Re: fai-chboot -c annoyance?
Holger Levsen writes: > Hi, > > On Wednesday 28 June 2006 11:39, Andreas Sindermann wrote: > > So I suggest to remove the checking of .disable files because one > > wants to create a new profile anyway. It might be even a better idea > > to remove the old .disable file automatically. > > why not a -f|--force switch ? -f fai_flags Set FAI_FLAGS. The flags must be comma separated. -F Set default values for FAI_FLAGS. This is the same as -f ver? bose,sshd,createvt Andreas
fai-chboot -c annoyance?
Hi, I'd like to suggest a slight change in the behaviour of 'fai-chboot -c' which I'm using to change quickly between several system setups (Ubuntu 5.04 and Ubuntu 6.06). After an installation the pxe boot profile /boot/fai/pxelinux.cfg/865F439C is renamed to 865F439C.disable so that the installed client can boot from it's freshly installed hard disk. 'fai-chboot -c' first checks whether such a .disable file exists and gives an error message instead of simply copying the template file: [EMAIL PROTECTED]:~# fai-chboot -ec ubuntu l36 Config for l36 is currently disabled. Copying aborted. This is somewhat annoying because I don't mind whether it was disabled due to an old installation (which might have been performed even half a year ago or whenever). So I suggest to remove the checking of .disable files because one wants to create a new profile anyway. It might be even a better idea to remove the old .disable file automatically. Thanks for listening... Andreas
Re: alsaconf over FAI
Peter Burger writes: > Hi there, > > I would config my alsa-sound with the command "alsaconf" over FAI. Therefore > I have to interact with the alsaconf-gui (press four times enter). Which > answers I have to put in the FAIBASE to solve this problem? > > Regards Peter You should do it once interactively and afterwards do a debconf-get-selections |grep to find out the corresponding settings for your packages. In the fai configuration space you then can add these lines to debconf/ so that your debconf settings will be available and recognized during the installation process. Andreas
Re: network interfaces exchanged
Sorry for answering myself..
Several things needed to be done to enable proper networking on
Ubuntu 6.06 (dapper):
a) /etc/resolv.conf is handled dynamically by the resolvconf
facility. For that reason it is necessary to add the following two
lines to $target/etc/interfaces instead of fcopy'ing a static
/etc/resolv.conf file (in the FAI example files this is done in the
scripts/30-interface script):
dns-search $DNSDOMAIN
dns-nameservers $DNSSRVS
b) During FAI installation I save the hardware address and the
interface name of the interface being used during installation in
/etc/mactab of the newly installed system (FAI examples:
scripts/30-interface):
# FAI Ubuntu kernel thinks something different than the normal Ubuntu
# kernel about the order of the network interfaces. So use nameif(1)
# which reads /etc/mactab for defined order of interfaces.
if [ -f $LOGDIR/dhclient.log ]; then
myif=`grep Listening $LOGDIR/dhclient.log | awk -F/ '{print $2}'`
mymac=`grep Listening $LOGDIR/dhclient.log | awk -F/ '{print $3}'`
echo "$myif $mymac" > $target/etc/mactab
fi
b) To reorder (only if necessary) the network interfaces I've created
a little script /usr/local/sbin/ifexchange on the installed FAI
client (based on the 'nameif'):
snip --
#!/bin/sh
# In case the system has two interfaces and the kernel accidently
# configured them in the wrong order, simply exchange them, i.e.
# eth0 becomes eth1 and vice versa.
# 1. Are there two interfaces (eth0 and eth1) at all?
if [ `grep -c eth.: /proc/net/dev` -ne 2 ]; then
exit 0
fi
# 2. If MAC address used during FAI installation is associated with the same
#interface used during FAI everything is fine, else the interfaces
#need to be exchanged.
