[CentOS] startx reboots my computer

Mon Nov 27 12:27:35 UTC 2006
Ted Miller <tmiller at broadcast.net>

>>Well, sometimes startx reboots my computer.  Other times is does nothing
>>except slow it to a halt for a while.  Sometimes is locks up.
>>
>>I am using init level 3.  At the command line, everything looks fine, and I
>>can do whatever commands I want.
>>
>>When I type startx, then things go down the tubes.
>>
>>What doesn't happen:
>>X never starts
>>No error messages get posted to dmesg
>>no log is generated in /var/log/xorg.0.log
>>
>>What DOES happen:
>>on a good run, I get the error message:
>>  /usr.../startx: fork: resource temporarily unavailable
>>on a bad run the computer either locks up or reboots, or I get a parade of
>>  notices that the comp is "out of memory, shutting down process xxxxxx".
>>running top -u tmiller -b>/tmp/top.log in another console generates a huge
>>  file.  After deleting the process entry lines (all end in startx, except
>>  the bash line), I get results like this:
>>
[snip]
>>************************************************************************
>>eventually peaking out at:
>>
>>top - 22:28:15 up 4 min,  3 users,  load average: 2.54, 0.74, 0.26
>>Tasks: 7696 total,   2 running, 7693 sleeping,   0 stopped,   0 zombie
>>Cpu(s): 10.5% us, 45.2% sy,  0.0% ni,  0.0% id, 43.8% wa,  0.0% hi, 0.6% si
>>Mem:   1024380k total,   896720k used,   127660k free,      264k buffers
>>Swap:  8385848k total,  1035964k used,  7349884k free,    10372k cached
>>
>>***************************************************************************
[snip]

>>This was done with the xorg.conf file deleted, so it isn't caused by that.
>>
>>Hardware:
>>AMD 3500+ x86_64 proc
>>1GB ram
>>8GB swap
>>3 x 160GB SATA drives
>>nvidia nForce4 chipset on Gigabyte MB
>>Gigabyte nvidia GeForce 6200 video card
>>firewall is separate ipcop box to ADSL
>>
>>Software:
>>Centos 4.4, fully updated
>>  started as a bare install (uncheck everything except KDE at software
>>  choice screen), then added only things I really want/need
>>vmware server (free version)
>>wine
>>
>>I had just spent a couple of days getting the install "tweaked" in
>>preparation to moving to it as my "workhorse" workstation environment.  I
>>had tweaked my dual monitor setup (a 21" landscape screen next to a 17"
>>portrait screen took a lot of false starts to get xorg.conf right), gotten
>>vmware working, and was still wondering where the fonts went for my wine
>>install.  It was working pretty well, and I was looking forward to using
>>it, and when I rebooted, I have this mess.

[snip]

>>"#rpm --validate --all>grep X" sqwauked about some missing *.so files, but
>>showed nothing corrupted.

Johnny Hughes was kind enough to suggest:
> While connected to the Internet, I would start out with the command:
> 
> yum groupinstall "X Window System" "KDE (K Desktop Environment)"

Tried that with no change.  I did find out how to avoid waiting the whole
process out, though.  Starting "#top -u me" in tty1, I went to tty2 and
logged in with my user name.  I noted the PID of my new bash shell.  I then
went to tty2 and typed "#startx".  No screens of stuff scrolling by in
tty2, so back to tty1, where it was obvious it was going wild.  Did a
"k -> 5471 -> 'Enter'", but that didn't do anything.  But "k -> 5471 ->
'Enter' -> 9 'Enter'" stopped the sorcerer's apprentice in his tracks.  A
great time saver, and keeps me from having to recover from a possible
lockup or reboot.

[snip]

> You didn't say if you installed the i386 distro of the x86_64 distro.

I have let most of that be dependency driven.  Anything I wanted to install
wanted to install a library, I let it go ahead.  Don't know if that was a
problem or not.  I also discovered the yum.log file (does it ever rotate?),
and could send you what I installed since the last successful "#startx" was
executed, and what I installed just before that, if that would be helpful.

> If you have 1GB ram and want this as a workstation, I would recommend
> the i386 distro as Shockwave (and other plugins for firefox),
> openoffice.org, etc. are only available as i386 programs.
>
> Operating i386 and x86_64 programs on an x86_64 machine can be hard and
> confusing (with duplicate packages of each arch required to be
> installed, etc.)

I am aware of the dual libraries, and the need to look at which distro is
involved, but I do not know if the yum/rpm dependencies always keep that
straight, or if I need to get more actively involved (always install the
i386 version of any library I have in x86_64?)

> That is where i would start, and go from there.

I started, so any more "go" instructions will be appreciated.  I am off to
work, so will not be back at this for 10 hours or more.

Ted Miller
Indiana