SNMP Monitoring of MX65W

[namenotoneman]

Hi People,

To add some context - we currently use PRTG to monitor around 700 different "sensors" on our various systems which includes UPS's, servers and comms/network equipment etc. Bandwidth/traffic sensors by default are "probed" every 30 seconds and this provides an accurate and timely picture of what is happening on a link. Some of the equipment we monitor is relatively "cheap", fit for it's specific purpose and quite happily responds and is monitored using the above methods.

Recently we rolled some of the "not so cheap" MX65W's and MS225 switches and set about configuring the SNMP requirements. We could walk the standard / cloud MIB's and therefore created the required sensors. As stated above the default probe interval is 30 seconds and in the case of the MX65W every second probes returned zero's as the value, in some cases two or three consecutive probes returned zero'. We compared the MX65W outputs to our "cheap" Cisco RV215 AP's - the RV215 are probed at the same interval and never miss a response.  

We went back to PRTG's developers and asked them why this could be happening? Their response was to increase the probe interval to 60 seconds and the likely cause is the Meraki doesn't update it SNMP counters that often. The 60 sec scan interval has significantly reduced the zero values returned.

Is there a definitive answer as to how often MX65W's update SNMP data? Can anyone supply official figures?

Regards.

Lee.

[PhilipDAth]

Are you scanning the cloud or the device itself?

[namenotoneman]

Hi,

Should have mentioned that - the device itself. The one I initially tested on was on the same LAN as the monitoring host and the others that exhibit the same issue are remote locations connected via the Auto-VPN feature.

Regards,

Lee.

[MerakiDave]

I believe the MX will update such counters every 30 seconds, but I'd open a case with Support to confirm that and also determine if they have a back-end mechanism to adjust that value.

[namenotoneman]

Hi,
Will do. I'll open a case and seek they input.
Cheers,
Lee.

[Sal_Collora]

I've just done some tests.  The MS225-48LP refreshes the values every 30 seconds, and the MX84 (the only model I have to test) does it every 60 seconds.  I set up PRTG to poll the MX every 65 seconds as I was still getting some zero-values in there.  Fundamentally, this needs vast improvement.  The box itself should write these to these values every 5 seconds like Cisco Catalyst switches.  You are able to capture the peaks and produce a much smoother graph.  While the Dashboard provides ease of use and configuration, much of the time, I find the specifics and granularity a little wanting.  I've deployed probably over 1000 Meraki devices, and I do like the technology, I just wish it had better telemetry capability.

[jwalker55]

We have an MX84 and notice the same issue. SNMP interface counters update every 25 seconds, making it pointless to try doing any kind of real-time monitoring of interface bandwidth. It's 2021, we can update these counters much more frequently. Hell, even $100 Ubiquiti access points update their interface counters every 2 or 3 seconds. 

[Crocker]

Adding in my agreement here.

We're in the process of migrating from a fully Cisco environment to Meraki devices at our satellites. While the Meraki equipment does satisfy most of the performance/security requirements at the satellites, they've been kind of a nightmare for any sort of monitoring or reporting.

Whether it's SNMP values requiring quite a bit more brain work to interpret, trying to set up alerting that is useful without being too spammy (via the Dashboard or third party tools), to trying to get a 'big picture' view of ~150 satellite locations, it's been a real struggle.