fai_if=`awk '{print $1}' /etc/mactab`
fai_mac=`awk '{print $2}' /etc/mactab`
current_if=`ifconfig -a|grep -i $fai_mac|awk '{print $1}'`
if [ $fai_if != $current_if ]; then
# Interfaces need to be exchanged!
# Get 'wrong' hardware address
wrong_mac=`ifconfig $fai_if|grep $fai_if|awk '{print $5}'`
# Shutdown network
ifdown -a
# Assing interfaces with exchanged hardware addresses:
nameif g $wrong_mac
nameif $fai_if $fai_mac
nameif $current_if $wrong_mac
# Bring up network
ifup -a
fi
-- snip --
This ifexchange script is called during the booting phase:
/etc/init.d/checkinterfaces:
-- snip --
#!/bin/sh
[ -x /usr/local/sbin/ifexchange ] && /usr/local/sbin/ifexchange
- snip --
$ROOTCMD update-rc.d checkinterfaces start 41 S 0 6 .
This seems to work at least for me. Perhaps there are others who might
want to try it out..
Andreas
Re: network interfaces exchanged
Henning Glawe writes:
> On Fri, May 19, 2006 at 10:49:11AM +0200, Andreas Sindermann wrote:
> > during the fai-installation the nfsroot-kernel sees the two network
> > interfaces eth0 and eth1 in a different order than the kernel used
> > during normal operation. The effect is that when booting from harddisk
> > no network is available. How does one care about this usually?
>
> use something like ifrename or nameif to rename the interfaces by looking at
> the MAC address. In testing/unstable, this functionality is provided by
> udev.
This seems to be exactly what I'm looking for. Unfortunately when
simply creating the /etc/mactab file
(in /scripts/FAIBASE/30-interface) with the following few lines
the installed system doesn't recognize that it should use 'nameif' to
reorder the interfaces. Does one have to create explicitely a
new boot-script or something in /etc/network/if-pre-up.d?
(this is on an Ubuntu dapper system) Or is there something something
in the /etc/udev/ region that needs to be changed?
# FAI Ubuntu kernel thinks something different than the normal Ubuntu
# kernel about the order of the network interfaces. So use nameif(1)
# which reads /etc/mactab for defined order of interfaces.
if [ -f $LOGDIR/dhclient.log ]; then
myif=`grep Listening $LOGDIR/dhclient.log | awk -F/ '{print $2}'`
mymac=`grep Listening $LOGDIR/dhclient.log | awk -F/ '{print $3}'`
echo "$myif $mymac" > $target/etc/mactab
fi
Andreas
--
Dr. Andreas Sindermann fon: +49 (221) 470-4201
Institut fuer Theoretische Physikfax: +49 (221) 470-5159
Universitaet zu Koeln
Zuelpicher Str. 77 mailto:[EMAIL PROTECTED]
D-50937 Koeln, Germany http://www.thp.uni-koeln.de/~sinder
network interfaces exchanged
Hi, during the fai-installation the nfsroot-kernel sees the two network interfaces eth0 and eth1 in a different order than the kernel used during normal operation. The effect is that when booting from harddisk no network is available. How does one care about this usually? Andreas
Re: newer fai kernel
Xavier Claessens writes: > Le lundi 08 mai 2006 à 12:25 +0200, Andreas Sindermann a écrit : > > Xavier Claessens writes: > > > I got this error on ubuntu dapper when running "fai-setup": > > > Kernel installed into the nfsroot. > > > FATAL: Could not open '/boot/System.map-': No such file or directory > > > cp: ne peut =C3=A9valuer `/usr/lib/fai/nfsroot/boot/vmlinuz-': Aucun > > fichie= > > > r > > > ou r=C3=A9pertoire de ce type > > > > > > Here is files I have in the nfsroot/boot: > > > config-2.6.16-fai-kernels memtest86+.bin patches-2.6.16-fai-kernels > > > System.map-2.6.16-fai-kernels vmlinuz-2.6.16-fai-kernels > > > > > > So the problem is the capital "S" for "System.map-" and fai-setup > > > searches without the capital "system.map-" > > > > What does your make-nfs-root.conf