I see those occasionally also. It is usually a prelude to an out of memory error and service crash.
We use the Keep-Alive feature to send messages from each syslog server to another. Enable Keep-alive in the Inputs section at the bottom of the menu tree.
On the sending server create a rule with an input source of 'Keepalive' and an action of 'Forward to another host'.
On the receiving server setup up a Timeout rule with the desired threshold and actions.