During most of the communications between the Harvest clients and the Harvest server side (broker and hservers), port 5101 is used.
However, when the user requests a check in or check out, a second port is employed for this purpose.
If the system has no firewalls, the ephemeral ports can be used on the network for this purpose, but if a firewall is in place between the Harvest server side and the clients, a block of ports must be opened in the firewall for this purpose and Harvest must be configured to use this predetermined block of ports for check in and check out.
This is configured as the direct connect port range.
The question is, how many ports should be configured for the port range?
Harvest Software Change Manager all versions and platforms
Harvest is currently built in such a way that an hserver process that connects to a port in the port range will maintain that connection, even after the checkin or check out is finished.
The connection to the port is only released when the hserver process is ended.
Therefore, you should configure your direct connect port range to have as many ports available as the number of hserver processes that might be running at any one time.
To find this exact number, look in the %CA_SCM_HOME%\HBroker.arg file (or $CA_SCM_HOME/HBroker.arg file on Linux/Unix) on your Harvest Server machine for an option named "-maxserver=". This will show the maximum number of hserver processes that can be executing at any point in time.
Therefore, the number you have specified for the "-maxserver=" option, should correspond to the number of ports in your direct connect port range.
More information on the Direct Connect Port Range can be found here: Configure Enterprise Communicator (PEC) - The Server Port Range
If an EFix is desired that would allow the hserver process to release its direct connect port after the check in or check out has completed, please contact support.