Symptom

Cisco Nexus 3K/9K switches can experience snmp memory leak and eventual kernel panic. There is no core getting generated due to low memory in the box. Device reports critical memory alert followed by reload due to kernel panic %$ VDC-1 %$ %PLATFORM-2-MEMORY_ALERT: Memory Status Alert : CRITICAL. Usage 85% of Available Memory ----- reset reason for module 1 (from Supervisor in slot 1) --- Reason: Kernel Panic Service: Version: 9.3(9) + From the logs we have high memory utilization in snmp 16427 1187815424 4294967295 1623404544 ffbbe3f0/ffbb927c snmpd +Processs SNMP is using more than 1Gb bash-4.3$ ps -aux |i snmp root 15891 0.0 0.4 208388 19416 ? Ss Jan05 0:10 /isan/bin/snmpmib_proc root 16427 2.2 30.7 1585356 1200968 ? Ssl Jan05 1073:43 /isan/bin/snmpd -f -s udp:161 udp6:161 tcp:161 tcp6:161 + From the SNMP memory detail , we see this libnetsnmp is using the maximum space 183 [0xf6e49000]libnetsnmp.so.0 22028286 22028316 601287650 601418786

Conditions

SNMP runningon n3k or n9k with multiple polling server. longer uptime.

Workaround

None

Further Problem Description