file contain? > > (grep KERNEL /etc/fai/make-fai-nfsroot.conf) > > > > For me the capital "S" is not a problem... > > > > Here my nfsroot/boot directory (looks pretty much the same): > > > > /usr/lib/fai/dapper-nfsroot/boot# ll > > total 1908 > > -rw-r--r-- 1 root root 531144 Apr 9 02:44 System.map-2.6.16-fai-kernels > > -rw-r--r-- 1 root root 32985 Apr 9 02:20 config-2.6.16-fai-kernels > > -rw-r--r-- 1 root root 94760 Oct 25 2005 memtest86+.bin > > -rw-r--r-- 1 root root 48 Apr 9 02:44 patches-2.6.16-fai-kernels > > -rw-r--r-- 1 root root 1268112 Apr 9 02:44 vmlinuz-2.6.16-fai-kernels > > > > Make sure that on the install server the /boot/fai/vmlinux-install > > (not in the nfsroot environment!!) is the same file as > > /usr/lib/fai/nfsroot/boot/boot/vmlinuz-2.6.16-fai-kernels: > > > > ls1:/usr/lib/fai/dapper-nfsroot/boot# cmp vmlinuz-2.6.16-fai-kernels > > /boot/fai/vmlinuz-install > > ls1:/usr/lib/fai/dapper-nfsroot/boot# echo $? > > 0 > > > > If "echo $?" gives something different than something is broken... > > > > Andreas > > [EMAIL PROTECTED]:/etc/fai# grep KERNEL /etc/fai/make-fai-nfsroot.conf > KERNELPACKAGE=/usr/lib/fai/kernel/linux-image-2.6.16-fai-kernels_1_i386.deb > > [EMAIL PROTECTED]:/usr/lib/fai/nfsroot/boot# cmp > vmlinuz-2.6.16-fai-kernels /boot/fai/vmlinuz-install > vmlinuz-2.6.16-fai-kernels /boot/fai/vmlinuz-install sont différents: > octet 501, ligne 5 > [EMAIL PROTECTED]:/usr/lib/fai/nfsroot/boot# echo $? > 1 > > don't know what's the problem. I just downloaded fai-kernels-1.10.3 and > installed it by "dpkg -i". > > Any idea to solve this problem ? Copy /usr/lib/fai/nfsroot/boot/vmlinuz-2.6.16-fai-kernels: $ cp /usr/lib/fai/nfsroot/boot/vmlinuz-2.6.16-fai-kernels /boot/fai/vmlinuz-install Then it should work properly (I think). Perhaps you want to save the original vmlinuz-install before. To the developers: I'm wondering whether it would be nice that the 'fai-chboot' script would check by default whether /boot/fai/vmlinuz-install is the appropiate boot-sector (perhaps a naming problem) or not. Currently I'm trying to install both i386 and amd64 architectures in parallel and often I forget to copy the correct vmlinuz-install by hand so that the installation process doesn't start properly... Andreas
Re: newer fai kernel
Xavier Claessens writes: > I got this error on ubuntu dapper when running "fai-setup": > Kernel installed into the nfsroot. > FATAL: Could not open '/boot/System.map-': No such file or directory > cp: ne peut =C3=A9valuer `/usr/lib/fai/nfsroot/boot/vmlinuz-': Aucun fichie= > r > ou r=C3=A9pertoire de ce type > > Here is files I have in the nfsroot/boot: > config-2.6.16-fai-kernels memtest86+.bin patches-2.6.16-fai-kernels > System.map-2.6.16-fai-kernels vmlinuz-2.6.16-fai-kernels > > So the problem is the capital "S" for "System.map-" and fai-setup > searches without the capital "system.map-" What does your make-nfs-root.conf file contain? (grep KERNEL /etc/fai/make-fai-nfsroot.conf) For me the capital "S" is not a problem... Here my nfsroot/boot directory (looks pretty much the same): /usr/lib/fai/dapper-nfsroot/boot# ll total 1908 -rw-r--r-- 1 root root 531144 Apr 9 02:44 System.map-2.6.16-fai-kernels -rw-r--r-- 1 root root 32985 Apr 9 02:20 config-2.6.16-fai-kernels -rw-r--r-- 1 root root 94760 Oct 25 2005 memtest86+.bin -rw-r--r-- 1 root root 48 Apr 9 02:44 patches-2.6.16-fai-kernels -rw-r--r-- 1 root root 1268112 Apr 9 02:44 vmlinuz-2.6.16-fai-kernels Make sure that on the install server the /boot/fai/vmlinux-install (not in the nfsroot environment!!) is the same file as /usr/lib/fai/nfsroot/boot/boot/vmlinuz-2.6.16-fai-kernels: ls1:/usr/lib/fai/dapper-nfsroot/boot# cmp vmlinuz-2.6.16-fai-kernels /boot/fai/vmlinuz-install ls1:/usr/lib/fai/dapper-nfsroot/boot# echo $? 0 If "echo $?" gives something different than something is broken... Andreas
