SpectroSERVER VNM threshold exceeded alarms
Article ID: 145585
Updated On:
Products
Issue/Introduction
The following alarms are seen especially when the SpectroSERVER process is restarted:
SPECTROSERVER IH CPU THRESHOLD EXCEEDED
SPECTROSERVER IH REALTIME THRESHOLD EXCEEDED
Environment
DxNetOps_Spectrum - 10.4.x/20.2.x/21.2.x
OS - Linux/Windows
Cause
SPECTROSERVER IH CPU THRESHOLD EXCEEDED
SYMPTOMS:
Less-than optimal SpectroSERVER performance, abnormally disconnected clients, and slow response times.
PROBABLE CAUSES:
An Inference Handler trigger has taken a longer-than-normal amount of CPU time, which might be contributing to the above symptoms.
It is also possible that this is a perfectly normal occurrence, and should be disregarded.
RECOMMENDED ACTIONS:
1) You can gauge the severity of the performance impact by looking at how long the IH trigger took (see the event). If it took 5010 ms and the threshold is 5000 ms, then there is minimal impact. If you are experiencing SpectroSERVER performance problems that seem to coincide with the timing of these events, please report this issue to SPECTRUM support. If you can correlate a specific user action to these events, that would be useful debugging information as well.
2) If you only occasionally see this alarm, it is probably just an anomaly and should be cleared. You can also increase the "ih_perf_cpu_threshold" setting in the $SPECROOT/SS/.vnmrc file to reduce the number of "false alarms". The default setting is 5000 milliseconds, and can be increased as needed to baseline performance in your particular SPECTRUM environment.
SPECTROSERVER IH REALTIME THRESHOLD EXCEEDED
SYMPTOMS:
Less-than optimal SpectroSERVER performance, abnormally disconnected clients, and slow response times.
PROBABLE CAUSES:
An Inference Handler trigger has taken a longer-than-normal amount of time, which might be contributing to the above symptoms.
It is also possible that this is a perfectly normal occurrence, and should be disregarded.
RECOMMENDED ACTIONS:
1) You can gauge the severity of the performance impact by looking at how long the IH trigger took (see the event). If it took 30010 ms and the threshold i
s 30000 ms, then there is minimal impact. If you are experiencing SpectroSERVER performance problems that seem to coincide with the timing of these events,
please report this issue to SPECTRUM support. If you can correlate a specific user action to these events, that would be useful debugging information as w
ell.
2) If you only occasionally see this alarm, it is probably just an anomaly and should be cleared. You can also increase the "ih_perf_realtime_threshold" setting in the $SPECROOT/SS/.vnmrc file to reduce the number of "false alarms". The default setting is 30000 milliseconds, and can be increased as needed to
baseline performance in your particular SPECTRUM environment.
Resolution
The event associated with the alarm will identify the Inference Handler where the issue occurred.
If these alarms are only seen during Spectro SERVER startup or only occasionally and there are no other performance issues noticed, these alarm can be ignored and cleared. These alarms do not clear automatically.
If these alarms are occurring frequently and other performance issues are noticed, please open a case with Spectrum Support for further analysis.
Feedback
