/srv/irclogs.ubuntu.com/2010/11/17/#ubuntu-installer.txt

evindeed, that's been the plan for some time.  DX is working on it (specifically Ted, last I checked)01:42
evthough yes, I do sit right next to mpt :)01:43
CIA-4usb-creator: evand * r327 usb-creator/ (3 files in 2 dirs):10:44
CIA-4usb-creator: * debian/control, usbcreator/frontends/gtk:10:44
CIA-4usb-creator:  - Remove python-gnome2 dependency by switching help_display_uri to10:44
CIA-4usb-creator:  gtk.show_uri. LP: #66128910:44
CIA-4usb-creator: evand * r328 usb-creator/debian/changelog: releasing version 0.2.2611:03
CIA-4ubiquity: evand * r4435 trunk/ (debian/changelog ubiquity/frontend/gtk_ui.py):11:07
CIA-4ubiquity: Expose the navigation control in the GTK frontend API.11:07
CIA-4ubiquity: Thanks Ying-Chun Liu!11:07
sorenIs there any particular reason why the root filesystem is placed at the start of the disk rather than at the end?13:49
sorenI'm sure there is, but what is it? :)13:49
sorenIt would be handy if they were at the end. If you install Ubuntu in a VM on a virtual disk and decide to expand the disk later, it's a bit of a bummer that you can't just extend the partition.13:52
sorenAlternatively, maybe it's time we use lvm by default? :)13:53
evlvm> what doesn't btrfs solve?14:45
evI appear to have lost the magic runes to shoving a partman-auto udeb kicking and screaming into a running d-i instance16:14
cjwatsondoes it involve new debconf templates?16:15
evyes16:16
evwhich I separately imported using debconf-loadtemplate after sed'ing out the underscores16:16
cjwatsonyeah, that doesn't really work16:17
evstill hanging on a description metaget16:17
evah16:17
evfail16:17
cjwatsonbecause you end up talking to the wrong debconf instance16:17
evright!16:17
evugh16:17
cjwatsonbuild the udeb, and edit some random victim postinst script (I usually use load-cdrom.postinst, after the call to anna)16:17
cjwatsonmake it do wget and udpkg -i16:17
cjwatson(or udpkg --unpack if the udeb you're using has a postinst, but partman-auto doesn't)16:17
evah, thanks!16:26
sorenev: Well, not everyone uses btrfs, but does btrfs let you add another block device to it like lvm does?19:13
corecodehi21:27
corecodewill the installer/preseed only use eth021:27
corecodeor will it probe other cards as well?21:27
corecodesomehow the automatic install on this machine didn't work out21:27
cruejoneshi, why am i able to user an ubuntu DVD iso loopback shared via http as a local mirror but standard debian fails with the same setup?22:01
cruejonesfor netinstalls22:01
cjwatsonto make that work you have to include a few extra packages on the iso (things that aren't on the netboot initrd but are in the cdrom initrd) and Debian may well not do that22:03
cjwatsonI don't think either of us really consider it a properly supported setup, it just happens to work in Ubuntu at the moment22:03
cruejonesthanks - any pointers on what files might be needed?22:04
cjwatsoncorecode: you should be able to preseed netcfg/choose_interface.  I think the default is something like the first interface that seems to have a link, or possibly just the first interface22:04
cjwatsoncruejones: sorry, I don't have a list22:05
cruejonescjwatson: any pointers on where to look?22:05
cjwatsonthere's a udeb manifest alongside the initrds22:05
cruejonesok22:05
cjwatsonhttp://ftp.debian.org/debian/dists/testing/main/installer-i386/current/images/ or something like that22:05
cjwatsonyou'd need to subtract the files in the netboot initrd from those in the cdrom initrd, and include the difference22:06
cruejonesalready having to use unionfs on iso to fix some symlink issues so I might be able to fake a mirror by adding those files to the iso via the union22:06
cruejonesand please consider it an important configuration for future support - network shared loopback isos is a classic scenario for space savings22:07
cruejonesand in my case I need to use a different non-debian based distro as the pxe/netboot install source so creating a local mirror is non-trivial22:08
cruejonescjwatson: so I just thought about it and I am confused as to why I need to care about the netboot vs. dvd initrd?  I am using the kernel/ramdisk from the netboot (mini.iso) which works just fine.  Then I loopback mount the main DVD and share it via http.22:11
cruejoneswhen I point the netboot installer to the dvd/iso http share it returns that it is a bad mirror - so I would assume I am missing some repo/mirror type files on the main DVD?22:12
cruejonesand thanks for helping - seeing as this is more pure debian question22:12
cjwatsoncruejones: because the way the installer works is that it only ships the necessary core of itself in the initrd, and fetches more bits of itself at run-time22:16
cjwatson(udebs)22:16
cjwatsoncruejones: the "necessary core" depends on the installation method - netboot, cdrom, etc.22:16