Re: newer fai kernel
Thomas Lange writes: > >>>>> On Sun, 7 May 2006 20:23:35 +0200 (MEST), Andreas Sindermann <[EMAIL > >>>>> PROTECTED]> said: > > > I'm trying to use kernel-image-2.6.8-fai_1_i386.deb within the nfsroot > > > Perhaps it would be nice to provide a newer fai-kernel package as > > 2.6.8 is already quite old (summer 2004?) and I think the issues with > fai-kernels 2.10.1 includes a 2.6.16 kernel. Include following line > into your sources.list file (even when using ubuntu) and you will get > the new packages. It's fai-kernels 1.10.3 and the line to be added to sources.list is (I think): deb http://www.informatik.uni-koeln.de/fai/download sarge koeln Thanks for the hint... :-) Andreas
Re: Kubuntu fai 2.9.1 problem
Thomas Lange writes: > >>>>> On Tue, 2 May 2006 12:04:12 +0200 (MEST), Andreas Sindermann <[EMAIL > >>>>> PROTECTED]> said: > > > Hi, > > currently I'm doing some experiments with Kubuntu 6.06 beta 2 > > (dapper). I'm trying to install the i386 version of Kubuntu on my > > Athlon64 machines to avoid the 32bit-chroot environment that is needed > > by a number of programs (acroread, firefox plugins and others). > > > fai-2.9.1 is coming with this distribution (all fai packages except > > the fai-kernels, but this is not a problem) and a strange issue occurs > > after setting up /etc/fai/make-fai-nfsroot.conf, /etc/fai/fai.conf and > > /etc/fai/sources.list: > Known bug in Ubuntu. > > https://launchpad.net/distros/ubuntu/+source/fai/+bug/34366 > > We wonder why this makes problems in ubuntu, because fai is using > dpkg-divert and it works without problems on Debian. I already thought about something with
Kubuntu fai 2.9.1 problem
Hi, currently I'm doing some experiments with Kubuntu 6.06 beta 2 (dapper). I'm trying to install the i386 version of Kubuntu on my Athlon64 machines to avoid the 32bit-chroot environment that is needed by a number of programs (acroread, firefox plugins and others). fai-2.9.1 is coming with this distribution (all fai packages except the fai-kernels, but this is not a problem) and a strange issue occurs after setting up /etc/fai/make-fai-nfsroot.conf, /etc/fai/fai.conf and /etc/fai/sources.list: 'make-fai-nfsroot' stops with the following message which I don't understand at all: [...] Setting up fai-client (2.9.1) ... Setting up fai-nfsroot (2.9.1) ... Configuration file `/etc/dhcp3/dhclient-script' ==> Deleted (by you or by a script) since installation. ==> Package distributor has shipped an updated version. What would you like to do about it ? Your options are: Y or I : install the package maintainer's version N or O : keep your currently-installed version D : show the differences between the versions Z : background this process to examine the situation The default action is to keep your current version. *** dhclient-script (Y/I/N/O/D/Z) [default=N] ? dpkg: error processing fai-nfsroot (--configure): EOF on stdin at conffile prompt Errors were encountered while processing: fai-nfsroot E: Sub-process /usr/bin/dpkg returned an error code (1) Removing `local diversion of /sbin/discover-modprobe to /sbin/discover-modprobe.distrib' [EMAIL PROTECTED]:/usr/lib/fai# Does anybody see why it doesn't simply select the default action (leave everything as is) and continues? Andreas
Re: LVM on FAI-installed computers
