RPC server for UDP requests

[Subir Singh]

Hi ,

I am facing a very strange problem .I have a RPC Server listening to UDP requests on HPUX11.
If this server process dies or is brought down due to any reason and restarted with the
clients still connected to the earlier process, then the new process is heavily affected by the previous set of clients and this causes the size of new process to increase exponentially.

This problem is not replicated for TCP . The only work around I have at the moment is to terminate the client processes. Is there any other solution ?

Thanks in Advance,
Subir

[Ron Kinner]

You might try setting
rpc_mgmt_set_com_timeout to a lower value. See if that helps. TCP ignores this value unless it's set to 10.

Ron

[Subir Singh]

Ron ,

The default timeout for UDP is 5 . I even tried 0 and 3 but it still does not have an effect.
I tried using "rpc_mgmt_is_server_listening"
to determine if the process has been restarted, but it works only if the server is completely down . The moment it comes back up ,it returns a TRUE value .

My aim is to determine somehow if the server process has been restarted without bringing any change in the existing server code.

Can you suggest a method ?

Thanks,
Subir