Resolution:
Description:
=========
After some routed EMS servers were brought down, messages routing to a central EMS server was delayed.

Environment:
==========
EMS version prior to 6.0

Resolution:
========
In following routing structure:

A central EMS server and a number of satellite EMS servers are configured with active-active routes. Messages are published to all satellite EMS instances on global destinations. The message consumers are running against the central EMS server meaning messages were routed from all "satellite" EMS servers to the central EMS server. When some of the satellite EMS servers were brought down it was noticed that the messages were piling up in the global destinations in the remaining satellite EMS servers.

The reason for this is in EMS server versions prior to 6.0 the same thread was used to handle  route reconnection and acknowledgement of routed messages. If the central server is blocked trying to reconnect to one of its routes then this explains why  acknowledgments are not being sent on a regular basis.

Since EMS 6.0, the route connects and route acknowledgments were put into two separate threads. This way if the route connection gets blocked then the route acknowledgments will not be held up.