superm1so normally in our regular recipe we make a different partition for /var/lib/  and format that14:59
superm1but since /host is always mounted, we need to bind mount the two14:59
superm1oh well then lets not do this for 8.1014:59
superm1if  performance is a problem i dont want people accounting it to different things and being set away from this project14:59
xivulonyep I do expect performance in 8.10 to be worse then 8.04 when it comes to ntfs writes15:01
xivulonsuperm1 you might want to simply create a /host/mythtv folder and bindmount that, if you want to try things, I do not remember if ubiquity honours the early-command hooks though15:12
davmor2evand: would it help to have a full list of what works and doesn't across the range?15:12
xivulonotherwise you could simply place that in custom-installation/hooks/early-command.sh and use that for testing purposes15:13
evanddavmor2: if you have the time, otherwise I can investigate that when I have more time to look into it15:13
xivulonevand, can you use early-command in ubiquity? or equivalent?15:14
superm1xivulon, well i think 9.04 will be a better focus to determine definitively and plan for this15:14
superm1xivulon, the frontend stuff will be good for now15:14
xivulonsuperm1 ok for me15:14
xivulonI assume there are no changes then on my side, you might want to test though with wubi 51015:15
evandxivulon: https://wiki.ubuntu.com/DesktopCDOptions15:15
xivulonwell superm1 ^ that is handy if you want to play with the backend as well15:16
xivulonpreseed/early_command=/path/to/script.sh15:16
xivulonFor wubi preseed/early_command is set to execute custom-installation/hooks/early_command.sh (if there is one)15:17
cjwatsoncasper implements preseed/early_command, so yes you can use it15:17
xivulonhmm /isodevice is renamed into /host in autopartition-loop, which is run after casper, not sure how that will impact things15:18
cjwatsonerm? autopartition-loop doesn't mention /isodevice15:19
davmor2evand: it's no probs I'm smoke testing anyway.  I think I have your email I'll did about a bit in MM and find out where the settings are and have a look through everything else :)15:20
evandok15:21
davmor2s/did/dig15:21
xivuloncjwatson don't think it does an explicit move, it mounts moves target to host15:21
xivulonbut target is originally called isodevice in casper15:22
cjwatsonxivulon: leaving it as /target would be, er, not a very good idea ;-)15:23
cjwatsonanyhow, it only moves it to /target temporarily so that it can use partman's mount.d scripts15:24
xivulonI agree was just mentioning that, since if superm1 is to use early-command, /host will not be around (or better will be around with a different name)15:24
cjwatsonoh, right, I see15:25
cjwatson(and yes, it's true that autopartition-loop looks around in /proc/mounts for the mount point of the partition you're using for / and moves it)15:25
cjwatsonwell, maybe lupin-casper should just mount it on /host instead, providing that works15:26
xivulonhave yet to provide the patch for dm-loop by the way, but have managed to waste the laptop keyboard for the second time in a year15:27
xivulonthis time it was beer15:27
cjwatsonxivulon: preseed/early_command is no use in superm1's case; he needs something that instructs partman to act differently, not something that instructs the live CD to act differently15:27
xivulonah true15:28
superm1eg it would need to be a patch to one of the partman apps15:28
cjwatsonpartman-auto-loop, surely15:29
superm1so something more worthwhile to worry about the next few months rather than now15:29
cjwatsoncan't it be just another loop-mount file in the Windows drive?15:29
xivulonyes, but I think the main purpose for the backend is to be able to use a LOT of space15:29
superm1the files are going to be quite large, so i dont think most people want a ~200 gig loop mount file sitting in their windows drive15:29
superm1(or more)15:30
xivulonso the idea is to use /host raw15:30
xivulonthat said with syncio patch, writes onto /host should be far worse than writes in a loopfile inside of /host...15:30
xivulonin fact superm1 if you wanted to just test things you could use success commands, copy the files from /var/lib/mythtv to /host/mythtv then add a line to /etc/fstab to bindmount that15:32
superm1xivulon, the expectation would be that it's ready immediately after partitioning15:33
superm1xivulon, because later steps do depend on it15:33
superm1oh but yeah i suppose that would work to just copy them in the success command15:33
xivulonwell move in fact...15:34
