Bug#186283: netinst cd problems (march24th sarge image)
Please skip over my repeated nitpicking items marked with a * and anything
else you feel will be fixed too soon to bother commenting on or that isn't
worth bothering with (eg extra junk due to "low" priority questions").
Testing today with default lilo image (just hit enter) and the iso:
sarge-i386-netinst.iso 30-Apr-2003 16:15 82.8M
...
Default choose language takes me to the set priority screen. This doesn't
seem like a nice thing to do for english is a second language (esl)
people, or even worse, people who speek no english.
Selecting 4, low. Selected the default "11. English (USA) (default)".
* Still no English (Canada), will that go in later?
Default selections in order:
"Detect a keybord and select layout" -> "ps2" -> "1. [us] American
(default)" -> "Detect CDROM devices and mount the CD in /cdrom"
-----
/var/lib/dpkg/info/cdrom-detect.postinst: 61: /sbin/discover: not found
Loading kernel modules
/var/lib/dpkg/info/cdrom-detect.postinst: 160: /sbin/discover: not found
[100.0%]
Debian Installer Main Menu
A CDROM has been detected.
CD autodetection worked. A CDROM reader has been found with the following CD
inserted : CDNAME You can continue the install.
[Press enter to continue]
-----
I have several questions here. Where is discover? Why is the main menu
text shown? Is the CD called CDNAME? Why do I have to hit enter to
continue the install?
The answer to most might be because I selected low, and so I really don't
mind... These are just minor cosmetic issues anyway.
Default selections in order:
"Load installer modules" -> "cdrom-retriever" -> "q"
* Again I would prefer "s to stop" and "r for restart list" or something a
* little less ambiguous than "q to end, b for begin"
* Waiting...
Hmm, it'd be nice to see a list of the packages that have already been
retreved/installed rather than just the current package.
* I'm looking forward to deselecting packages as you described
Selected default: "Detect network hardware and load kernel drivers for it"
Why does it go to detect all hardware now? That option should be in the
menu first.
Selected default: "yes"
The following were detected:
ux Floppy Driver
Linux IDE-SCSI emulation layer
Hmm, why was the first prompt ">" and the second "[default = ]>"...
Probably because I selected null, but I don't think it'd be desirable to
set the same IRQ and IO settings for other devices.
Selected "6. ne" which is *not* the default. Hit enter at the ">" prompt.
-----
Note: /etc/modules.conf is more recent than /lib/modules/2.4.20-1-386/modules.de
p
modprobe: Can't locate module ne
Detecting HW and loading kernel modules
An error occured while running 'modprobe -v ne'
Something must have gone wrong.
[Press enter to continue]
-----
Switching to the second console and running "nano /lib/modules/2.4.20-1-386/modules.dep"
and then searching for ne.o reveils that ne.o isn't in the list? It wasn't
found.
The module ne.o is in the directory
/lib/modules/2.4.20-1-386/kernel/drivers/net
I think I could force the module to load using modprobe at the command
line, but I won't for now.
Switching back to the first console and hitting enter...
-----
ethdetect's postinst exited with status 256
installer[48]: Configuring 'ethdetect' failed with error code 256
installer[48]: Menu item 'ethdetect' failed.
installer[48]: Lowering debconf priority limit from 'low' to 'low'
Debian Installer Main Menu
-----
Selecting defaults: "Detect network hardware and load kernel drivers for
it" -> "yes"
Selecting now default: "ne"
Setting "IRQ=10 IO=0xcc00" (Default is null)
-----
Note: /etc/modules.conf is more recent than /lib/modules/2.4.20-1-386/modules.de
p
modprobe: Can't locate module ne
Detecting HW and loading kernel modules
An error occured while running 'modprobe -v ne'
Something must have gone wrong.
[Press enter to continue]
----
hitting enter, same errors as before after this and I get into a loop.
Switching to the second console, going to the dir with ne.o and typing
insmod ne gives me of course a list of unresolved symbols... Rather than
figuring out the correct dependancies for ne, I'll just move on.
