We have a similar environment where we have co-lo equipment, or equipment that is controlled by the customer or a 3rd party which changes frequently and we are never informed of those changes. However we are always called for stats on such devices (even if they are changed).
Since all the fields are searchable by IP or if it has a name a name, it works. (In many calls, the customer's don't know the IP, or the name (as it could be different under different internal or external DNS's which we or even they (if it's their customers calling up) don't see), so the first 15+ minutes is usually a hunt for /something/ to look into.
Normally with other tools we add them by IP since names/dns is pretty useless in this case. With observium I just wrote a quick script that just added entire ranges to the host table as Hst_A.B.C.D for each. If you have a /LOT/ of them, this does not scale (old problems as to why bind was invented), you can just create a local name-server on the observium system and do the exact same thing there for entire subnets which avoids the linear lookup of a host table.
Simple but gets the job done.
On 5/9/2016 2:50 AM, Tom Laermans wrote:
On 09/05/2016 01:57, Ivan Jukic wrote:
Hostnames are not necessary when monitoring via SNMP, in fact IP Addresses are easier. Second we host many IP address ranges for our clients (Public and Private), so this will be a nightmare to manage. As they tend to change hostnames (and rightly so) all the time. Many of them have managed services.
So you are saying you don't know and don't care about what you are actually monitoring? As it seems they (rightly so, apparently?) swap out devices daily while keeping them on the same IP address...
Tom _______________________________________________ observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium