"gdm mysteriously murdered"

"gdm mysteriously murdered"

Post by Trygve H?geb » Tue, 28 Sep 1999 04:00:00



I'm in dire need of some good advice right now. When I booted my
system today ( RH 6.0, ATI Rage Pro - Mach 64-chip) everything went
well until the last bit when the X-server starts. The screen went all
black, and there were som occasional flickerings with a few seconds
in-between. I tried pressing ctrl-alt-F1 but was stuck with the black
screen for a while, until suddenly I was switched over to the
text-mode terminal. There were some repeated error messages on the
screen:

"According to /var/run/gdm.pid, gdm was already running ( three-digit
pid nr. ) but seems to have been *ed mysteriously."

The message was repeated about eight times, with only the pid-number
changeing ( growing larger ). Finally it read:

"INIT: Id "x" respawning too fast: disabled for 5 minutes."

Well., I checked with ps aux: no processes called "gdm", and none with
any of the listed pid-numbers.

5 minutes later, as promised, the flickering started again, went on
for a short while, and I was taken back to the terminal-window. Same
kind of error-messages this time. I tried reducing the color
resolution in XF86Config to get a lower refresh-rate, but it didn't
help, and this really shouldn't matter as nothing is changed here
since yesterday, when eveything worked. In fact nothing much at all
have been chaged  since then, but still this very annoyng problem has
occured.

No "red" messages during booting the system, but I notice that
shutting down the X-Font server under shutdown now gets a "failed".

Trying startx from the terminal generated the same situation.

Any advice now is very much appreciated. I'm desperate.

TIA

Trygve

 
 
 

"gdm mysteriously murdered"

Post by Reid Rivenburg » Tue, 28 Sep 1999 04:00:00


 >> Trygve H?geb?l writes:

 > I'm in dire need of some good advice right now. When I booted my
 > system today ( RH 6.0, ATI Rage Pro - Mach 64-chip) everything went
 > well until the last bit when the X-server starts. The screen went
 > all black, and there were som occasional flickerings with a few
 > seconds in-between. I tried pressing ctrl-alt-F1 but was stuck with
 > the black screen for a while, until suddenly I was switched over to
 > the text-mode terminal. There were some repeated error messages on
 > the screen:

 > "According to /var/run/gdm.pid, gdm was already running (
 > three-digit pid nr. ) but seems to have been *ed
 > mysteriously."

 > The message was repeated about eight times, with only the
 > pid-number changeing ( growing larger ). Finally it read:

 > "INIT: Id "x" respawning too fast: disabled for 5 minutes."

 > Well., I checked with ps aux: no processes called "gdm", and none
 > with any of the listed pid-numbers.

 > 5 minutes later, as promised, the flickering started again, went on
 > for a short while, and I was taken back to the
 > terminal-window. Same kind of error-messages this time. I tried
 > reducing the color resolution in XF86Config to get a lower
 > refresh-rate, but it didn't help, and this really shouldn't matter
 > as nothing is changed here since yesterday, when eveything
 > worked. In fact nothing much at all have been chaged since then,
 > but still this very annoyng problem has occured.

 > No "red" messages during booting the system, but I notice that
 > shutting down the X-Font server under shutdown now gets a "failed".

 > Trying startx from the terminal generated the same situation.

 > Any advice now is very much appreciated. I'm desperate.

Hi,

Perhaps your font server (xfs) somehow didn't get started?  You could
do a ps and see if it's listed.  If it isn't, you could try:

  /etc/rc.d/init.d/xfs start

And see if there are any problems.  You'll want to make sure that xfs
is being started at runlevel 5.  (There's a runlevel editor in the
control-panel.)

I think what you're seeing, by the way, is the X server continuously
trying to start, but failing.  I'm guessing from your "font server
failed" bit that that's the cause of it failing.

Good luck,
Reid

 
 
 

"gdm mysteriously murdered"

Post by Trygve H?geb » Tue, 28 Sep 1999 04:00:00



Quote:>Perhaps your font server (xfs) somehow didn't get started?  You could
>do a ps and see if it's listed.  If it isn't, you could try:

>  /etc/rc.d/init.d/xfs start

I did, and it didn't show up. Strange, because I noticed XFontserver
started with positive response during the boot-process.

OK, I tried starting it again, it says it does, but obviously doesn't.
I tried xfs with -status, and I get this message:

"xfs dead but subsys locked"

That seems like a not-too-good response, but maybe it brings me closer
to a solution. Problem is, it doesn't tell _me_ that much. More help
please.

Quote:>And see if there are any problems.  You'll want to make sure that xfs
>is being started at runlevel 5.  (There's a runlevel editor in the
>control-panel.)

well, the control-panel is a bit out of my reach now that I can't get
X up.

