EPM 11.1.2 - Service unavailable errors using EPM System
By Marc Schumacher on Mar 07, 2012
Users experience intermittent Service Unavailable errors using different components of the Oracle EPM System. A blank page with the following error message shows up:
The service is unavailable.
Reviewing the application specific log files and testing the WebLogic application servers in the back end does not show any issues. As the client can still connect to the web server and the Java Application Server is working fine in the back end, it is clear that there is some issue in between web and application server. To track this down, logging for the WebLogic web server plug-in needs to be enabled. Setting up logging depends on the web server, so we need to distinguish between Internet Information Services (IIS) and Oracle HTTP Server (OHS).
Oracle HTTP Server (OHS)
To enable logging for the WebLogic web server plug-in, the following lines need to be added to each "LocationMatch" section:
Internet Information Services (IIS)
On Internet Information Services there is a configuration file for each instance of WebLogic. The following lines need to be added to all of these files:
The configuration files (iisproxy.ini) can be found here:
Once the error reoccurs check the log files for the following error message:
PROTOCOL_ERROR: Backend Server not responding
In addition, check the number of sockets used on the server using the "netstat -an" command. Also check the number of available ports on the server. If you are using Windows 2008 use the following command:
netsh int ipv4 show dynamicport tcp
On Windows 2003 you need to check the following Windows Registry setting: (Default 5000)
If the number of used dynamic ports (determined using netstat) is close the number of available dynamic ports, this range needs to be extended.
The number of available dynamic ports can be increased as follows:
netsh int ipv4 set dynamicport tcp start=1025 num=64508
A reboot is needed to activate these changes.