thread leak in 1.0.2?

Hello --

We have an application that utilizes McKoi (1.0.2/JDK 1.4.2/W2K) as a standalone application, both McKoi and the client application are on separate JVM's on the same box.  

The client accesses McKoi via JDBC, we are seeing an alarming amount of thread bloat.  We are not doing any writes, only reads.  The behaviour consists of the thread pool increasing while never recycling the threads.

We do not see any exceptions being thrown in our logs.

Has anybody seen this before?  Any ideas?

Many thanks.