Hi all,

Trying to monitor various routerboards using collectd. Snmpwalk using SNMPv3 works great, I can connect to the routerboards no problem.

In the case of collectd, I can connect to one big switch and get a response, or two small switches and get a response, and more than that and the attempts to fetch data time out as follows:
Can anyone confirm whether they use and have had joy with SNMP and collectd?

(I know the routeros plugin exists, I need this to be SNMP).

Turns out all routerboards default to the same engine-id by default. This confuses net-snmp, and in turn collectd-snmp, which use one engine definition for all unrelated connections. One router works, all other routers reject the packets, and trigger the timeouts.

Workaround is to manually set a unique engine-id under the /snmp menu for each router.

Bug raised here: https://github.com/net-snmp/net-snmp/issues/120