11-16-2015 06:13 AM - edited 11-16-2015 06:16 AM
I have the following bottleneckmon settings on a switch:
"bottleneckmon --status" output:
Bottleneck detection - Enabled
Switch-wide sub-second latency bottleneck criterion:
Time threshold - 0.500
Severity threshold - 50.000
Switch-wide alerting parameters:
Alerts - Yes
Latency threshold for alert - 0.100
Congestion threshold for alert - 0.500
Averaging time for alert - 60 seconds
Quiet time for alert - 60 seconds
But the bottleneck monitoring doesn't seem to work.
We had an issue when an SVC was maxing out the 8G bandwidth for several hours.
When I check a historical performance graph for the relevant port in BNA, it shows that the port was doing 780 MB/sec (rx) for many hours non-stop.
I'm wondering why we did not receive any bottleneckmon alerts from this.
If I check directly on the switch with errdump, I also don't see a single congestion bottleneck warning.
Also, when I issued "bottleneckmon --show" at the time that the problem existed,
it was just showing "number of bottlenecked ports: 0",
this while "portperfshow -tx -rx" for the relevant port was constantly showing an RX rate of more than 700 MB/s.
I wonder why bottleneckmon is silent.
Any idea what I am missing here?
(switch firmware level is 7.3.1a)
11-18-2015 05:07 AM
where do you want to see those alerts?
maybe check snmp settings? if you use snmp to forward the alerts to a common system where you collect everything, besides bna
or you invest some time and read yourself into MAPS
in maps there is nomore bottleneckmon instead you use maps policies to monitor the port performance attributes
beware of port fencing and port decoomision in MAPS