Mathias Friman writes: > I have a volume group named "vg_data" and three lv's named "lv_home", > "lv_usr" and "lv_var" respectively, each with a ext3 filesystem. > > I try to mount the lv's in fstab: > /dev/vg_data/lv_home /home ext3 defaults 0 0 > > I get as far as > Setting up LVM Volume Groups... > Reading all physical volumes. This may take a while... > Found volume group "vg_data" using metadata type lvm2 > /dev/vg_data: opendir failed: no such file or directory > /dev/vg_data: opendir failed: no such file or directory > /dev/vg_data: opendir failed: no such file or directory > > and at the end of the boot, none of the volumes are being mounted. > > I can manually run the lvm-script and mount the volumes from the command > line. > > "# vgchange -a y" says > 3 logical volume(s) in volume group "vg_data" now active > > "lsmod" shows that dm_mod is loaded. > and the devices mapped in /dev/vg_data is soft links > lv_home -> /dev/mapper/vg_data-lv_home .. etc. Hm, perhaps as a workaround right before the lvm startup script you simply add another script (with FAI) containing the vgchange line mentioned above? Andreas
Re: FAI & Kubuntu
In principle it is possible. I have a running Kubuntu (Ubuntu with a single KDE meta-package added which overrides GNOME) environment installed from a FAI Debian install server (which at least at that time was the most simple way to do it). But this is quite old (Kubuntu 5.04 with FAI 2.8.4 so I can't tell something concerning the recently split up fai packages in connection with Kubuntu 5.10 (or even 6.04). Anyway, the main idea was (if I remember correctly) to remove the original Debian debootstrap package on the install server and to replace it with the Ubuntu debootstrap package to be able to create an Ubuntu-nfsroot environment on the install server. A big advantage is to have an install server of same architecture (e.g. i386) as all the client machines to be installed. Otherwise you have to setup an interim install server of the clients' architecture on which you only create the nfsroot environment which then needs to be copied to the 'real' install server (e.g. we have an i386 (Xeon) install server but all our clients are x86_64 (Athlon64) machines so I had to prepare the nfsroot on one of the clients first and then copied it to the i386 server). With the next Ubuntu release end of april I will upgrade my installation and most probably will write down some more helpful notes on the installation process... :-) Andreas
Re: Biarch installation to amd64?
Steffen Grunewald writes: [...] > BTW, are there fai-kernels for amd64 available somewhere? (Not that I need > them, I've got to build my own kernel for Areca support anyway...) See http://www.informatik.uni-koeln.de/fai/download/amd64/ Andreas
Re: SATA support in install kernel for AMD64?
Allan Bailey writes: > Andreas Sindermann <[EMAIL PROTECTED]> wrote: > >Allan Bailey writes: > > > > > > Does anyone have an install kernel for 2.6.x that supports SATA drives? > > > > > > Or can someone point me to the sources/configs for building my own? > > > > > > >Take the latest AMD64 kernel on > >http://www.informatik.uni-koeln.de/fai/download/amd64/ > >I think it's a 2.6.10 kernel with correct working SATA support. > > Ok. I actually have tried to use this kernel. > > What I have is the vmlinuz-2.6.10-fai and System.map-2.6.10-fai files for > booting, > as well as the /lib/modules/2.6.10-fai copied to the nfsroot/. > > It still cannot find the SATA drive, and I believe it is because it is not > loading the > kernel module. Try to put the kernel module into the file /class/20-hwdetect.source. For me it's the sata_nv module as I'm using a NVIDIA chipset board. kernelmodules="rtc floppy usbkbd ide-disk ide-cd sata_nv" Andreas
Re: SATA support in install kernel for AMD64?
Allan Bailey writes: > > Does anyone have an install kernel for 2.6.x that supports SATA drives? > > Or can someone point me to the sources/configs for building my own? > Take the latest AMD64 kernel on http://www.informatik.uni-koeln.de/fai/download/amd64/ I think it's a 2.6.10 kernel with correct working SATA support. Hope this helps Andreas
Re: kernel installation problems
Mike Henson writes: > I see no kernel references in software.log. The only NEW packages it > lists are GRUB. Where is it supposed to copy the kernel from? I have added this to the package_config/DEFAULT file: PACKAGES aptitude-r I386 memtest86+ kernel-image-2.6-686 Depending on the architecture of your client to be installed you also could define (just as an example): PACKAGES aptitude-r AMD64 memtest86+ kernel-image-2.6.8-11-amd64-k8 Good luck Andreas
Re: kernel installation problems
Mike Henson writes: > Greetings, > > I have a 2.8.4 FAI install server that I cannot get to install a > kernel. My install client PXE boots, partitions the disk and appears > to work up until invoking task_configure where I get the following > logged to shell.log: > > Searching for GRUB installation directory ... found: /boot/grub . > Testing for an existing GRUB menu.list file... found: /boot/grub/menu.lst . > Searching for splash image... none found, skipping... > /bin/ls: /boot/vmlinuz-*: No such file or directory > Updating /boot/grub/menu.lst ... done > > If I ctrl-c before rebooting and check /tmp/target/boot grub is there > but the kernel is missing. I have also tried LILO and the kernel does > not install either. Can someone enlighten me on what might be > happening? I cannot find any references to kernel installation in the > docs. What is the software.log file saying about the installation of the kernel (should be located somewhere before the grub stuff above)? Andreas
Re: Ubuntu/Kubuntu and FAI
Holger Levsen writes: > > you can remount the /dev directory to /tmp/target/dev > > mount --bind /dev /tmp/target/dev Hi, thanks for the hint, this was very helpful. Using this in form of a hook (see below) and resolving some package issues which were resulting of some Debian/Ubuntu differences and using the correct installserver architecture (AMD64) to create the nfsroot environment (I used a Intel Xeon server before but the client PC to be installed are AMD64 machines, that's why I had problems with the kernel/grub stuff) lead to a successfull Ubuntu installation process. Here the hook to dynamically create the /dev directory in the nfsroot environment: [EMAIL PROTECTED]:/opt/fai/hooks$ cat /opt/fai/hooks/instsoft.FAIBASE #!/bin/sh # make /dev available in the chroot-environment mount --bind /dev $FAI_ROOT/dev && echo Remounted /dev to $FAI_ROOT/dev #HG: work around buggy kernel postinst if [ ! -s $target/etc/kernel-img.conf ] ; then cat > $target/etc/kernel-img.conf <<-EOF do_boot_enable = no do_initrd = Yes EOF fi Nevertheless, when booting the installed PC with a precompiled AMD64 Ubuntu kernel (linux-image-amd64-k8) leads to the following screen (little bit mixed up with the grub menu entries): Booting 'Ubuntu, kernel 2.6.10-5-amd64-k8 ' kernel direct mapping tables upto 101000 @ 8000-c000 root (hd0,0) Filesystem type is ext2fs, partition type 0x83 kernel /boot/vmlinuz-2.6.10-4-amd64-k8 root=/dev/sda1 ro quiet splash [Linux-bzImage, setup=0x1600, size=0x13e7d2] initrd /boot/initrd.img-2.6.10-5-amd64-k8 [Linux-initrd @ 0x3fbd2000, 0x40e000 bytes] savedefault boot . Decompressing Linux...done. Booting the kernel. audit(1123516210.213:0): initialized Starting Ubuntu... /sbin/init: 428: cannot create /dev/null: Read-only file system /sbin/init: 429: cannot open dev/console: No such file Kernel panic - not syncing: Attempting to kill init! At the moment I don't hav any idea where to start looking as the fai.log logfile doesn't look too bad: http://www.thp.Uni-Koeln.DE/~sinder/fai/09082005/fai.log The warnings during setting up the kernel package don't seem to be important as /initrd.img and /vmlinuz seem to be created. The error.log logfile contains the following lines (btw, what should be used instead of /proc/ide/hd?)