Quote:

>I think what you're seeing, by the way, is the X server continuously
>trying to start, but failing.  I'm guessing from your "font server
>failed" bit that that's the cause of it failing.

You might be right there. Thanx for the response.
Now I hope someone can lead me even closer to a happy ending.

Trygve

 
 
 

"gdm mysteriously murdered"

Post by Stan » Tue, 28 Sep 1999 04:00:00


Hello, I'm just a beginner, but I had a very similar message about two weeks
ago.  I couldn't start X so I had to try to run Xconfigurator again(or
XF86setup[whatever you use]).  I don't remember if I had to reinstall the
rpm's for XFree86 but you might be able to try reinstalling the rpm's if you
have the cd.
If somebody with more experience finds this advice to be totally off base,
please correct me
Good Luck,
Stan
 
 
 

"gdm mysteriously murdered"

Post by Reid Rivenburg » Tue, 28 Sep 1999 04:00:00


 >> Trygve H?geb?l writes:


 >> Perhaps your font server (xfs) somehow didn't get started?  You
 >> could do a ps and see if it's listed.  If it isn't, you could try:
 >>
 >> /etc/rc.d/init.d/xfs start

 > I did, and it didn't show up. Strange, because I noticed
 > XFontserver started with positive response during the boot-process.

 > OK, I tried starting it again, it says it does, but obviously
 > doesn't.  I tried xfs with -status, and I get this message:

 > "xfs dead but subsys locked"

 > That seems like a not-too-good response, but maybe it brings me
 > closer to a solution. Problem is, it doesn't tell _me_ that
 > much. More help please.

Just guessing some more here, but does it help to remove the file
"/var/lock/subsys/xfs" before trying to start xfs?  Before deleting
it, does it look like it's old?  That file is probably used to
determine if another xfs is already running.

 >> And see if there are any problems.  You'll want to make sure that
 >> xfs is being started at runlevel 5.  (There's a runlevel editor in
 >> the control-panel.)

 > well, the control-panel is a bit out of my reach now that I can't
 > get X up.

Heh.  That was more for if starting xfs by hand made things work
again.

 >> I think what you're seeing, by the way, is the X server
 >> continuously trying to start, but failing.  I'm guessing from your
 >> "font server failed" bit that that's the cause of it failing.

 > You might be right there. Thanx for the response.  Now I hope
 > someone can lead me even closer to a happy ending.

We're getting there!

Reid

 
 
 

"gdm mysteriously murdered"

Post by D. Stimit » Tue, 28 Sep 1999 04:00:00


> >> Trygve H?geb?l writes:



>  >> Perhaps your font server (xfs) somehow didn't get started?  You
>  >> could do a ps and see if it's listed.  If it isn't, you could try:

>  >> /etc/rc.d/init.d/xfs start

>  > I did, and it didn't show up. Strange, because I noticed
>  > XFontserver started with positive response during the boot-process.

>  > OK, I tried starting it again, it says it does, but obviously
>  > doesn't.  I tried xfs with -status, and I get this message:

>  > "xfs dead but subsys locked"

>  > That seems like a not-too-good response, but maybe it brings me
>  > closer to a solution. Problem is, it doesn't tell _me_ that
>  > much. More help please.

> Just guessing some more here, but does it help to remove the file
> "/var/lock/subsys/xfs" before trying to start xfs?  Before deleting
> it, does it look like it's old?  That file is probably used to
> determine if another xfs is already running.

>  >> And see if there are any problems.  You'll want to make sure that
>  >> xfs is being started at runlevel 5.  (There's a runlevel editor in
>  >> the control-panel.)

>  > well, the control-panel is a bit out of my reach now that I can't
>  > get X up.

> Heh.  That was more for if starting xfs by hand made things work
> again.

>  >> I think what you're seeing, by the way, is the X server
>  >> continuously trying to start, but failing.  I'm guessing from your
>  >> "font server failed" bit that that's the cause of it failing.

>  > You might be right there. Thanx for the response.  Now I hope
>  > someone can lead me even closer to a happy ending.

> We're getting there!

> Reid

I had this problem on RH 6 after moving /usr/X11R6/ to another
partition, and using a sym link at /usr/ to point to it. There are sym
links further down in this subdirectory which are relative links, and
they were not pointing to their correct locations anymore. The problem
was solved by pointing them back to the right spot with absolute paths
instead of relative.

If you are using tcsh shell, the following aliases can help to locate
things (put them in .tcshrc or .cshrc if no .tcshrc, and source
.tcshrc):
alias fn 'find . -name "\!*" -xdev -print'
alias fnln 'find . -name "\!*" -type l -xdev -print'
alias fnhd 'find . -name "\!*" -type f -xdev -print'
alias fndir 'find . -name "\!*" -type d -xdev -print'
alias fnexe 'find . -name "\!*" -type f -perm -0700 -xdev -print'

fn will find all files or directories of the name pattern (regular
expressions accepted); fnln will do similar to any symbolic link (this
is the useful one in this case); fnhd will find only hard file links;
fndir will find only directories; fnexe will find only files chmod for
executable.

cd to /usr/X11R6/, fnln *, and see if they all point to a
real/exisiting file or directory. gdm will claim to be mysteriously
*ed if they don't (in RH6 these will point to /etc/rc.d/ init
scripts and X config files in /etc/X11/).