xivulonnot for deployment,  but should be ok for a quick test15:34
superm1nonetheless, i'd prefer to defer this for 9.04 for backend/frontend stuff when i've got a lot more time to dedicate to it15:34
xivulonI agree15:34
cjwatsonxivulon: a target-config hook would be more appropriate than a success command15:37
cjwatsoneven that isn't perfect but at least it runs just after copying files15:37
davmor2evand: Should m-a work in alternate?16:02
cjwatsonevand: should usb-creator be seeded somewhere?16:03
xivulondavmor2, did you check if m-a works in wubi?16:03
davmor2xivulon: no it doesn't :(16:03
xivuloncould you install in verbose mode and provide the installation logs?16:04
davmor2xivulon: well nothing obvious is transferred across anyway16:04
xivulonno rush16:04
davmor2xivulon: can do16:04
davmor2xivulon: running now.  This is where I select verbose from the grub line before the install starts right?16:27
evandcjwatson: indeed, I'm thinking it should go on the live CD (if we have space), but I need to make sure it handles that situation well16:41
evand(not trying to install to the device that contains the live system)16:41
evanddavmor2: I don't believe we ever turned it on in the alternate CD, but there should be nothing preventing it from working there if we were to (can't this cycle given the amount of time left)16:42
cjwatsondesktop seems like one obvious place for it16:42
davmor2evand: np's16:42
evandok16:43
davmor2evand, xivulon: www.davmor2.co.uk/install/casper.log www.davmor2.co.uk/install/partman www.davmor2.co.uk/install/syslog www.davmor2.co.uk/install/version www.davmor2.co.uk/install/initial-status.gz for the contents of var/log/installer/* on the verbose wubi install16:49
xivulondavmor2, I do not see anything obvious in the logs, I can see ma-apply being launched and os-probes finding windows on sda1... evand any hint?17:01
evandxivulon: not sure without looking into it myself.  I'd first check to see if anything showed up in ~ right after install, to see if it's just an issue of m-a not writing the correct data.17:06
evandBut m-a is low priority for me at the moment.  There are more serious bugs in the installer that I need to resolve first before I can look at m-a bugs.17:07
xivulonsure17:08
CarlFKsudo update-grub; "Found kernel: /boot/vmlinuz-2.6.27-5-server, Updating /boot/grub/menu.lst ... done"; grep "5-server" /boot/grub/menu.lst = nothing17:32
CarlFKam I missing something, or did it not do what it should have done?17:33
davmor2evand: mail sent.  How do I check for anything showing up in ~ easily is it just a case of monitoring it on another screen?17:35
evanddavmor2: once the install finishes, switch to a vt and look in /target/home/whatever17:36
davmor2no probs I'll try it now and let you know17:36
evandok17:36
evandbe sure to use -a in ls17:37
davmor2np's17:37
davmor2evand: nothing only examples shows up in /target/home/tester18:26
davmor2and about 4 . files18:26
evandwhat are the . files?18:27
davmor2bashrc profiles I think I'll see if I can enter in through rescue before it gets chance to fire any more up18:29
davmor2evand: .bash_logout .bashrc .profile and examples link18:31
davmor2I can re-run and see if there was anything else if you want18:32
evandah, no, that should be sufficient18:32
davmor2evand: Anything else you need?18:40
evandnope, thanks though18:40
davmor2evand: What about info on Vista MM?18:41
evandIn order to support Windows Mail I'll need to know where it stores the settings (registry keys, usually).  I can find that out when developing said support though.18:43
davmor2okay np's18:45
kirklandcjwatson: our HP contact sent me an email with the following... http://pastebin.ubuntu.com/54727/19:07
kirklandTheMuso: ^19:08
cjwatsonthe problem is more likely "Apt update failed" than the dmraid bit19:11
cjwatsonin order to see why we'd need the whole log19:11
cjwatsonbut it may well be the cluster of problems around bugs 273979 and 27634919:13
CIA-52tasksel: cjwatson * r1378 ubuntu/ (4 files in 2 dirs):20:54
CIA-52tasksel: Update Ubuntu tasks from seeds, adjusting descriptions of samba-server,20:54
CIA-52tasksel: tomcat-server, and virt-host tasks.20:54
CIA-52tasksel: cjwatson * r1379 ubuntu/debian/changelog: releasing version 2.73ubuntu1020:55
LogicBoxhey can anyone help with this i put the install disk in it comes up to boot menu letting me choose run, install, verify, etc.. and what happens is i click on install (have tried just running it from cd) and it shows the ubuntu logo and the status loading bar and then it just goes into a screen with text about Busybox v1.1.3 with some more txt then gives me what looks like a command line with...21:10
