> sh ~/bin/memused.sh                                                 
> Total = 38644K
> sjg:10215$ pstat -s
> Device      1K-blocks     Used    Avail Capacity  Type
> /dev/wd0b       32640    30804     1836    94%    Interleaved
> /dev/sd0b       24488    23652      836    97%    Interleaved
> /dev/sd1b           0  *** not available for swapping ***
> /dev/sd2b       32668  *** not available for swapping ***
> Total           57128    54456     2672    95%

I blew away the 13 perl processes that were running swatch, and now I
see: 

sjg:10438$ sh ~/bin/memused.sh; pstat -s
Total = 30464K
Device      1K-blocks     Used    Avail Capacity  Type
/dev/wd0b       32640    21108    11532    65%    Interleaved
/dev/sd0b       24488    14164    10324    58%    Interleaved
/dev/sd1b           0  *** not available for swapping ***
/dev/sd2b       32668  *** not available for swapping ***
Total           57128    35272    21856    62%

That's a discrepency of only 14% compared to previous 30%

More importantly, killing the perl processes did recover 19M of swap,
which seems reasonable.  

The perl script in question fork's and execs itself daily.
Certain perl operations are known to leak memory, so perhaps that's
all this is.

--sjg