I've configured a SNMP monitor using the snmp_dca parent monitor.
The SNMP traffic is well working between our LTM and the node on which the monitor is attached.
( the goal is attached this SNMP monitor to a node so that when the node is running high cpu/mem , it's deactivated by the F5 LTM and all application pool using that node will be rebalanced to remaining health nodes)
When I check the logs from the snmpdca.log , I well see that my current CPU is 99% but the node stay up from LB point of view.
Node Name: /DCS-LB-Partition/FLE-SP02
Node Address: ::ffff:X.X.X.X
SNMP version: v1, community = <sensitive data redacted>, port: 161
Agent Type: WIN2000
.18.104.22.168.22.214.171.124.2.0 - 8387532.000000
.126.96.36.199.188.8.131.52.184.108.40.206 - 99.000000
.220.127.116.11.18.104.22.168.22.214.171.124 - 4096.000000
.126.96.36.199.188.8.131.52.184.108.40.206 - 10067042.000000
.220.127.116.11.18.104.22.168.22.214.171.124 - 20813567.000000
i = 0, util = 0.000000, threshold=70.000000, coeff=1.000000
i = 1, util = 50.000000, threshold=50.000000, coeff=1.000000
i = 2, util = 48.367692, threshold=90.000000, coeff=2.000000
Main:Weight =19 (weight1 = 0.000000, weight2 = 19.417098)
Can someone explain me why the node on which the monitor is attached, does not go down ?
The CPU is constant at 99% for a while but this does not seems to make the node down on LTM .
Thanks in advance for your help.
It's a monitor and then I would exprect that when reaching the treshold, it should deactivate the node on which it's applied. In the same way it does for any other monitor like HTTP that put the node down when it does not answer the configured return code.
Strange that I would need to use an irule or policy for that.
Thanks in advance.
SNMP is design for external monitoring.
HTTP or pool based monitoring is internal monitoring for the f5 to action.
They are different things, but you might be able to get a irule that looks for a log saying the snmp has been triggered and action something. But that's not my area of knowledge.