LogicBox...a blinking cursor21:10
evandLogicBox: if you hit f6 at the boot menu, then hit the end key and remove 'quiet splash' from the end of that string and hit enter, it should show you a more detailed error.21:11
LogicBoxok like me try that21:12
LogicBoxi didnt see anyting that said Quiet splash21:14
LogicBoxwhat pops up afte rthe gui spalsh screen is this "BusyBox v1.1.3 (Debian 1:1.1.3-5Ubuntuv12) Built-in shell (ash) Enter 'help' for a list of built-in commands.  then it has (initramfs)_ on two lines below21:16
LogicBoxthe _ is a blinking cursor21:16
cjwatsonboot CD, press Enter to dismiss language screen, press F6, then a "Boot Options" line appears with "quiet splash --" at the end21:16
cjwatsonthis is all before you actually boot Linux21:16
LogicBoxya21:16
cjwatsonwhereas the stuff you're describing is after the CD boot menu disappears21:16
LogicBoxi was trying to install this on a dell dimension 240021:17
LogicBoxright after i say install21:17
LogicBoxbut i did restart and go into the boot menu and pressed f621:17
cjwatsonright, and what happened?21:17
LogicBoxhere i will write my options hold 1 sec21:17
LogicBoxOther Options: acpi=off  noapic nolapic edd=on free software only21:18
cjwatsondismiss that (press escape)21:18
cjwatsonthere should be a "Boot Options" line21:18
cjwatson(it sounds like you pressed F6 twice by mistake)21:18
LogicBoxnope f6 only once21:20
LogicBoxright now its sitting at the Ubuntu menu, the Try ubuntu, install, check CD, Test mem, boot from hdd21:20
cjwatsonnever mind, just dismiss that menu and you should see a "Boot Options" line stretching all the way across the screen just above the menu bar at the bottom21:20
LogicBoxi tried this on a newer dell and it worked just fine21:21
LogicBoxtried the cd21:21
LogicBoxya i saw that line21:21
cjwatsonok, at the end of it is "quiet splash --"21:21
LogicBoxes21:21
LogicBoxyes21:21
cjwatsondelete "quiet splash" from that line21:21
cjwatson(by the way, this is almost certainly a kernel bug not an installer bug, so all we're doing here is helping you gather information that you can pass onto the kernel guys later)21:22
LogicBoxahh21:22
LogicBoxok an error poped up that well it repeated itself saying end_request: I/O error, dev fd0, sector 021:23
cjwatsonyou can skip that error21:23
LogicBoxnow its just sitting at that busybox msg again21:23
cjwatsonwere there no other messages?21:23
LogicBoxnot that i can see21:24
cjwatsontype "cat /casper.log"21:24
LogicBoxi see ha:cache flushes supported21:24
cjwatsonwithout the quotes of course21:24
LogicBoxhda: hda121:24
cjwatsonthis will be easier if you have a digital camera21:24
cjwatsontranscribing into IRC is hopeless21:24
LogicBoxok i typed that and took a pic21:26
LogicBoxloading it into photoshop 1 sec21:26
LogicBoxcj u want me to err i will uploaded it to my photo bucket21:27
LogicBoxhttp://i485.photobucket.com/albums/rr213/SilverRavage/IMG_01791.jpg21:28
LogicBoxsorry the iPhone takes a shitty pic21:28
cjwatsonok, now "dmesg"21:29
LogicBoxok a bunch of crap just scrolled through the screen really fast21:30
cjwatsonyou can use "dmesg | more" instead to get it displayed in pages; use space to display the next screen21:30
LogicBoxare u ganna want a screen cap of all of them?21:32
cjwatsonthe kernel guys typically ask for dmesg output, yes21:35
LogicBoxis there any way to pull it like maybe to a usb key21:35
LogicBoxmake it to a txt21:35
cjwatsonremember that they have very little to go on to start with; they don't have your machine, and just telling them the model name actually helps a whole lot less than you'd think21:35
cjwatsonhow much Linux experience do you have? it's possible, but it will be tricky if you haven't used the command line much21:36
cjwatsonthe usual facilities aren't available at this level21:36
LogicBoxmeh not much, ive installed ubuntu b 4 and played around with it but thats about it.21:36
cjwatsonit would involve inserting the USB key, figuring out the device name and mounting it on a scratch directory, say /mnt, then 'dmesg > /mnt/filename'21:37
cjwatsonif the device doesn't become available immediately you'd have to poke udev a bit21:38
LogicBoxok i will try and see what i can do21:38
cjwatsonother things you'll need will be "ls /sys/block" and "cat /proc/bus/pci/devices"21:39
cjwatsonoh, and "uname -a" and "cat /proc/version_signature"21:39
cjwatson(see https://wiki.ubuntu.com/KernelTeamBugPolicies)21:40
