[sf-lug] Pointers for how to track down source of system hang

Charles N Wyble charles at thewybles.com
Thu Jan 31 15:19:03 PST 2008


I can say unequivocally that I *am* using the proprietary nvidia 
driver.  This is to provide the 3D support that Compiz requires.  I can 
certainly try running the system without it installed for some period of 

[Charles N Wyble] 

I believe in a previous message you stated that you saw the hang even with
Compiz turned off? Is the hardware 32 or 64 bit and if 64bit are you running
32 or 64 bit version of Ubuntu? 

As to the gap time between faults, I can only give a rough estimate, as 
it's variable.  Some times the system hangs multiple time in one day 
with a gap time ranging from < an hour to > 4+ hours.  However, other 
times the system remains stable for several days.

[Charles N Wyble] 

This sounds like it may be a hardware issue especially the fact that it
doesn't respond to sysrq sequences. Although this may be due to the kernel
not supporting those. 

May I recommend the following page:



may be of interest.

Installing the network crash dump utility is highly recommended. 
Also see 

You may also find Oprofile useful.

I know lots of info. :) 

Linux crash reporting/handling isn't the best unfortunately. 

As of today I've gone back to running without Compiz but have re-enabled 
the audio (it's a necessity for some of the development work I'm doing) 
and thus far it's remained stable, but that's no indication of whether 
or not it will remain so.

[Charles N Wyble] 

Interesting. Are you running a Xen kernel by chance? 

As an aside, it has at times remained very stable over as long as a week 
or more with both Compiz running and the sound driver installed, which 
makes me think it's going to be something else.  The key for me is to be 
sure I'm checking all relevant logs and that I have all possible logging 
enabled.  I've stated the four (4) files I'm aware of, but would love 
feedback on other sources of information.

 [Charles N Wyble] 

Well the system logs are useful for a lot of things. However a kernel
crash (called an oops) requires a fair amount of special monitoring
tools. I wish all the distros came with it turned on and configured
out of the box. 

Sorry for the brain dump format of this e-mail. I hope to get the
Info out there and spark more discussion. 

More information about the sf-lug mailing list