SNMP disk space.

L

lbl

Hi

I think i need some fresh eyes on this one, im trying to calculate the free disk space via snmp (net-snmp)

But it seems to be way off and im not shure why ...

Any clues ?

/lbl

Code: 
[lbl@atom1 ~]$ cat check_snmp_disk 
#!/usr/local/bin/bash

# This script takes:
# <host> <community> <mountpoint> <megs>

snmpwalk="/usr/local/bin/snmpwalk"
snmpget="/usr/local/bin/snmpget"

calc_free()
# takes <size> <used> <allocation>
{
echo "$1 $2 - $3 * 1024 / 1024 / p" | dc
}

if result=`$snmpwalk -v2c -c $2 -Oq $1 hrStorageDescr | grep "$3$"`
	then
		index=`echo $result | sed 's/.*hrStorageDescr//' | sed 's/ .*//'`

		args=`$snmpget -v2c -c $2 -Oqv $1 hrStorageSize$index hrStorageUsed$index hrStorageAllocationUnits$index | while read oid j ; do printf " $oid" ; done`

		free=`calc_free$args`

		if [ "$free" -gt "$4" ]
			then
				echo "DISK OK: mount $3 free $free MB."
				exit 0
			else
				echo "DISK CRITICAL: mount $3 free $free MB."
				exit 2
		fi
	else
		echo "DISK CRITICAL: $3 dosent exist or snmp isent responding."
		exit 3
fi
[lbl@atom1 ~]$ df -h
Filesystem            Size    Used   Avail Capacity  Mounted on
/dev/mirror/ides1a    496M    236M    221M    52%    /
devfs                 1.0K    1.0K      0B   100%    /dev
/dev/mirror/ides1e    496M    1.4M    455M     0%    /tmp
/dev/mirror/ides1f    103G    3.6G     91G     4%    /usr
/dev/mirror/ides1d    2.9G    1.8G    860M    68%    /var
/dev/raid3/sata0      1.8T    1.6T     61G    96%    /storage/download
/dev/raid3/sata1      1.8T    911G    749G    55%    /storage/pub
devfs                 1.0K    1.0K      0B   100%    /var/named/dev
[lbl@atom1 ~]$ df 
Filesystem         1K-blocks        Used     Avail Capacity  Mounted on
/dev/mirror/ides1a     507630     241164    225856    52%    /
devfs                       1          1         0   100%    /dev
/dev/mirror/ides1e     507630       1480    465540     0%    /tmp
/dev/mirror/ides1f  108230294    3783806  95788066     4%    /usr
/dev/mirror/ides1d    3017358    1895590    880380    68%    /var
/dev/raid3/sata0   1892045722 1676425126  64256940    96%    /storage/download
/dev/raid3/sata1   1892045722  955594880 785087186    55%    /storage/pub
devfs                       1          1         0   100%    /var/named/dev
[lbl@atom1 ~]$ ./check_snmp_disk localhost public / 100
DISK OK: mount / free 260 MB.
[lbl@atom1 ~]$ ./check_snmp_disk localhost public /tmp 100
DISK OK: mount /tmp free 494 MB.
[lbl@atom1 ~]$ ./check_snmp_disk localhost public /usr 100
DISK OK: mount /usr free 101998 MB.
[lbl@atom1 ~]$ ./check_snmp_disk localhost public /var 100
DISK OK: mount /var free 1095 MB.
[lbl@atom1 ~]$ ./check_snmp_disk localhost public /storage/download 100
DISK OK: mount /storage/download free 210566 MB.
[lbl@atom1 ~]$ ./check_snmp_disk localhost public /storage/pub 100
DISK OK: mount /storage/pub free 914502 MB.
[lbl@atom1 ~]$
 
Seems right.

Hmm it seems that im actualy calculating it correctly ...

df is way off tho ...

Can anyone point me in the right direction for understanding how this works ?

