This article explains why the event message RTPERF_CPU_THRESHOLD_EXCEEDED does not appear in the syslog every time (second) a PFE CPU excessive load event happens. For information on how to troubleshoot the syslog message, RTPERF_CPU_THRESHOLD_EXCEEDED, consult: KB19086 - Syslog message: RTPERF_CPU_THRESHOLD_EXCEEDED [juniper.net] .
The event message RTPERF_CPU_THRESHOLD_EXCEEDED appears in the syslog due to excessive load; however, after appearing one time then it does not keep appearing every second. It doesn't appear again until the PFE CPU threshold returns to normal and the excessive load happens again.
This example helps explains the symptom and cause. The following log messages are reported in the syslog:
PERF_MON: RTPERF_CPU_THRESHOLD_EXCEEDED: FPC 0 PIC 0 CPU utilization exceeds threshold, current value=91 PERF_MON: RTPERF_CPU_USAGE_OK: FPC 0 PIC 0 CPU utilization returns to normal, current value=84 PERF_MON: RTPERF_CPU_THRESHOLD_EXCEEDED: FPC 0 PIC 0 CPU utilization exceeds threshold, current value=100 RTPERF_CPU_USAGE_OK: FPC 0 PIC 0 CPU utilization returns to normal, current value=84
Once the value exceeds 85, the "RTPERF_CPU_THRESHOLD_EXCEEDED:..." is reported. The next "RTPERF_CPU_THRESHOLD_EXCEEDED.." log will only be reported after the value is restored to a normal value (lower than 85). In this time interval, even if the overload happens, the "RTPERF_CPU_THRESHOLD_EXCEEDED:..." log does not get reported.
This is a by design, as the trigger of the syslog message RTPERF_CPU_THRESHOLD_EXCEEDED is PFE load from normal range turned over the threshold.
Note : If the message occurs for only a short period of time in the log messages, it can be ignored (temporary CPU utilization spikes are common and expected). The RTPERF_CPU_THRESHOLD is a system parameter that cannot be configured manually.