Problem with correct settings on duplicate sysName machines

Help
,
#1

hey folks,

I’ve got a question about allow_duplicate_sysName …
I have currently set this to false because I have enabled autodiscovery and don’t want to have hosts with multiple IPs to be added multiple times. For most machines this does work fine, because for example my switches and routers all propagate their FQDN as sysName via snmp.

Now I see that this is a caveat since I have several linux machines (mostly debian) where the snmp sysName is the hostname by default, instead of the FQDN, and since I got many machines where the hostname is “srv1” and the rest of the FQDN is different, I end up with only one of them added, while the rest of them is blocked because of “duplicate sysName”.
I see that setting allow_duplicate_sysName to true would resolve this, but as far as I understand, I then have the risk that machines with multiple IP addresses would be discovered multiple times, which happened to me with my first test installation.

How can I solve this without manually setting the sysName in the snmpd.conf on all these machines? Why does snmpd even do that? According to RFC 3418, sysName is defined as “By convention, this is the node’s fully-qualified domain name.”

Frank

#2

The multiple IPs of the machines are in the same subnet?

#3

Depends. There are some hosts which contain several IPs from the same network, but I also got several machines that have IPs from different subnets (for example backup and monitoring nodes).

#4

This topic was automatically closed 91 days after the last reply. New replies are no longer allowed.