Support Center > Search Results > SecureKnowledge Details
"snmpd" daemon occasionally consumes CPU at very high level on machine with several hundred interfaces
Symptoms
  • "snmpd" daemon occasionally (every ~5 seconds) consumes CPU at very high level (spikes).

    Example output from "top" command:

      PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  P COMMAND 
     3072 admin     17   0 25788  10m 6572 S   90  0.1  17:17.01 1 snmpd
    
  • On Cluster, as a result of high CPU load, the CUL mechanism is triggered.
    Example from /var/log/messages file:
    FW-1: [cul_load_freeze][CUL - Cluster] Setting CUL FREEZE_ON, high kernel CPU usage (85%) on local Member 1, threshold = 80%

  • Output of "strace" utility attached to SNMPD daemon shows during the issue (during spike in CPU load) that SNMPD daemon is querying various interface information on the machine.

    Example:
    ioctl(13, SIOCGIFCONF, {31264, {{"lo", {AF_INET, inet_addr("127.0.0.1")}}, {"eth1-05", {AF_INET, inet_addr("172.30.56.43")}}, {"eth1-06", {AF_INET, inet_addr("172.30.62.43")}}, {"eth1-07", {AF_INET, inet_addr("172.30.76.51")}}, {"eth1-08", {AF_INET, inet_addr("172.30.26.14")}}, {"Sync", {AF_INET, inet_addr("172.30.84.17")}}, {"Mgmt", {AF_INET, inet_addr("172.30.93.58")}}, {"bond1.797", {AF_INET, inet_addr("172.30.122.66")}}

Cause

"snmpd" daemon loads the interface table every ~5 seconds. On a machine with several hundred interfaces (e.g., 900 VLAN interfaces) this operation might consume the CPU at high level.


Solution
Note: To view this solution you need to Sign In .