Integration with Meraki MX devices

[lowellg]

We upgraded to a new version of Cacti (1.2.17) and could not get graphs to show for our Meraki MX devices. Everything was working fine in our older Cacti environment. (We actually did a new install to move from a no longer supported version of CentOS to a current version of Ubuntu Linux).

It turns out the "Downed Device Detection" default setting under the "Availability/Reachability Options" for the device was "SNMP Uptime". Once I changed to "Ping or SNMP Uptime" the graphs worked!

[macan]

Is something in log?

Try to *Enable Device Debug on any Meraki MX devices.

You can try increase snmp timeout too.

[p4ben]

Apologies for resurrecting an old thread, but I'm seeing what I believe is the same issue. I can add MX and other Meraki devices, but RRD files are never created for them. I've enabled debug for the device and see the following.

Code: Select all

2023-12-11 19:45:27 - SPINE: Poller[Main Poller] PID[586369] PT[140636212684352] DEBUG: Device[DEVICE-NAME] HT[1] DEBUG: HOST COMPLETE: About to Exit Device Polling Thread Function
2023-12-11 19:45:27 - SPINE: Poller[Main Poller] PID[586369] PT[140636212684352] Device[DEVICE-NAME] HT[1] Total Time: 0.078 Seconds
2023-12-11 19:45:27 - SPINE: Poller[Main Poller] PID[586369] PT[140636212684352] Device[DEVICE-NAME] HT[1] NOTE: There are '25' Polling Items for this Device
2023-12-11 19:45:27 - SPINE: Poller[Main Poller] PID[586369] PT[140636212684352] Device[DEVICE-NAME] HT[1] DQ[SNMP - Interface Statistics] RECACHE OID: .1.3.6.1.2.1.1.3.0, (assert: < output: )
2023-12-11 19:45:27 - SPINE: Poller[Main Poller] PID[586369] PT[140636212684352] Device[DEVICE-NAME] HT[1] DQ[SNMP - Interface Statistics] Extended Uptime Result: , Is Numeric: 0
2023-12-11 19:45:27 - SPINE: Poller[Main Poller] PID[586369] PT[140636212684352] Device[DEVICE-NAME] HT[1] DQ[SNMP - Interface Statistics] Legacy Uptime Result: U, Is Numeric: 0
2023-12-11 19:45:27 - SPINE: Poller[Main Poller] PID[586369] PT[140636212684352] DEBUG: Device[DEVICE-NAME] HT[1] RECACHE: Processing 1 items in the auto reindex cache for '10.97.2.1'
2023-12-11 19:45:27 - SPINE: Poller[Main Poller] PID[586369] PT[140636212684352] DEBUG: Device[DEVICE-NAME] HT[1] In Poller, About to Start Polling 

Other devices (Cisco IOS and NXOS for example) poll just fine, so I don't believe this is a permissions issue even though the graph debug message states:

Code: Select all

ERROR: opening '/var/www/html/cacti/rra/DEVICE-NAME_traffic_in_4417.rrd': No such file or directory

I think this may be related to the Meraki devices either not reporting uptime, or not reporting it in a format that Cacti understands. I've tried disabling the "Availability/Reachability Options" for the device, and setting them to ICMP only, (ICMP pings are successful), but Cacti still won't create RRD files after I create new MX devices. I'm at a loss. Cacti version is 1.2.25, but I think this actually started after an upgrade that resulted from remediation for CVE-2022-46169.

In short, everything looks like it should work for Meraki devices, SNMP polls are successful, I am able to see a list of interfaces and create graphs, but no RRD files are generated.

[lowellg]

For me, under Availability/Reachability Options > Downed Device Detection is "Ping or SNMP Uptime" and under Ping Method is "ICMP Ping". With these settings I got my Meraki devices to show graphs.

[p4ben]

I've tried that, and ICMP ping is successful. Everything really looks like it should be working, and I see no obvious errors, but the RRD files just never get created. Nothing I see in the debug info or logs gives any indication as to why though. I'm not able to find out if it's documented behavior for Cacti to ignore hosts when it can't determine uptime. The devices are showing as "UP" in Cacti, so my assumption is that it should try to create graphs and RRDs. I understand devices not being polled and RRD files not being created if Cacti determines the devices are in a down state, but that doesn't seem to be the case here.

Data sources seem ok from what I can see. I'm really stumped. This is for both MX firewalls and MS switches. Every other device I've added, Cisco, Juniper and Palo Alto has worked exactly as expected, but there seems to be some aspect of the Meraki devices that causes problems. The only immediately obvious difference is that Cacti doesn't recognize the uptime reporting of the Meraki devices, but using ICMP or disabling the checks altogether has no impact on the issue. The devices were definitely working in the past, and unfortunately I can't pinpoint the change beyond it seeming to coincide with the patch for the CVE.

[p4ben]

Also, to add. Realtime graphs for Meraki hosts work, so I'm reasonably certain everything is ok with reachability and SNMP. I just seems like something with the graphs or possibly the data sources isn't right which is causing the RRD files to never get created.

[Osiris]

Use SNMP only. Get off version 1.2.17 ASAP. 1.2.25 is real solid. 1.2.26 coming soon. Then 1.2.x will only get security updates.

[p4ben]

I'm on 1.2.25. What do you mean by use SNMP only? I've tried SNMP Uptime, SNMP Desc, and SNMP GetNext and I still don't get RRD files generated. The interface query runs successfully, and I'm able to get realtime graphs, so I'm confident SNMP is working. I'm using the Generic SNMP Device template.

[Osiris]

Check the poller cache and then put the device in debug mode to see you are getting data. Report back.

[p4ben]

Screenshots of debug info from the logs and poller cache attached.

debug log

debug log.png (32.17 KiB) Viewed 862 times

poller cache

poller cache.png (91.03 KiB) Viewed 830 times

[TheWitness]

For that device, locate the devices host_id and then run spine by hand manually like follows:

./spine -R -V 3 -S -H host_id

The host_id is a number. It should show all the values.