> I am having unfortunate experiences with Linux.
> It seems that Netscape can cause it to lock up so tightly that I need to
> hit the power switch to get it back. This is NOT good.
>
> It seems that the OS should be able to detect a rogue application
> and take some corrective action -- just letting me log in on a different
> terminal to kill the offending process would be sufficient.
>
> Of course, there is the possibility that my configuration is incorrect...
>
> Can someone help out?
>
> Ted
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.rutgers.edu
> Please read the FAQ at http://www.tux.org/lkml/
I have this _exact_ same problem. I've noticed tihs especially on 2.2.x. It
seems that any application that goes astray can bring the system to a near
halt. The system doens't die (ping from other machines get responses) but its
*so* slow that it takes minutes for keyboard/mouse actions to have responses
in X/console.
I am currently running Redhat 6.0 and 2.2.9. I have 96 megs of ram and 96 of
swap. I am planning on adding 128 mroe megs of physical ram, however I don't
think this is giong to help. An application that goes astray can still eat up
all my memory and cause the system to be unusable. Last time this problem
happened, it was due to a perl script that had a function that recursed
endlessly. I had called this script in an xterm and the system had alreadyb
ecome _way_ to slow for me to do anything about by the time I noticed.
I walked to another machine to try to ping the the slowed down one, and sure
enough there were responses.Luckily enough, when I walked back to the
original machine, X had died ( and thus the child perl process died ) and
with the exception of several other daemons dying, the machine had returned
to a normal _usable_ state.
Is there anything I can do about this? Someway to have the kernel regulate on
non-root process that are taking up too much resources? Or some key combo to
throw me to a console with a high priority. Or something?
--- Nadeem
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/