Re: Tux-T3-Still Crashes Under Stress

[Ingo Molnar <mingo elte hu>]

On Thu, 15 Mar 2001 RodneyLynThomas@netscape.net wrote:

> I am still trying most of your new versions as you release them. Tux
> T3 crashed my Linux system as before. At a seemingly random time on my
> 4-way machine, three processors went to zero while the other one
> pegged. Let me know if there is any info that I can provide to you.

please try the T6 patch, T3 had a memory leak, which would eat up RAM
slowly. (But probably your problem is not related to the memory leak,
because the leak has been introduced recently.)

such lockups/weird system states can be debugged in a number of standard
ways:

A) it would be most useful to get a 'readprofile' snapshot of the
incident, by logging in over the console of the system and doing:

	readprofile -r; sleep 30; readprofile > toingo

and also send me a 'gettuxconfig' done while the lockup is happening.  (i
already have your gettuxconfig while the system is 'normal'.)

or:

B) if the system has locked up to such a degree that logging in over the
console or running 'gettuxconfig' is not possible anymore, then please do
the following things:

1) run 'top' from over the network, and paste & send me the last screen it
   displayed. This should at least help us determine whether this is VM
   related or not.

2) attach a serial console to the system and do RightALT-PrtScn a few
   dozen times. In the TUX kernels this should display properties of the
   currently running process on random CPUs - which again will help us to
   fix this bug. The captured Call Traces will pinpoint

since i cannot reproduce this problem (yet), i need more info from you to
debug this.

	Ingo