You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
When CheckSystem=enabled is set in nsclient.ini, the system will encounter noticeable performance issues for a few minutes every 20 minutes.
Steps to reproduce: Start NSCP. No external monitoring or remote access is necessary to reproduce
.\nscp.exe test
Expected Behavior
NSCP should not cause system-wide performance issues.
Actual Behavior
The performance issues are most noticed in response times for IIS web pages (identified by the timeTaken field in the IIS logs).
These performance issues can be exactly correlated with ~3% CPU utilization by WmiPrvSe.exe, and this line in the Microsoft-Windows-WMI-Activity/Operational log:
WMIProv provider started with result code 0x0. HostProcess = wmiprvse.exe; ProcessID = 3296; ProviderPath = %systemroot%\system32\wbem\wmiprov.dll
Having enabled WMI trace logging (https://docs.microsoft.com/en-us/windows/desktop/WmiSdk/tracing-wmi-activity) via Wevtutil.exe sl Microsoft-Windows-WMI-Activity/Trace /e:true it does not appear that a large number of WMI queries are being executed during these times. The only notable queries in the trace logs pertain to network interfaces (executed by the ClientProcessId owned by nscp.exe)
According to that thread, it looks like the PDH queries should not cause this lag, but only WMI queries will. The NSCP source code seems to use PDH for most counters, but the networking counters still appear to use WMI.
I don't see an easy way to disable network counters from the CheckSystem module, but this would be my first guess as to diagnosing this issue.
Issue and Steps to Reproduce
When
CheckSystem=enabled
is set innsclient.ini
, the system will encounter noticeable performance issues for a few minutes every 20 minutes.Steps to reproduce: Start NSCP. No external monitoring or remote access is necessary to reproduce
Expected Behavior
NSCP should not cause system-wide performance issues.
Actual Behavior
The performance issues are most noticed in response times for IIS web pages (identified by the
timeTaken
field in the IIS logs).These performance issues can be exactly correlated with ~3% CPU utilization by
WmiPrvSe.exe
, and this line in theMicrosoft-Windows-WMI-Activity/Operational
log:Having enabled WMI trace logging (https://docs.microsoft.com/en-us/windows/desktop/WmiSdk/tracing-wmi-activity) via
Wevtutil.exe sl Microsoft-Windows-WMI-Activity/Trace /e:true
it does not appear that a large number of WMI queries are being executed during these times. The only notable queries in the trace logs pertain to network interfaces (executed by theClientProcessId
owned bynscp.exe
)This appears to be the same scenarios as described here: prometheus-community/windows_exporter#89
According to that thread, it looks like the PDH queries should not cause this lag, but only WMI queries will. The NSCP source code seems to use PDH for most counters, but the networking counters still appear to use WMI.
I don't see an easy way to disable network counters from the
CheckSystem
module, but this would be my first guess as to diagnosing this issue.Details
Additional Details
NSClient++ log:
The text was updated successfully, but these errors were encountered: