Bug#253032: alpha test rc1 installation failed on various reasons

2004-06-07 Thread maks attems
hello vorlon,

thanks for your detailed answer!

On Sun, 06 Jun 2004, Steve Langasek wrote:

 On Sun, Jun 06, 2004 at 08:38:53PM +0200, maximilian attems wrote:
  1) detecting network loads module DAC960
 consoles without seeing any grave message, i assume kernel panic
..
 
 The detecting network hardware is a bit of a misnomer, actually, as it
 does detect other hardware and load drivers if they're available.
 
 You say you were able to change consoles after this message appeared,
 but the installer was dead?  Did the system freeze completely, or were
 you still able to switch consoles?

yes i was able to switch consoles, but couldn't open the vc2,
no error message on the other, d-i did not respond to any other
keystroke.

this DAC960P controller is a beast, i'm not shure if that driver is smp safe 
i saw some work on this driver lately
http://linux.bkbits.net:8080/linux-2.4/related/drivers/block/DAC960.c?nav=index.html|src/|src/drivers|src/drivers/block
on a similar machine i remember having worked with a 2.2.20 up,
as last known good from the old debian installer,
unfortunatly its kernel panics on boot (tried with different floppies)
so i want basicaly to get sarge up on that mini scsi disc to be
able later too search for an kernel were that raid actually works.

 
  4) boot loader menu entry jumps back to base-install instaed of calling
 swriteboot witch correct arguments:
..
 The bootloader option depends on you first having a successfully
 installed system.  It is possible to kludge this, but the correct fix is
 to get a working base installer onto the CD as discussed above.

okay 

  5) boot fails on kernel panic unable to mount root=/dev/sda1
.. 
 Do you know for sure that /dev/sda1 is the correct root partition for
 your system?  What filesystem did you use?  This may also be an artifact
 of the above errors (e.g., failure to build a correct initrd for the
 system).

i used ext2 on a root partition that took 520 mb from the end of the drive
that has 540 mb. that was the only partition.  20 mb in front were left 
untouched for aboot.
i assume your initrd guess is valid.

i may be able to retest alpha images this friday on that machine.
tc1 images may be already better then?

best regards
maks



signature.asc
Description: Digital signature


Bug#253032: alpha test rc1 installation failed on various reasons

2004-06-06 Thread maximilian attems
Package: installation-reports

INSTALL REPORT

Debian-installer-version: alpha test rc1 from may 30
uname -a: 
Date: 6. june 2004 ~16h
Method: sarge-alpha-businesscard.iso, boot from scsi cdrom,
target ont scsi hard disc.

Machine: Digital Server 4000 - family Rawhide
Processor: 4
Memory: ~ 800 Mb
Root Device: SCSI Maverick hard disc
Root Size/partition table: 520mb from the end of a ~540 mb scsi drive
   
Output of lspci and lspci -n:


Base System Installation Checklist:
[O] = OK, [E] = Error (please elaborate below), [ ] = didn't try it

Initial boot worked:[O]
Configure network HW:   [E] 1)
Config network: [O/E] 2)
Detect CD:  [O]
Load installer modules: [O]
Detect hard drives: [O]
Partition hard drives:  [O]
Create file systems:[O] 
Mount partitions:   [O]
Install base system:[E] 3)
Install boot loader:[E] 4)
Reboot: [E] 5)

Comments/Problems:

this is the same machine as #242659 *)
unfortunately ended again without a working os on it.

1) detecting network loads module DAC960
   don't know how that raid controller helps for networking?
   anyway everything dead, last thing i could do was to change
   consoles without seeing any grave message, i assume kernel panic
   - please blacklist that controller for network detection.
   rebooted with DEBCONF_PRIORITY=medium to not load that module

2) another error while installing the base system on a clean target:
   couldn't download dpkg
   the dhcpd daemon of that ass slow network is configured with
   renewal in 150 seconds so that may have interferred.
   on the next run i had more luck regarding network.

3) install the base system fails after 70% because of apt-utils
   vc3 shows:
   Errors were encountered while processing: apt-utils
   umount: /target/dev/pts: No such file or directory
   umount: /target/dev/shm: No such file or directory
   umount: /target/proc/bus/usb: Invalid argument

   changed to vc2, chrooted into /target, cd /var/cache/apt/archives,
   dpkg -i *deb
   apt-get install aboot libdb4.2 kernel-image-2.4.26-generic

4) boot loader menu entry jumps back to base-install instaed of calling
   swriteboot witch correct arguments:
   tried to install aboot from debian main-menu,
   that asked if i want to proceed on an unclean base install,
   which i thought to have cured above somehow, but arrrgh
   it brings me back to 3)
   redone aboves steps then outside of chroot:
   sbin/swriteboot -c1 dev/sda boot/bootlx
   (hmm it's a wild guess but seems to work somehow, 
stumbeled a bit but /dev/sda seemed not to exist only above /target)
   edited /etc/aboot.conf to reflect the root partition on 1 partition.
   