cjwatson/proc/bus/pci/devices replaces the lspci command there since lspci isn't available at this point21:40
cjwatsononce you've collected that information, you should file a bug on https://bugs.launchpad.net/ubuntu/+source/linux/+filebug saying that your CD drive isn't being detected21:41
cjwatsonalso remember to quote the version of Ubuntu you're using ;-)21:41
cjwatsonLogicBox: https://wiki.ubuntu.com/DebuggingCasper may help you21:44
ago_evad, on 247461, is umenu shipped with the DVD a different version?22:16
evandago_: umenu is the same on all CDs22:18
evanderr22:18
evandyeah, it is.  Wubi just isn't on the DVDs.22:19
ago_what about umenu?22:19
ago_I remember fixing umenu not to show the wubi entry so that it could be used on DVD22:20
evandIt's on all the images, DVDs included.22:20
ago_evand: one possible explanation for the bug above is that we bumped the  version number (checked by umenu) to .1 but that new umenu revision didn't make the DVD22:22
TheMusokirkland: What were you pointing me to?22:23
Fragadelichello22:23
kirklandTheMuso: dmraid not found error, possibly known apt error in the installer22:23
kirklandTheMuso: see that pastebin22:23
Fragadeliccjwatson - are you online now?22:25
Fragadelicanyone have any idea why the installer would hang at 94% trying to reconfigure popularity-contest on a remastered version of Ubuntu?22:31
cjwatsonFragadelic: run it with DEBCONF_DEBUG=developer on the kernel command line, post /var/log/syslog22:35
cjwatson(or /var/log/installer/syslog if that exists)22:35
kirklandcjwatson: i'm curious... would it make sense to just keep DEBCONF_DEBUG=developer through alpha/beta, and just switch it off before rc?22:36
cjwatsonno, because it logs passwords22:37
FragadelicI'll try that - killing the reconfigure lets it move on without issues.  Its only on a remaster - original is fine.22:38
cjwatsonhttp://bugs.debian.org/cgi-bin/bugreport.cgi?bug=35711822:38
cjwatsonkirkland: ^-22:38
Fragadelicits almost like the popularity-contest package is originally prepared in a certain way and after an initial install it starts with this behaviour22:38
kirklandcjwatson: gotcha22:39
cjwatsonI'm not sure how much help debconf debugging will be, but at any rate I certainly need to see the log22:40
cjwatsonFragadelic: alternate installer or desktop installer?22:40
Fragadeliccjwatson: regular desktop installer - has been happening since I think 7.10 but 8.04 for sure and 8.04.1.  I had to have remastersys remove the popularity contest and ubuntu-standard packages so install would work normally but I'd prefer to leave it there.  I also tried many variations of removing and purging the package and installing it without configuring,etc - will have to change remastersys to22:43
Fragadelicleave popularity-contest in and test.  will take a bit to remaster and test but I'll be back.  you want it posted here or in the list?22:43
cjwatsonFragadelic: in a bug report on ubiquity, ideally22:50
cjwatson(https://bugs.launchpad.net/ubuntu/+source/ubiquity/+filebug)22:51
cjwatsonFragadelic: though you could link to the bug here22:51
Fragadeliccjwatson: I'll do that but it doesn't really relate to an original ubuntu install - it only happens on remasters.  Thanks for your time.22:55
cjwatsonFragadelic: I understand, but I'm keen to fix this sort of thing if it does turn out to be a bug since it means fewer questions for us to answer22:55
Fragadeliccjwatson:  sounds good - up until now I've just been putting workarounds in remastersys to get around certain things in casper and ubiquity22:57
cjwatsonit may turn out that we can't fix it, in which case I'll mark it invalid with advice22:58
Fragadelicis there a list of other debug options for the command line?  I know about break=bottom as I stumbled upon that accidentally but it would be nice to have any others so I could help folks with their backups even more.22:59
Fragadelici tried checking the package info for the different versions but couldn't really find a lot of info about ubiquity and casper which is why I had to spend hours going through all the differnet scripts - lol23:00
Fragadelicif only I'd know I could contact you all like this without actually being part of the team a year ago it would have  made things much easier for me - I sent you an email eons ago but never got a response.  I'm assuming now that it might have been due to the fact that I use a yahoo email address for portability reasons.23:04
=== cjwatson_ is now known as cjwatson

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!