I’ve been tasked with the job of implementing disk checking via snmp in nagios. I’ve created a perl script with net::snmp to pull the data but i’ve run into a problem. when doing ( hrStorageSize / hrStorageUsed ) * 100 I consistently get back data which is off by ~ 5%. I can only assume this is due to journaling. Multiplying the data by 1.1 or 1.07 seems to generate data that is closer to what I see in df.
My question I suppose is…am I using the wrong MIB or am I forced to run a local client on these machines to get accurate reporting?
don’t know the net::snmp stuff.
Here i simply use a script which makes a df-k on the partition i want and returns the percentage value on a private MIB. maybe it’s easier (problem is you need to modify snmpd.conf on each machine you want to check)
Luca
PS: Could the 5% be the root-reserved space? Edited Thu May 26 2005, 11:24PM ]
Granted a big chunk of them are all the same, but there are still over a hundred servers with individual disk monitoring needs. I really do not want to push a configs to every server.
I’m not sure if it’s a reserve of some kind, though it could be. this problem effects solaris as well as linux. Though NFS mounts do not suffer from this problem.
(problem is you need to modify snmpd.conf on each machine you want to check)