🚀 go-pugleaf

Here is what I am doing: my application is using IB for statistics of client
application requests.  Every time a client application makes a request, I
update that users record with a timestamp.  During my application load test
(50000 request an hour) everything seemed ok, until the next morning.  My
stats table has only about 30 records in it, and all other tables combined
have about 150 records, but the file itself has run up to 12.8M.  Is this
normal?  Basically, during my load test, I update the same 5 or 6 records
over and over again.

Also, I restarted IB to reclaim all the memory that was allocated (see my
previous post) and when it started up, one of the ibserver processes went
immediately to 100% cpu utilization, and allocated 65M of memory (ouch!).
It ran in this state for about 10 minutes, and eventually ate up 101M
(double ouch!) before it finally calmed back down.  Was it doing some kind
of maintenance/clean up?

I guess the major question is, am I doing something wrong, or should I
expect this kind of behavior? Anyone else using JDBC and seeing these kind
of anomalies?

These things kind of scare me since I cannot afford 101M of memory being
taken up (it has to run in parallel with my application).  Can I limit it in
some way?

I do not do commits in my code, since I am using JDBC and the Java reference
says the driver defaults to auto commit and you have to specifically turn it
off.