A few times the past couple months, my NetBSD/i386 1.6.1_STABLE and then
1.6.2_STABLE system has locked up.

My mouse is unmovable in X and nothing is updated on the console. Usually
when this happens I log in remotely to it and kill blackbox. (And then
type blackbox in an xterm to continue my work.)

But lately the system has been unpingable.

Today, it happened when building qt3-libs.

Now, I see gcc's cc1plus is doing:

  PID USERNAME PRI NICE   SIZE   RES STATE      TIME   WCPU    CPU COMMAND
 1279 reed      64    0   111M   76M RUN        3:23 57.91% 57.91% cc1plus

and sometimes "biowait" state. I am surpised on the amount of memory gcc
needs.

And I dislike how my system is sometimes so sluggish :)

top tells me:

load averages:  1.93,  1.69,  1.24                                     13:11:07
55 processes:  1 runnable, 53 sleeping, 1 on processor
CPU states: 98.0% user,  0.0% nice,  0.5% system,  0.5% interrupt,  1.0% idle
Memory: 51M Act, 35M Inact, 212K Wired, 5532K Exec, 20M File, 18M Free
Swap: 193M Total, 166M Used, 27M Free

Any suggestions on sysctl's I should tweak for possibly improved
performance?

I am guessing that my system became unpingable because I ran out of
memory.

It seems like at one time I modified a kernel so it would sleep for awhile
when out of memory and then try again. (I will need to look to see if I
still have this patch.)

I guess I should just add some more memory (although for years I haven't
needed a lot for doing basically same work).

 Jeremy C. Reed

 	  	 	 open source, Unix, *BSD, Linux training
	  	 	 http://www.pugetsoundtechnology.com/