vfs.fs.size[<fs>,used] gives bad value

[dcrandall]

Hi,

zabbix-1.1beta9

vfs.fs.size[/tmp,used] returns 84.21 MB
df-h /tmp USED collumn reports 5.1 MB

I'm seeing this for all of the file systems that I'm monitoring used space with vfs.fs.size[,used].

On the other hand vfs.fs.size[<fs>,total] returns exactly correct total size.

Problem with the "used" mode perhaps?

[Alexei]

Please provide us with platform details.

[dcrandall]

Sorry, should have done that...

FreeBSD 5.4-STABLE on both zabbix server and client.

I figured out some more details...
So I decided to switch to SNMPv2 agent.
SNMP also reports unexpected values for hrStorageUsed... exactly half of what df says.
Turns out that...
hrStorageAllocationUnits.9 = INTEGER: 2048 Bytes
So I set the multiplier to 2048 and this made the data from SNMPv2 agent accurate.

The numbers from zabbix_agent don't fit that profile though.
For instance df -k / gives a USED value of 79194
zabbix_get -s<hsot> -k"vfs.fs.size[/,used] gives 99488
I'm not sure what that means in terms of the multiplier.

dan

Please provide us with platform details.

[jasonj]

Same here...

I have the same errors on RHEL3, RHEL4, FC3 and FC4

022988:20060427:110017 Got empty string from [********] IP [] Parameter [vfs.fs.size[/usr,free]]

This is not the only thing that has starting having issues after upgraded to beta9

net.if.in for lo, eth0, and eth1 on many hosts... inconsistantly

Most of them end with something like

023035:20060427:110713 Assuming that agent dropped connection because of access permissions
023035:20060427:110713 Started network errors for [******]
023035:20060427:110713 Host [******]: another network error, wait for 5 seconds

UPDATE::
Using the cli program zabbix_get works with no problems
zabbix_get -s ********* -k vfs.fs.size[/usr,used]
14629776

[dcrandall]

Using the cli program zabbix_get works with no problems
zabbix_get -s ********* -k vfs.fs.size[/usr,used]
14629776

I do get a value from vfs.fs.size[/usr,used], it's just that the value is way off.