Linux or UNIX agent is not responding

On Linux or Solaris systems, a No current data available... error message may appear in RadarScan or you may notice missing performance statistics. These symptoms are indications of an up.time agent problem. Use the following steps to help troubleshoot and isolate the root cause of the issue.

NOTE: If you are troubleshooting an HP/UX agent, use the root user instead of the nobody user.

1. Login to the system on which the agent is running. Note that this may not work correctly if launching a telnet connection from a Windows system.

   # su - nobody
   $ telnet localhost 9998
   

   You should receive a connection with no prompt (although the prompt may appear if the session has hung). Type sadc_cpu and press Enter.

   You should see a line of numbers and be returned to the prompt.

   If this step works as described, the server is listening on the appropriate socket and the agent binary is working correctly. If this process does not work, inetd may not have been restarted or configured properly (make sure the agent service is defined in inetd.conf and services). Alternatively, permissions to the uptmagnt binary may not be correct (ensure the agent is owned and executable by the nobody user).

2. Test if CPU and disk statistics gathering is working on the agent by running the following commands (that the agent binary runs to collect CPU and disk information):

   # su - nobody
   $ /opt/SPYNuptm/bin/perfparse.sh cpu
   $ /opt/SPYNuptm/bin/perfparse.sh disk
   

   After each command you should see rows of numbers. If you receive any error messages, review the Before Contacting Support section of this article.

3. Test the monitoring station and the network by running the following commands on the monitoring station:

   # su - uptime 
   $ cd /opt/SPYNuptm/bin
   $ ./perfget -c AgentName -f /dev/stdout -q
   

   Replace AgentName with the name of the system that you are troubleshooting.

   If the command was successful then the agent is responding and there are no networking issues. If the command fails, port 9998 between the monitoring station and the agent may be blocked or the monitoring station may not be able to resolve the agent.

   Run the following command:

   $ ./perfget -c AgentName -f /dev/stdout -s
   

   If this command fails, the agent is not properly gathering CPU stats and explains why RadarScan is showing an error. This normally means that either the sar utility is not functioning properly or that the agent did not properly delete its lock file.

4. Check for a lock file on the agent (Solaris only) by running the following command:

   # rm -rf /tmp/*.lock*

Before Contacting Support

If you need further assistance troubleshooting this issue, run the following commands on the agent and email the output to [email protected].

# su - nobody
$ telnet localhost 9998
... get a connection and then type the next line and hit enter
sadc_cpu
$ grep 'uptmagnt' /etc/inetd.conf
$ grep 'uptmagnt' /etc/services
$ ls -l /opt/SPYNuptm/bin
$ ls -l /tmp/*lock*
$ ksh -x /opt/SPYNuptm/bin/perfparse.sh cpu
$ ksh -x /opt/SPYNuptm/bin/perfparse.sh disk