The scenario is: The situation is this:
What happens now is the following:
The strange thing is this "write" operation that gets "stuck".
Could a switch config or OS/application decide to close established connections that are not used after some delay? (Apparently set at 10 minutes). Why would this "intermediary" device cause a timeout of 18 seconds when we try to write in the "old connection"?
asked 10 Jul '13, 06:35
edited 10 Jul '13, 09:42
sounds like there is a firewall between your client and server. After some time of inactivity in the TCP connection (no packet sent in either direction), it will drop the session in its internal tables.
If the client sends a new packet, the firewall will drop that packet, as there is no 'open' session any longer. The client may/will re-try to send the packet for a certain amount of time (18 seconds in your case) and then give up and show the error message.
So, if there is a firewall involved, please ask the firewall admin to increase the "IDLE timeout" for those sessions. The firewall could be on the SQL server as well!
answered 10 Jul '13, 07:09
Kurt Knochner ♦
edited 10 Jul '13, 07:23