Currently when push monitor receives a network error, it continues to retry its monitoring. If the server is in fact dead, this causes us to try to connect to it in two monitors simultaneously.

A more efficient implementation would be to kill the push monitor when it receives an error and immediately schedule a server check in the monitor. If the server is alive, the push monitor will be shortly reinstated. If the server is dead, it will only be attempted to be monitored by the main monitor.

The benefit of stopping push monitoring is to reduce resource consumption in the network layer.