High cpu in monitoring because of SnmpSubagent andf SnmpMaster processes

[Frank_Veen]

Hi all,

Can anybody help explain why the snmpMaster and snmpSubagent processes are using a lot of cpu during snmp polls? Maybe misconfiguration?

Running image is 16.1.3.6-patch1-9

In our monitoring we see that every few minutes the cpu utilisation of one of our X440's is abnormal high. Our monitoring server uses snmp polls every minute to get the switch cpu utilization (among other things).

This is happening when there is almost no traffic even, for example during nights. See the image below the cpu utilization every few minutes around 40%:

Now the top command on the switch shows more info, clearly the SnmpMaster and SnmpSubagent are using a lot of CPU during the poll:

Kind regards,

Frank

[Frank_Veen]

Hi,

Thanks for helping out,
We are using snmp v2, the server is a remote (Zabbix) monitoring server.
The are no relevant messages in the logging.

Do I need to explicitly add the deny all in the acl ?

The configuration on switch:

X440 # show configuration | incl snmp
configure snmp sysName "X440"
configure snmp sysLocation "xxx"
configure snmp sysContact "xxx"
# Module snmpMaster configuration.
configure snmp access-profile "snmpACL" readonly
X440.2 #

The snmpACL.pol file:

entry allow_zabbix_bnc {if match all {
source-address x.x.x.x/32;
}
then {
permit;
}
}

Kind regards,

Frank

[Frank_Veen]

Are you sure you are employee? 

[Tripathy__Priya]

Default config is nothing but If you forget your password while logged out of the CLI, you can use the boot loader to reinstall a default switch configuration, which allows access to the switch without a password. Note that this process re configures all
switch settings back to the initial default configuration.

[Frank_Veen]

What default config are you referring to?