Code: 
[lbl@atom1 ~]$ df
Filesystem         1K-blocks        Used     Avail Capacity  Mounted on
/dev/mirror/ides1a     507630     241164    225856    52%    /
devfs                       1          1         0   100%    /dev
/dev/mirror/ides1e     507630       1480    465540     0%    /tmp
/dev/mirror/ides1f  108230294    3783924  95787948     4%    /usr
/dev/mirror/ides1d    3017358    1895916    880054    68%    /var
/dev/raid3/sata0   1892045722 1676425126  64256940    96%    /storage/download
/dev/raid3/sata1   1892045722  955594880 785087186    55%    /storage/pub
devfs                       1          1         0   100%    /var/named/dev
[lbl@atom1 ~]$ echo "5 k 1892045722 1676425126 - 1024 / 1024 / p" | dc
205.63182
[lbl@atom1 ~]$ echo "5 k 64256940 1024 / 1024 / p" | dc
61.28019
[lbl@atom1 ~]$ echo "5 k 1892045722 955594880 - 1024 / 1024 / p" | dc
893.06911
[lbl@atom1 ~]$ echo "5 k 785087186 1024 / 1024 / p" | dc
748.71748
[lbl@atom1 ~]$ df -h
Filesystem            Size    Used   Avail Capacity  Mounted on
/dev/mirror/ides1a    496M    236M    221M    52%    /
devfs                 1.0K    1.0K      0B   100%    /dev
/dev/mirror/ides1e    496M    1.4M    455M     0%    /tmp
/dev/mirror/ides1f    103G    3.6G     91G     4%    /usr
/dev/mirror/ides1d    2.9G    1.8G    859M    68%    /var
/dev/raid3/sata0      1.8T    1.6T     61G    96%    /storage/download
/dev/raid3/sata1      1.8T    911G    749G    55%    /storage/pub
devfs                 1.0K    1.0K      0B   100%    /var/named/dev
[lbl@atom1 ~]$
 
The answer is simpler than you might think. Notice how the answers are 5% different? UFS2 reserves 5% disk usage for the root user. Storage as reported by df will go to 105% disk usage if written by root. However, the disk usage reported by net-snmp will only to 100%.

To verify this, check the size of the disks as reported by snmp and by df (don't worry about used space, just total space). You should see the discrepancy there.
 
gordon@ said:
UFS2 reserves 5% disk usage for the root user
Click to expand...
You sure?

6.2 and 8.0 both have this:

Code: 
$ grep MINFREE /usr/src/sys/ufs/ffs/fs.h
 * MINFREE gives the minimum acceptable percentage of filesystem
#define MINFREE         8

lbl: See the "-m" option in the man pages for newfs and tunefs
 
Nice

Thanks ... just what i was looking for ...

If somebody wants the script its here in the new version.:

Code: 
[lbl@atom0 ~/scripts/nagios-plugins]$ cat check_snmp_disk 
#!/usr/local/bin/bash

# This script takes:
# <host> <community> <mountpoint> <megs>

snmpwalk=/usr/local/bin/snmpwalk
snmpget=/usr/local/bin/snmpget

calc_free()
# takes <size> <used> <allocation>
{
if result=`echo "($1*$3*.92/1024-$2*$3/1024)/1024" | bc`
	then
		echo "$result"
	else
		echo "DISK UNKNOWN: Cant calculate free space."
		exit 3
fi
}

fetch_details()
#takes <host> <community> <index>
{
if result=`$snmpget -v2c -c $2 -OqvU $1 hrStorageSize.$3 hrStorageUsed.$3 hrStorageAllocationUnits.$3 | while read oid ; do printf "$oid " ; done`
	then
		echo "$result"
	else
		echo "DISK UNKNOWN: Cant fetch details."
		exit 3
fi
}

if result=`$snmpwalk -v2c -c $2 -Oqs $1 hrStorageDescr | grep "$3$"`
	then
		index=`echo $result | sed 's/hrStorageDescr\.//' | sed 's/ .*//'`

		details=`fetch_details $1 $2 $index`

		free=`calc_free $details`

		if [ "$free" -gt "$4" ]
			then
				echo "DISK OK: mount $3 free $free MB."
				exit 0
			else
				echo "DISK CRITICAL: mount $3 free $free MB."
				exit 2
		fi
	else
		echo "DISK UNKNOWN: $3 dosent exist or snmp isent responding."
		exit 3
fi
[lbl@atom0 ~/scripts/nagios-plugins]$
 
To get disk usage, I found I only needed to uncomment one line in snmpd.config:

vim /etc/snmpd.config
Code: 
begemotSnmpdModulePath."hostres" = "/usr/lib/snmp_hostres.so"

vim /etc/rc.conf
Code: 
bsnmpd_enable="YES"

And then oddly, on some servers the SNMP description (used to look up a specific file system in nagios) is simply "/" and on others with the same OS version, it is something like "/, type zfs, dev ...".
 
You must log in or register to reply here.
Back
Top