Some system DCI parameters unavailable

Started by BillLortz, November 19, 2014, 08:12:00 PM

Previous topic - Next topic

BillLortz

I have a moderate sized NetXMS installation with NetXMS monitoring 80 individual remote sites plus the local master servers.   Each remote site is basically identical to the other sites.

At only one site, NetXMS is returning "Unsupported" for the "system.uptime" and "system.cpu.usage" DCI variables when it polls.   But, if I right click on the node, and use tools->info->agent->Supported Parameters, both of those parameters appear.   This behavior of declaring them unsupported is recent.

Other system parameters such as system.servicestate and process.count are working fine on that same node.

I've seen Unsupported occasionally appear in what I assume are race conditions on system startup.   In those cases, if I re-enable, restart the agent, and re-poll, it won't reflag them as unsupported.  I tried this for this node and it didn't solve the issue.   I restarted the server and it didn't help.   I used Windows Update to bring the server current and restarted and it didn't solve the issue.

The agent is version 1.2.9.        The server was just upgraded to 1.2.17 a couple days ago, but this problem was happening when the server was 1.2.9.   We haven't had time to upgrade the 80 agents yet.   79 of the 80 are working just fine.   The DCI configuration is managed by templates and not individually for each site, so I don't think there is a typo in the DCI configuration.    I've verified the agent configuration is identical to other sites - we have an internal update management system that pushes out the configuration files which are common to all locations.

I believe that this issue is likely to be a windows configuration corruption issue.    What components of Windows does the agent use to return uptime and cpu usage?   Is it WMI or something else?   At a former job, I occasionally saw WMI corruption on earlier versions of Windows, but don't want to rebuild WMI's database if it isn't involved.     The remote nodes are all running Windows 8.

Any suggestions would be appreciated.

Victor Kirhenshtein

Hi,

agent uses Windows performance counters for accessing CPU usage and uptime. Check that winperf subagent is loaded and do not report any errors on startup. It is possible that performance counters are corrupted or PDH is disabled altogether.

Best regards,
Victor

BillLortz

Victor:

The information was very useful.   

Thanks

Bill