I'll describe what apparantly is happening, the question is if there are some salt/tuxedo time-outs in play:
A client application, on a Tandem platform , accesses our webservices published on Tuxedo by the use of SALT.
When a peticion in launched, we see a connection established
tcp 0 0 172.21.10.13:7777 10.241.250.11:2582 ESTABLISHED
that remaines established for 5 minutes, and than disconnects. Is this 5 minutes some default timeout for the GWWS process ?
PS we are in a testing environment so basically the client is sending a ECHO request every 15 minutes.
On the other hand, if the client increases the frequency of ECHO's to every 10 seconds,
apparantly the TCP/IP connection seems to be re-used and does not die after 5 minutes.
Although I'm not certain, I don't believe SALT times out client connections. Are the timeouts causing problems or is this just something you noticed? Many client SOAP stacks have built-in timeouts to try and allow a connection to be reused, but not tie up a connection for long idle periods.
Oracle Tuxedo Chief Architect
SALT will keep a network timeout if no activity from the client connection. The default value is 5 minutes. You can change it if needed by this link http://edocs.bea.com/salt/docs20/admin/admin.html#wp1038031.
Thanks Todd, Wayne,
this network timeout sure seems to be the guilty one!
I've changed it, lets see if it's fixed but sure looks like it.
The thing is that I was looking for GWWS tuning here :
but should have looked here