? dmesg.log:Warning: /proc/ide/hd?/settings interface is obsolete, and will be rem oved soon! dmesg.log:ide2: Wait for ready failed before probe ! dmesg.log:ide3: Wait for ready failed before probe ! dmesg.log:ide4: Wait for ready failed before probe ! dmesg.log:ide5: Wait for ready failed before probe ! fai.log:Couldn't find any package whose name or description matched "jove" fai.log:dpkg: warning, architecture `amd64' not in remapping table fai.log:dpkg: warning, architecture `amd64' not in remapping table software.log:Couldn't find any package whose name or description matched "jove" software.log:dpkg: warning, architecture `amd64' not in remapping table software.log:dpkg: warning, architecture `amd64' not in remapping table The 'jove' error messages is due to a missing amd64 jove package but doesn't lead to any problems (I have removed that package from the package list). Has anybody solved this kernel panic problem before? Google is not too helpful in this special case... Thanks again Andreas
Re: Ubuntu/Kubuntu and FAI
Andreas Jobs writes: > On Wed, Aug 03, 2005 at 02:54:11PM +0200, Andreas Sindermann wrote: > > a) For some reason the grub package is not installed and the > > installation log files (esp. software.log) is not very informative > > why. Effect is that the system is left in an unbootable state at the > > grub> prompt. > >=20 > > b) When entering the needed information (kernel to be used etc) by > > hand at this prompt the system is booting but for whatever reasons in > > a very bad state. It seems that the /dev directory contains only very > > few devices like /dev/sg and things like that. Especially /dev/sd* are > > missing where the system has been installed on. > > I had a similar problem. In my case fai wasn't able to update the deb packa= > ges > during the installation. First my $FAINFSROOT/etc/apt/sources.list wasn't > pointing to the correct location and during my second try the > "host-to-be-installed" had no access to the mirror (wrong IP configuration;= > in > my case).=20 Fortunately, this seems to work without any problems. The packages can be accessed without any problems from the nfsroot environment as well as from the host-to-be-installed... But talking about /etc/fai/sources.list I have the impression that the base.tgz file created by make-fai-nfsroot is not really complete as the logfile http://www.thp.uni-koeln.de/~sinder/fai-setup.log says dpkg: dependency problems prevent configuration of ubuntu-base: ubuntu-base depends on at; however: Package at is not installed. ubuntu-base depends on fdutils; however: Package fdutils is not installed. ubuntu-base depends on info; however: Package info is not installed. ubuntu-base depends on logrotate; however: Package logrotate is not installed. ubuntu-base depends on mailx; however: Package mailx is not installed. ubuntu-base depends on ppp; however: Package ppp is not installed. ubuntu-base depends on pppconfig; however: Package pppconfig is not installed. ubuntu-base depends on pppoeconf; however: Package pppoeconf is not installed. dpkg: error processing ubuntu-base (--configure): dependency problems - leaving unconfigured Errors were encountered while processing: ubuntu-base W: Failure while configuring base packages. This will be attempted 5 times. Creating base.tgz For some reason the /etc/fai/make-fai-nfsroot.conf says FAI_DEBOOTSTRAP_OPTS="--exclude=pcmcia-cs,ppp,pppconfig,pppoe,pppoeconf,dhcp-client,exim4,exim4-base,exim4-config,exim4-daemon-light,mailx,at,fdutils,info,modconf,libident,logrotate,exim" so that explicitely the packages mentioned in the warning above ar excluded and I'm wondering if there is any specific reason for excluding these packages from base.tgz. Andreas
