Re: TCP packet out of state Most likely the problem is related to the database opening connections, then not using them for over an hour (the default TCP service timeout). You can increase the timeout for certain services - check the advanced button on the service definition. You could try putting it up to 2 hours. Another thing to try is changing the tcp_keepalive_interval on the hosts, using ndd. (This assumes HP-UX or Solaris, there will be some similar way of changing it on other OSes). Default keepalive_interval is 2 hours, but Check Point's default timeout is 1 hour. If an OS has a session with no traffic up for two hours, it sends a tcp keepalive packet. If it gets no response, it closes the session. If you set the keep alive interval to under an hour, then Check Point should register the session as still open, and reset the timer. Thirdly, I have heard of some issues with the sqlnet service defined in Check Point. You might want to do some research into that, I think there are some things you can do with that. Have a look in SecureKnowledge. |