 
 
 

"gdm mysteriously murdered"

Post by Trygve H?geb » Thu, 30 Sep 1999 04:00:00




Quote:>I had this problem on RH 6 after moving /usr/X11R6/ to another
>partition, and using a sym link at /usr/ to point to it. There are sym
>links further down in this subdirectory which are relative links, and
>they were not pointing to their correct locations anymore. The problem
>was solved by pointing them back to the right spot with absolute paths
>instead of relative.
[snip]
>cd to /usr/X11R6/, fnln *, and see if they all point to a
>real/exisiting file or directory. gdm will claim to be mysteriously
>*ed if they don't (in RH6 these will point to /etc/rc.d/ init
>scripts and X config files in /etc/X11/).

Well I checked it out, but could't find any invalid links.
Thanks anyway.

Trygve

 
 
 

"gdm mysteriously murdered"

Post by Trygve H?geb » Thu, 30 Sep 1999 04:00:00



>Hello, I'm just a beginner, but I had a very similar message about two weeks
>ago.  I couldn't start X so I had to try to run Xconfigurator again(or
>XF86setup[whatever you use]).  I don't remember if I had to reinstall the
>rpm's for XFree86 but you might be able to try reinstalling the rpm's if you
>have the cd.

I could be willing to try that too, but my problem is that at this
moment I can't execute anything that could take more than 4 1/2
minutes. - see another posting in this thread.

Quote:>If somebody with more experience finds this advice to be totally off base,
>please correct me
>Good Luck,

Nothing's off base right now. I'm ready to crawl up the walls, and
even a good luck- wishing is contributing towards me not smashing the
keyboard to little pieces

Trygve

 
 
 

"gdm mysteriously murdered"

Post by Trygve H?geb » Thu, 30 Sep 1999 04:00:00



Quote:> >> Perhaps your font server (xfs) somehow didn't get started?  You
> >> could do a ps and see if it's listed.  If it isn't, you could try:

> >> /etc/rc.d/init.d/xfs start

> > I did, and it didn't show up. Strange, because I noticed
> > XFontserver started with positive response during the boot-process.

> > OK, I tried starting it again, it says it does, but obviously
> > doesn't.  I tried xfs with -status, and I get this message:

> > "xfs dead but subsys locked"

>Just guessing some more here, but does it help to remove the file
>"/var/lock/subsys/xfs" before trying to start xfs?  Before deleting
>it, does it look like it's old?  That file is probably used to
>determine if another xfs is already running.

Looks like the lockfile works as intended. The xfs - file is generated
as the system boots, and the font server claims to start. Likewise it
disappears as I do the "/etc/rc.d/init.d/xfs stop" even if that gets a
[FAILED]-message.

Quote:> >> I think what you're seeing, by the way, is the X server
> >> continuously trying to start, but failing.  I'm guessing from your
> >> "font server failed" bit that that's the cause of it failing.

I tried reinstalling the xfm package, but that didn't change anything.

Another source of frustration now ( except from having my favourite OS
rendered practically useless ) is the constant attempts to start gdm
every 5 minutes. This interrupts whatever I'm doing and leaves me with
a blinking screen for half a minute. It even sends what halfway
composed commando I have at the prompt for execution which could be
potentially dangerous. Any way I could stop this?

Any other packages I could reinstall that might solve the mess?

Trygve

 
 
 

"gdm mysteriously murdered"

Post by V. Bala » Thu, 30 Sep 1999 04:00:00


I reported the same problem in another thread. It appears (from
/var/log/messages) to have something to do with the non-existence of a
/tmp/:0.XAuth file. I still haven't figured what is wrong with xauth,
but in my case, from the console, I can do the following to get gdm up:

remove the file /var/run/gdm.pid;
then gdm -nodaemon

Seems to work then. If any guru can tell me what went wrong with xauth,
and how I can fix it so that gdm starts from inittab I (and Trygve too:)
will be grateful.
--

Balaji         1 609.452.6516
SGI/GFDL Princeton University

 
 
 

"gdm mysteriously murdered"

Post by Reid Rivenburg » Thu, 30 Sep 1999 04:00:00


 >> Trygve H?geb?l writes:

I'm just about stumped.  Hopefully, the other poster's idea about the
xauth problem will help.  You could delete everything under /tmp
(including dot files) to see if anything under there is causing
problems.

