I would tend to think the issue is in the APS/Essbase communication, and not between the client and the APS server, due to the Essbase error code.
We see these messages on Windows machines due to limitations in the TCPIP stack default socket configuration. The Windows default configuration sets the 'max' socket number to 5000 (instead of a possible 65530 something) with a socket timeout of 240 seconds. Any TCPIP activity that uses a socket thus has a smaller pool of sockets to choose from and, to compound matters, holds onto the socket for 4 minutes before releasing it. I did a blog post on the details: http://timtows-hyperion-blog.blogspot.com/2007/12/essbase-api-error-fix-geeky.html.
It didn't used to be a problem when computers were slower and couldn't use that many sockets but it is becoming a big deal now. We see similar issues all the time with the Outline Extractor running against big outlines (and making many, many calls to the server in s a short period of time). I don’t know exactly what the Smart View servlet is doing internally but it is conceivable that it is doing many, many calls to Essbase in quick succession.
All of this being said, I don’t know if there is a similar issue with TCPIP configuration on AIX. I may be worth a look.
Tim Tow
Oracle ACE Director
Applied OLAP, Inc
|