Troubleshooting › SNMP Queries Using Investigations › Situation 1-- Server Incident › Analyze the SNMP Performance Data › Server Refused or Unresponsive Sessions Incidents Analysis

Server Refused or Unresponsive Sessions Incidents Analysis

If you see unresponsive or refused sessions (server running out of available threads) along with expired sessions it indicates that resources are not being released by the application.

For server refused or unresponsive session incidents it indicates that the server is not responding to requests for new sessions in a timely manner.

Top CPU Processes or Memory Processes

In analyzing the CPU resources, ensure that the processes responsible for hosting the application are running. For example, in the process list for a SQL Server sqlserver.exe should be listed in the processes for CPU or memory. The process should be running and not consuming all the CPU. If the CPU is high, near 70 or 100 percent, identify the process that is consuming the CPU. If this process is non-critical, stop the process or restart it. The process might be locked or in a poor state that is unnecessarily taking processing power away from serving critical applications.

If the memory utilization is over 80 percent, identify the process that is consuming the memory. If this process is not critical, stop the process or restart it. The process might be locked or in a poor state, unnecessarily taking critical memory resources away from the critical processes. If the process that is consuming the memory is critical, restarting the process should clear its memory cache and let it restart normally.

Disk Space

If the expected processes are not listed, they potentially might have failed because of disk space limitations. Check the drives to ensure that the server has enough disk space to launch the applications. Then log in to the server to see if the processes are running and are able to start. Restarting the services or processes to restore connectivity might be required.

Interface Statistics

If the processes are running and operating normally, then it is possible the NICs are experiencing issues making the server appear slower or unavailable. Check the interface statistics. Ensure that all the interfaces are listed properly and are connected.

When the SNMP data looks good, you can initiate an operational investigation, including:

• Review the number of active sessions, the Time_Wait, and other TCP/IP parameters on the server to ensure they are not exhausting TCP port availability.
• Compare Server Connection Time with other applications running between remote office and server to see if the increase is across the board.