We ever doubted there's some problem with the cluster, sync issue or even the NIC card, the sync interface is connected to hub (does not support multicast)
The cluster seems normal, and we even tried to switch off one of the gateway and it still happened.
It seems there's some setting has been change on the kernal for the timeout value, we are going to take a look of the packets of the source machine and see if there's a reset packet by the firewall or something.
We have configured the tcptimeout value in Global properties, some improvements, but it could not solve.
Quote:
Originally Posted by northlandboy  Do you mean disconnect after an hour's idle time? Or shorter than that?
Is the cluster maybe flapping? Check your logs to see if you've got active/standby switching frequently. Related to that, check that they're synchronising properly. |