5) boot fails on kernel panic unable to mount root=/dev/sda1
   when entering aboot, a quick 'l' lists aboot.conf entries
   choosing 1 boots up the kernel, but
   Please append a correct root= boot option
   kernel panic: vfs unable to mount root fs on 08.01
   also treid with bootargs 
   devfs=mount,dall root=/dev/scsi/host0/bus/0/target0/lun0/partition1
   but same result.


 hope that helps for further alpha development.
 best regards
 maks


*) i will close that bug report,
   as scsi disc were recognised after boot (no need to mount) 
   those base system trouble were cured,
   and the menu entry of aboot showed up.


signature.asc
Description: Digital signature


Bug#253032: alpha test rc1 installation failed on various reasons

2004-06-06 Thread Joey Hess
maximilian attems wrote:
 Debian-installer-version: alpha test rc1 from may 30

We have never release a version of the installer called rc1. Please be
more careful about your version numbers and tell us from what url you
downloaded the image.

-- 
see shy jo


signature.asc
Description: Digital signature


Re: Bug#253032: alpha test rc1 installation failed on various reasons

2004-06-06 Thread maks attems
On Sun, 06 Jun 2004, Joey Hess wrote:

 maximilian attems wrote:
  Debian-installer-version: alpha test rc1 from may 30
 
 We have never release a version of the installer called rc1. Please be
 more careful about your version numbers and tell us from what url you
 downloaded the image.

s/rc1/tc1/
test canditate 1, url:
http://cdimage.debian.org/pub/cdimage-testing/sarge_d-i/alpha/tc1/

a++ maks



signature.asc
Description: Digital signature


Bug#253032: alpha test rc1 installation failed on various reasons

2004-06-06 Thread Steve Langasek
On Sun, Jun 06, 2004 at 08:38:53PM +0200, maximilian attems wrote:
 1) detecting network loads module DAC960
don't know how that raid controller helps for networking?
anyway everything dead, last thing i could do was to change
consoles without seeing any grave message, i assume kernel panic
- please blacklist that controller for network detection.
rebooted with DEBCONF_PRIORITY=medium to not load that module

The detecting network hardware is a bit of a misnomer, actually, as it
does detect other hardware and load drivers if they're available.

You say you were able to change consoles after this message appeared,
but the installer was dead?  Did the system freeze completely, or were
you still able to switch consoles?

 3) install the base system fails after 70% because of apt-utils
vc3 shows:
Errors were encountered while processing: apt-utils
umount: /target/dev/pts: No such file or directory
umount: /target/dev/shm: No such file or directory
umount: /target/proc/bus/usb: Invalid argument

Is there anything above the Errors were encountered while processing:
line?  This error may be far back in the syslog output.  Is this related
to your installation of libdb4.2, mentioned below?

Ah, it seems this was the problem that prompted the debootstrap upgrade,
which was slow to arrive on alpha.  And due to personal time constraints
and autobuilder problems, the current debian-installer package has still
not been built on alpha, so this fix has not propagated into the sarge
d-i images.

... and building the current debian-installer package depends on having
a current version of the anna udeb in the archive for alpha; so it looks
like it may yet be a day or two before we can get usable sarge tc1
images into the archive for this arch. :/

Ok, I've gotten anna built and uploaded, and I'm short-circuiting to get
debian-installer built -- it's now uploaded as well.  Joey, can we get
the sarge CD build immediately following next dinstall declared tc1 for
alpha?

 4) boot loader menu entry jumps back to base-install instaed of calling
swriteboot witch correct arguments:
tried to install aboot from debian main-menu,
that asked if i want to proceed on an unclean base install,
which i thought to have cured above somehow, but arrrgh
it brings me back to 3)
redone aboves steps then outside of chroot:
sbin/swriteboot -c1 dev/sda boot/bootlx
(hmm it's a wild guess but seems to work somehow, 
 stumbeled a bit but /dev/sda seemed not to exist only above /target)
edited /etc/aboot.conf to reflect the root partition on 1 partition.

The bootloader option depends on you first having a successfully
installed system.  It is possible to kludge this, but the correct fix is
to get a working base installer onto the CD as discussed above.

 5) boot fails on kernel panic unable to mount root=/dev/sda1
when entering aboot, a quick 'l' lists aboot.conf entries
choosing 1 boots up the kernel, but
Please append a correct root= boot option
kernel panic: vfs unable to mount root fs on 08.01
also treid with bootargs 
devfs=mount,dall root=/dev/scsi/host0/bus/0/target0/lun0/partition1
but same result.

Do you know for sure that /dev/sda1 is the correct root partition for
your system?  What filesystem did you use?  This may also be an artifact
of the above errors (e.g., failure to build a correct initrd for the
system).

Thanks for the feedback, max.  I'm sorry the installer couldn't be in
better shape for your testing, but I'm confident we'll have it working
again soon.

-- 
Steve Langasek
postmodern programmer


signature.asc
Description: Digital signature