Symptoms:
False Alarm from cisco_monitor About SNMP Agent not Responding.
We get alarms indicating snmp agent is not responding but when we test using the configuration GUI, the SNMP agent IS responding.
Additionally, some segfaults have been reported in /var/log/messages. Example:
/var/log message "segfault at 0000000000000066 rip 000000000041034c rsp 0000000043b5b740 error 4"
Environment:
Linux RedHat with Cisco_monitor probe installed.
Resolution:
There's an issue with Red Hat Linux where the pthread_kill() function is non-POSIX compliant. So, if invalid threads are passed to the pthread_kill() function, it will throw a segmentation fault which is centered around the non-posix compliant method of handling pthread_kill() if passed an invalid thread ID (which RHEL has not fixed).
The resolution as this writing is to use 32-bit Red Hat Kernel.