Buggy Debian Etch netinst.
 

I made a "netinst" of Debian Etch a few days ago using the 2 iso images one after the other (long iso about 160MB and the shorter one) and the installations were very buggy. The latest netinst was made with the 160MB iso file (debian-40r0-i386-netinst.iso).

First I was able to login as root through the GUI, but not as a user.
I received the following messages:
"The greeter application seems to be crashing. Attempting to use another one."
"Cannot find or run the base session script. Running the Gnome failsafe session instead."
"This is Failsafe Gnome session ........ fix problems in your installation."

After trying to fix the problem for a day or two, that stopped working, I was only able to login from the command line as root. The display kept toggling between GUI and command line and then gave the message:
"The display server has been shut down about 6 times in the last 90 seconds. It is likely that something bad is going on. Waiting 2 minutes before trying again on display :0."

I have tried all the the suggestions I found on the net and only one worked, which was to stop GDM.

I then removed GDM and rebooted, this gave me a stable GUI as root but using the "switch user" option does not work. However, I can login as a user from the command line although I receive a warning "/etc/profile permission denied." and startx does not work from a user login.

So I have 2 problems that I like to fix although there are more than 2 steps that failed.

1) I was thinking of installing a replacement for GDM (XDM?).
Is there a better solution? Any hint or advice welcome.

2) It seems the users are not created properly. This is users that had their own directory in the previous installation (Sarge), in a /home partition. There was 4 users and not one can login.

I thought the previous files might interfere with the install, I renamed the directories and created new empty ones after the install, then copied the files (not the .files) to these empty directories but it did not help.

------------------------------------------------------------------------------

I have read somewhere about a "user profile" but I don't know how to make one nor if that would be the solution.

I have one more question, that is: should I expect other "breaks" in the install? There are already failures about hal, avahi, exim, linux printing, .dmrc, .gnome2/accels and possibly others to be discovered yet. If that is the case I am thinking about switching to ubuntu because I cannot reinstall my old Debian Sarge, it looks like one of the 14 CDs is faulty. I get a message to the effect that some applications failed to install.

Any suggestion welcome.

Thank you for your help.

This sounds like you're running out of disk space, and the operating system is failing as gracefully as it can. By default, there's a certain amount of extra space reserved for root, so that's why root is functional. How big are your partitions?

(This happened to me when I switched from Sarge to Etch, and my habitual 3gig OS partition wasn't QUITE big enough. It took me a while to figure out what the heck was going on!)

Thank you for your reply, I should have plenty of space, here is the result of "df" after the buggy install.

Filesystem 1K-blocks Used Available Use% Mounted on
/dev/hda1 520824 119476 374892 25% /
tmpfs 453396 0 453396 0% /lib/init/rw
udev 10240 96 10144 1% /dev
tmpfs 453396 0 453396 0% /dev/shm
/dev/hda3 1032056 431504 548112 45% /home
/dev/hda8 1032056 34092 945528 4% /opt
/dev/hda7 1032056 34140 945480 4% /tmp
/dev/hda5 5039560 1849900 2933656 39% /usr
/dev/hda6 508020 385579 96210 81% /usr/local
/dev/hda9 5039560 882744 3900812 19% /var
/dev/hda13 5039560 141260 4642296 3% /var/www
/dev/hda11 2064144 48496 1910768 3% /var/www/admin
/dev/hda10 2064144 33260 1926004 2% /var/www/cgi-bin
/dev/hda14 52817124 390236 49743872 1% /var/www/files
/dev/hda12 2064144 89492 1869772 5% /var/www/html

Yeah, the disk space looks good. I vaguely recall something like this happening to me one time when I copied files over and had accidentally changed some files in /home/username/ from username ownership to root ownership. See if manually chown'ing a user's /home directory (recursively) helps.

did you have separate partition and mount points for /home/<usernames> in the sarge install?

hmmm
 

etc/profile should be owned by root with permissions of 0644. If this isn't loaded by bash when you log in, your PATH variable may not contain directories you need.

Also, make sure the user home directories you moved or renamed are owned by the user, with permissions of 0755.

Thank you all for your answers.

I had stuffed permissions and ownership but correcting this and reinstalling made no difference.

I have checked every point raised and everything was correct. Having nothing to lose, I installed Ubuntu thinking I might learn something but it works well without any adjustment at all except for the Apache2 config files.

Because of the time it takes to debug that sort of problem, I will use Ubuntu and, when I have time, I will try another Etch "netinst" on a spare machine on which I can format the entire drive.

I will update this thread if I discover anything new.

So you did a netinstall over top of an old Sarge install ?

Why didn't you just edit your sources.list and do a dist-upgrade ?

Thank you for the advice, the reason is lack of experience.