JunOS SNMP LogicMonitor Performance

On one of our Juniper Virtual Chassis, we noticed we that we were getting constant and correct responses via SNMP for all our Juniper DataSources (CPU/DOM/State) except for the Network Interfaces DataSource. (“SNMP_Network_Interfaces”)

Upon running the collection script, we normally ran into a timeout error. We were expecting around 200 Interfaces to be monitored after filtering.

We were thinking of spinning up another collector and setting the timeout to a higher value, but that seemed like wasted resources for just 1 Juniper device.

After some research we stumbled across this article (https://forum.checkmk.com/t/junos-snmp-performance-hints/41429)

Using the following 2 commands on our JunOS device allowed us to filter out a large number of sub-interfaces along with any internal interfaces.

set snmp filter-interfaces interfaces \.0$

set snmp filter-interfaces all-internal-interfaces

We tested this on another Juniper device that we wanted to cut down on monitoring sub interfaces and saw that it went from pulling 250 interfaces when running the script down to around 150 before the active discovery filtering.

Once these changes were made to our problematic JunOS device, we almost instantly started getting interface traffic in LogicMonitor.  

I thought I’d shared this in case anyone was having any issues monitoring their Juniper devices via SNMP.

The article I referenced goes into more detail on the specifics of the commands.