cjwatsonfor netboot initrds it has to include the network configuration components etc. in the initrd; for cdrom initrds it needs to know how to mount the cdrom22:16
cjwatsonhowever, when netbooting, there are bits that still need to be performed *eventually*, just not necessarily so early that they have to be in the initrd22:17
cjwatsonso the udebs for those tasks need to be available somehow22:17
cruejonesok, but in theory what I am doing is a pure netinstall, just that my mirror is really a loopback mount of the main DVD shared via http22:18
cjwatsonmind you returning "bad mirror" is actually something a bit different, now that I read it22:18
cruejonesright22:18
cjwatsonyou may well have the same problem later though22:18
cjwatsonfor Debian, could we move this to #debian-boot on irc.oftc.net please?22:18
cjwatsonthat's the corresponding Debian channel22:19
cruejonesok, thanks22:19
corecodewhat could be the reason that list-devices disk returns empty?22:52
cjwatsoncorecode: what's in /sys/block (excluding the ram* entries)?22:56
corecodeloop22:56
cjwatsonthen the kernel hasn't found your disks22:57
corecode0f:04.0 SCSI storage controller: LSI Logic / Symbios Logic SAS1064 PCI-X Fusion-MPT SAS (rev 02)22:57
corecodeyea, but why22:57
cjwatsoncould be because the module for that disk isn't in the installer initrd for some reason22:57
cjwatsonwhat's the numeric pci id for that?22:57
cjwatsonnever mind, looked it up, 1000:005022:58
corecodeyep22:58
corecodelspci -nv over 9600bps is a bit slow22:58
corecodeon that machine in particular22:58
cjwatsonlsmod | grep mptsas23:01
corecodenop23:01
cjwatsonwhat release?23:01
corecodemaverick23:02
corecodeobscure23:03
corecodelet me reboot23:03
cjwatsonno wait!23:03
corecodemaybe that fixes something23:03
corecodeok23:03
cjwatsonnano -v /var/lib/dpkg/status23:03
cjwatsonCtrl-w message-modules ENTER23:03
cjwatsondoes that show anything?  if so what's the Status field?23:03
corecodenothing23:03
cjwatsonno match?23:04
corecodeno match23:04
cjwatsoncan you extract /var/log/syslog and put it on paste.ubuntu.com for me?23:04
corecodeyes23:04
corecodehang on23:04
corecodeif i can get scp on this box23:04
cjwatsonanna-install openssh-client-udeb23:05
corecodehm, that just clears the screen23:05
cjwatsonyes, but now you have scp23:05
cjwatsonor should do23:06
corecodeno :/23:06
corecodesomething is a bit wrong23:06
cjwatsontry "save debug logs" from the installer main menu23:06
cjwatsonsounds like your installer is incapable of installing extra components of itself for some reason23:06
cjwatsonwhat installer image are you using?23:06
corecodethe maverick amd64 netboot image23:06
corecodebut that worked for all other machines23:07
cjwatsonno, which url23:07
cjwatsonexactly23:07
corecodei can't tell right now23:08
cjwatsonok, well, you mentioned a network configuration problem23:08
corecodeyes23:08
cjwatsonhow did you resolve that, and what did you do afterwards?23:08
corecodei did manually dhclient eth123:08
corecodeand then let the installer continue23:08
corecodethen it must have rebooted, but didn't finish the installation + preseed post script23:09
corecodehmmm23:09
cjwatsonsounds like some bit of network configuration wasn't properly set up23:09
corecodemaybe there is some nfsv4 issue there as well23:09
cjwatsonand therefore it couldn't fetch installer components off the network23:09
corecodebut that's unrelated23:09
corecodequite possible23:10
cjwatsonyou're probably right to reboot at this point23:10
corecodehow can i find out why it didn't manage to pull up the network?23:11
cjwatsonread through /var/log/syslog I guess23:11
cjwatsoneverything should be in there somewhere, though it's pretty verbose23:11
corecodehttp://pastebin.ca/199505723:11
corecodeit keeps talking about eth1: link is not ready23:13
corecodebut that seems to be due to a missing ifconfig up23:14
cjwatsonnot sure I have any idea, sorry23:14
cjwatsonnormally you shouldn't need ifconfig up to get a link indication23:14
corecodeNov 17 23:05:25 kernel: [ 3720.533354] ADDRCONF(NETDEV_UP): eth1: link is not ready23:14
corecodeNov 17 23:05:25 dhclient: DHCPDISCOVER on eth1 to 255.255.255.255 port 67 interval 323:14
corecodeNov 17 23:05:27 kernel: [ 3722.590977] e1000e: eth1 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: None23:14
corecodeNov 17 23:05:27 kernel: [ 3722.594115] ADDRCONF(NETDEV_CHANGE): eth1: link becomes ready23:14
corecodemagical23:14
cjwatsonthe MII link indication is what tells you that you can do ifconfig up23:15
cjwatsonas I understand it23:15
cjwatsonbut this isn't particularly my field of expertise, so "I don't know, sorry"23:15
corecodeyea23:15
corecodemmm23:15
corecodeoh also the kernel doesn't seem to reliably number the network cards the same way23:22
cjwatsonno, that has been a bit of a saga to try to fix I understand23:28
corecodehmm23:41

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