Denis Tonn wrote to Will Honea on 04-27-1998
Thanks, Denis.  That confirmed most of what I figured was going on,
altho the 'preswap' had not occurred to me.  The one thing I was really
unsure about was whether the system ever did a true 'garbage
collection' to compact the swap file - you seemed to indicate that this
was a low priority task, which makes perfectly good sense to me.
I was really looking for some more background in answer to the
persistent complaints about the swap file not shrinking - as if it
should instantly reflect a decrease in memory demand.  The TCP/IP/LAN
leak seems to persist in the latest fixes under some unusual but
repeatable conditions - the author of InJoy found a case last week
where afinet.sys grabbed an additional 16k every time a specific
sequence was followed - but it's not a big worry.  I'm sure he reported
it.  I know that the TCP/IP stack got a really thorough going over in
the last few months with lots of memory leak APARs reported and fixed. 
With up times of 30-60 days, I have no real problems but I appreciate
your confirmation of the reasonableness of a very slow and possibly
incomplete shrinkage of the swap file after an overcommit caused it to
grow.  Thanks. 
Will Honea 
--- Maximus/2 2.02
---------------