Monitoring disk space

scottyp55 · 03-19-2010, 03:19 PM

We are running IPmonitor to monitor the disk usage on our Linux servers. It does not seem to coincide with what is reported when running df -h. For example on a Red Hat 5.3 server - our IPmonitor shows that 85% is used on the /usr partition, however when I do a df -h on the server it shows that 91% is used.

Does anyone know why there would be a discrepancy? IPmonitor uses SNMP.

Thanks,
Scott

smoker · 03-19-2010, 03:39 PM

Maybe IPmonitor uses a different way of getting the usage.

Code:

[smoker@kids ~]$ du -sh /boot
7.8M    /boot
[smoker@kids ~]$ df -h /boot
Filesystem            Size  Used Avail Use% Mounted on
/dev/hda1              99M   12M   82M  13% /boot
[smoker@kids ~]$

Of course they would still have to calculate a percentage somehow, which seems silly as it's there in the df output.

MensaWater · 03-19-2010, 03:40 PM

There is reserved space that system processes (root etc...) can continue writing to but regular user processes can't. The used % in df is based on what the user can access rather than what the system can access.

scottyp55 · 03-22-2010, 09:31 AM

Thanks for the response.
So is there a way for SNMP to accurately report the disk usage?

kirukan · 03-22-2010, 10:09 AM

Nagios, Mrtg & Snmptrap etc..

MensaWater · 03-22-2010, 10:32 AM

Quote:

Originally Posted by scottyp55

Thanks for the response.
So is there a way for SNMP to accurately report the disk usage?

It depends on what's in the MIB for what you're monitoring. I haven't done a lot with SNMP (will be doing more this years since I finally got the newer Nagios installed). However, most MIBs give you tons of metrrics so it may just be you need to pick a different metric than the one you're using now from the same MIB.