bind 9.2.4 caching cpu level increases flush helps

[dannyz at belgonet.com]

hello;

i got a box running 5 bind instances with 3 gig ram configured in it;

it is only doing cache name server but when the process runs for one
or two days it grows to about 250M in memory space for each instance.
... it is doing the requests fine, but when running for about 5  days
the non-active cpu level goes up to 7 - 10 % instead of 0,5% , the
only solution to get the cpu level back to the normal level is to do a

rndc flush 

to flush the cache or to reload the process.  Does anybody know why
this happens and how this could be solved?  I already tried to tune it
by changing the maitenance intervals for the named server but this
doesn't seem to be a solution.  Also after a week or so the memory
used is almost 3 gis of ram.  Is there a way to FREE the memory that
is being allocated during the named process after we did a flush cache
?

thnxs