 > Another source of frustration now ( except from having my favourite
 > OS rendered practically useless ) is the constant attempts to start
 > gdm every 5 minutes. This interrupts whatever I'm doing and leaves
 > me with a blinking screen for half a minute. It even sends what
 > halfway composed commando I have at the prompt for execution which
 > could be potentially dangerous. Any way I could stop this?

You could do a "telinit 3" to go into non-X mode.

 > Any other packages I could reinstall that might solve the mess?

I don't recall what version of XFree86 you're running.  You might
consider upgrading to the 3.3.5 rpm's on Red Hat's rawhide site.
Here's an example:

http://ftp.rge.com/pub/systems/linux/redhat/rawhide/i386/RedHat/RPMS/...

I've been using that version without problems.  (Of course, it may
have nothing to do with your problem....)

Good luck,
Reid

 
 
 

"gdm mysteriously murdered"

Post by Trygve H?geb? » Thu, 30 Sep 1999 04:00:00


Well, tonight I finally managed to start X again.
As V.Balaji suggested I tried to start gdm with the -nodaemon option but
that didn't seem to make any difference. Thanks anyway for the
suggestion. What finally did the trick was to delete the entire content
of /tmp. Big thanks to you Reid, for all your support.
I should of course have made a note of everything I deleted, but I'm
afraid  I wasn't in a very scientific frame of mind at the moment. What
I think I have noticed is that I now have a .XO-lock file in /tmp which
wasn't there when I emptied the directory.

Again, thank you all for your suggestions

 
 
 

"gdm mysteriously murdered"

Post by Reid Rivenburg » Thu, 30 Sep 1999 04:00:00


 >> l  writes:

 > Well, tonight I finally managed to start X again.  As V.Balaji
 > suggested I tried to start gdm with the -nodaemon option but that
 > didn't seem to make any difference. Thanks anyway for the
 > suggestion. What finally did the trick was to delete the entire
 > content of /tmp. Big thanks to you Reid, for all your support.  I
 > should of course have made a note of everything I deleted, but I'm
 > afraid I wasn't in a very scientific frame of mind at the
 > moment. What I think I have noticed is that I now have a .XO-lock
 > file in /tmp which wasn't there when I emptied the directory.

 > Again, thank you all for your suggestions

Great, glad it works now.  When in doubt, start deleting files! :)

Reid

 
 
 

"gdm mysteriously murdered"

Post by Anthony Joseph Meizeli » Fri, 01 Oct 1999 04:00:00


ahh..  yes...  I had a simular problem today...  x wouldn't start...
Ironically enough, it was the .xsession-errors file in my home directory
that was causing the space to be used up, and therefore caused xfs to be
unable to start.

oh well....
tony

Yesterday Reid Rivenburgh expounded thusly on Re: "gdm mysteriously *ed":

Quote:>  >> l  writes:

>  > Well, tonight I finally managed to start X again.  As V.Balaji
>  > suggested I tried to start gdm with the -nodaemon option but that
>  > didn't seem to make any difference. Thanks anyway for the
>  > suggestion. What finally did the trick was to delete the entire
>  > content of /tmp. Big thanks to you Reid, for all your support.  I
>  > should of course have made a note of everything I deleted, but I'm
>  > afraid I wasn't in a very scientific frame of mind at the
>  > moment. What I think I have noticed is that I now have a .XO-lock
>  > file in /tmp which wasn't there when I emptied the directory.

>  > Again, thank you all for your suggestions

> Great, glad it works now.  When in doubt, start deleting files! :)

> Reid

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

                         Visit my web page at:
                    http://www.veryComputer.com/~ameizeli/

TOO BAD YOU CAN'T BUY a voodoo globe so that you could make the earth spin
real fast and freak everybody out.
                -- Jack Handley, The New Mexican, 1988.

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

 
 
 

"gdm mysteriously murdered"

Post by root » Tue, 09 Nov 1999 04:00:00



> Well, tonight I finally managed to start X again.
> As V.Balaji suggested I tried to start gdm with the -nodaemon option but
> that didn't seem to make any difference. Thanks anyway for the
> suggestion. What finally did the trick was to delete the entire content
> of /tmp. Big thanks to you Reid, for all your support.
> I should of course have made a note of everything I deleted, but I'm
> afraid  I wasn't in a very scientific frame of mind at the moment. What
> I think I have noticed is that I now have a .XO-lock file in /tmp which
> wasn't there when I emptied the directory.

> Again, thank you all for your suggestions

I am having the same problem with my X server.  Lo and behold, when
I deleted the .X0-lock file in /tmp the X server actually booted up!  Now
I just have one more problem.  As any user, I can get to the logon screen.
However, only root can actually successfully login.  If a normal user
attempts a login, the computer totally freezes (not even ctrl-alt-del
works).  Any suggestions?

Paolo