Ubuntu/Kubuntu and FAI
Hi, using Henning Sprang's How-To as a guideline I managed to successfully use the fai-setup/fai-make-nfsroot scripts for an Ubuntu installserver on a based on a Sarge system. Thank you for providing it, that was very helpful... Also the client PC to be installed is successfully booting over the network using the installserver. The install process is also performing. However, there are two problems (probably related to each other) which I currently cannot resolve, perhaps somebody on the list has seen them before: a) For some reason the grub package is not installed and the installation log files (esp. software.log) is not very informative why. Effect is that the system is left in an unbootable state at the grub> prompt. b) When entering the needed information (kernel to be used etc) by hand at this prompt the system is booting but for whatever reasons in a very bad state. It seems that the /dev directory contains only very few devices like /dev/sg and things like that. Especially /dev/sd* are missing where the system has been installed on. The reasons seem to lie in the base.tgz file created by the fai-make-nfsroot script. When checking with 'tar tvzf base.tgz ./dev' I can see that it contains only an empty /dev directory. On the other hand the udev package is installed which should populate the /dev directory. Currently when booting the client PC with the nfsroot environment a /dev with all needed devices is created (by udevd), but of course in /tmp/target/dev no devices are available so all $ROOTCMD referring this directory with /dev/sda (e.g. during grub-configuration) won't find the appropiate devices. Changing /etc/udev/udev.conf on the running nfsroot environment one could change /dev to /tmp/target/dev but this cannot work as the udevd process is already started in /etc/rcS/S04udev whereas the new root partition on the harddisk to be installed is mounted later (of course) as the corresponding devices need to be created first... I hope this is not too confusing... :-) Anyway, perhaps some people have already solved these problems and can give me some adivce... The logfile for the nfsroot-creation (where esp. the base.tgz archive is created) is available: http://www.thp.uni-koeln.de/~sinder/fai-setup.log And also the FAI-logfiles of the installation process itself are available: http://www.thp.uni-koeln.de/~sinder/software.log http://www.thp.uni-koeln.de/~sinder/FAI_CLASSES http://www.thp.uni-koeln.de/~sinder/additional.var http://www.thp.uni-koeln.de/~sinder/boot.log http://www.thp.uni-koeln.de/~sinder/cfengine.log http://www.thp.uni-koeln.de/~sinder/debconf.data http://www.thp.uni-koeln.de/~sinder/debconf.log http://www.thp.uni-koeln.de/~sinder/dhclient.log http://www.thp.uni-koeln.de/~sinder/disk_var.sh http://www.thp.uni-koeln.de/~sinder/dmesg.log http://www.thp.uni-koeln.de/~sinder/error.log http://www.thp.uni-koeln.de/~sinder/fai.log http://www.thp.uni-koeln.de/~sinder/fcopy.log http://www.thp.uni-koeln.de/~sinder/format.log http://www.thp.uni-koeln.de/~sinder/fstab http://www.thp.uni-koeln.de/~sinder/partition.sda http://www.thp.uni-koeln.de/~sinder/shell.log http://www.thp.uni-koeln.de/~sinder/status.log http://www.thp.uni-koeln.de/~sinder/syslog.log http://www.thp.uni-koeln.de/~sinder/updatebase.log http://www.thp.uni-koeln.de/~sinder/variables.sh Thanks in advance!! Andreas -- Dr. Andreas Sindermann fon: +49 (221) 470-4201 Institut fuer Theoretische Physikfax: +49 (221) 470-5159 Universitaet zu Koeln Zuelpicher Str. 77 mailto:[EMAIL PROTECTED] D-50937 Koeln, Germany http://www.thp.uni-koeln.de/~sinder
