This feature controls how long the Connector in Bridge mode waits before forcibly disconnecting the server side of the session after a client session ends.

Default: 50ms

When a client application opens a socket and connects to the connector, a second socket/connection to the server is created. The Connector in bridge mode then simply transfers data between these two sockets. 

When a client application brutally closes a connection without following the proper disconnection protocol, the server will not know about that disconnect until the connector closes the Connector<>server socket. If the connector closes the Connector<>server connection too soon after a client disconnect, there is a chance that the proper disconnection messages will be missed, if sent late. If the connector does not close this connector<>server connection, it would stay open indefinitely, using memory and resources that would otherwise be reclaimed.

The default of 50ms is very conservative and will fit most environments where client applications disconnect properly. When the volume of connections opened and never closed exceeds a certain level, the timeout must be tuned (lowered) to close idle connections faster, or the available resources will get used up.

Many times this situation is caused by health checks, especially from monitoring scripts and load balancers checking port liveness. Many of these check do not gracefully close the connection, triggering the need for tuning the Connector.

If connections still exist after the timeout interval, they are forced closed, and a warning will be printed in the connector logs.

C>S ended. S>Cstreaming did not finish within bridgeServerToClientForcedCloseTimeout=500(ms). Will be closed anyway !

This timeout is adjusted via the tpm property property=bridgeServerToClientForcedCloseTimeout in milliseconds.

For example, to change the delay to 100 milliseconds, the following property should be set in your /etc/tungsten/tungsten.ini file:

property=bridgeServerToClientForcedCloseTimeout=100