We have discovered a problem at some customer sites where connections build up which eventually leads to a high CPU state on the webNetwork relay server.
The problem comes from a bug in SSL , Java and Jetty (the internal web server used in webNetwork).
This can happen on version of webNetwork 6.2.1 thu 6.3.0 (current release)
The problem shows up predominately on larger systems where they have thousands and thousands of connections every day. Some connections in the OS are not getting closed completely and start to get back logged. The OS starts spending large amounts of time trying to close these connections leading to a high CPU state.
If you use a tool like TCPView (provided in the \stoneware\util\windows) folder you can view the number of TCP connections in the close_wait state. This is the tcp state that causes the problem.
There are two current work arounds.
1) Schedule periodic restarts of the webNetwork service on the relay. Depending on your usage this could be ever few days or more.
2) If you have a load balancer, you can off load your SSL to the load balancer and that will eliminate the problem.
Stoneware is currently working on a fix which includes an updated Java version along with an updated Jetty web server in the product. Unfortunately this is not a simple task and is taking many development hours and we are shooting for April to have a update available to customers.