Large scale htdig

Greetings everyone-

I saw that you are using Apache, MHonArc and htdig, which is exactly 
what I am working on setting up.  I am working on implementing htdig 
for over 100 mailing lists, and I am having some issues.  Currently 
there are over 520,000 documents for htdig to index, and we have been 
hitting a number of problems.  At this point it takes days to index all 
the records and we are hitting memory allocation errors as well (so 
far, we haven't even had a successful first index!).  I was wondering 
if anyone could offer some words of advice on setting up htdig on such 
a large scale.

Any help would be greatly appreciated!

-=Aaron