Selecting "9. Detect hardware and load kernel drivers for it (full version)
Selecting "10. Partition a hard drive"
Selecting "1. /dev/ide/host0/bus0/target0/lun0/disc (Generic 1234/ 100MB)"
It warns me:
"Warning: invalid flag 0x0000 of partition table 4 will be corrected by w(rite)"
In bochs log file, I think at the same time it said:
-0786176432i[HD ] enabling LBA mode
selecting n->p->1->1->+10M n->p->2->21->203 t->1->82 t->2->83 a->2
10MB swap, ~90MB linux set as active
w
Selected "2" to finish.
Selecting "13. Configure and Mount Partitions"
-----
modprobe: Can't locate module xfs
modprobe: Can't locate modlue lvm-mod
modprobe: Can't locate module linear
Segmentation fault
partconf's postinst exited with status 35584
installer[48]: Configuring 'partconf' failed with error code 256
installer[48]: Menu item 'partconf failed.
installer[48]: Lowering debconf priority lmit from 'low' to 'low'
-----
And back to the main menu, just like last time.
On 25 Apr 2003, Martin [ISO-8859-1] Sjögren wrote:
> fre 2003-04-25 klockan 23.07 skrev Drew Scott Daniels:
> > Hmm, when I get to "Configure and Mount Partitions" it says:
> > Journalled Block Device driver loaded
> > LVM version 1.0.5+(22/07/2002) module loaded
> > md: md driver 0.90.0 MAX_MD_DEVS=256, MD_SB_DISKS=27
> > md: raid0 personality registered as nr 2
> > Segmentation fault
> > partconf's postinst exited with status 35584
> > installer[44]: Configuring 'partconf' failed with error code 256
> > installer[44]: Menu item 'partconf' failed.
> > [...]
> > And I'm back to the main menu. I had setup a 100MB image and had the
> > installer create one 10MB swap partition and then one 90MB linux
> > partition, both set with the proper types and with the linux partition
> > marked as bootable.
>
> Damn. It might be the raid code, it's untested. Do you have anything in
> your /proc/mdstats? Does /usr/lib/partconf/find-partitions segfault too?
>
[...]
> Older versions didn't probe for the raid modules and stuff, so it might
> be that...
>
# cd /tmp
# cat /proc/mdstats
cat: /proc/mdstats: No such file or directory
# /usr/lib/partconf/find-partitions
/dev/ide/host0/bus0/target0/lun0/part1 /host0/bus0/target0/lun0/part1 9
.8 MiB
/dev/ide/host0/bus0/target0/lun0/part2 |.@|.@.@.@.@.@.@.@.@.@.@.@.@
.@.@.@.@.@..@..@I.@I.@O.@O.@..@..@..@..@.@.@..@..@.@.@ 90.1 MiB
#
Note, some characters may not have pasted properly.
Perhaps this is a bochs glitch, but I see no more messages in bochs' log.
Going back to the installer...
-----
installer[4878]: Using default fdisk configuration!
Command (m for help): p
Disk /dev/ide/host0/bus0/target0/lun0/disc: 104 MB, 104767488 bytes
16 heads, 63 sectors/track, 203 cylinders
Units = cylinders of 1008 * 512 = 516096 bytes
Device Boot Start End Blocks
Id System
/dev/ide/host0/bus0/target0/lun0/part1 1 20 10048+
82 Linux swap
/dev/ide/host0/bus0/target0/lun0/part2 * 21 203 92232
83 Linux
-----
I wonder if there are any errors here. I have a 104,767,488 byte image as
hda. My bochs config contains the recomended line:
ata0-master: type=disk, path="debtest.img", cylinders=203, heads=16, spt=63
From this, can you deduce if the hard drive was incorrectly partitioned?
> > I'll try the install again another day. I'm not going to bother hacking
> > around this glitch. I tried looking up the error code in partconf, but it
> > seems that it's generated by some algorithm that would take some time for
> > me to decipher.
>
> The exit code 35584? It's not something I've invented. :)
>
>
It looked to be generated. I saw no hard coded value. How is it generated?
Not that It really maters to me if this bug is fixed.
Going back to the installer again...
Selecting "16. Mount/Unmount a partition (OLD)"
Selecting default "1. Create filesystem (OLD) (default)"
Selecting the one linux partition...
* "q to end", -end +stop or +quit
q->ext2->yes
So far so good, I'll write more tomorrow when I get back to this.
Drew Daniels
Reply to: