Check memory and swap usage on Windows or Linux Server using SNMP queries

On linux system this plugin creates false alerts by exceeding thresholds without any reason. The way the memory usage is calclulated is just plain wrong. It tries to tell us that most systems have 98% usage just because the buffers for I/O are using them but they have to be subtracted before taking the thresholds into account.

I have this plugin up and running and it works really well. One issue that I am having is that I cannot get the Warning and Critical state to trigger an alarm. I have -w 80 -c 90 set and I have a server at 98% ram utilization, but the status in Nagios shows as Green status. Any suggestions?

I do some change to the script to support faulty Windows 2000.

Defined 2 new OID:

OID_TOTAL_W2K="HOST-RESOURCES-MIB::hrMemorySize"
OID_USED_W2K="HOST-RESOURCES-MIB::hrSWRunPerfMem"

and then added after

MEMORY_USED_ID=`$CMD_SNMPWALK -t 2 -r 2 -v 1 -c $COMMUNITY $HOSTNAME $OID_TAGMEMORY | $CMD_GREP -i 'Physical Memory\|Real Memory' | $CMD_AWK '{ print $1}' | $CMD_AWK -F "." '{print $NF}'`

this test:

if [ -z "$MEMORY_USED_ID" ]; then
MEMORY_USED_ID="0"
fi

finally, in main block:


if [ $MEMORY_USED_ID != "0" ]; then
MEMORY_TOTAL=`$CMD_SNMPGET -t 2 -r 2 -v 1 -c $COMMUNITY -OvqU $HOSTNAME ${OID_TOTAL}.${MEMORY_USED_ID}`
SWAP_TOTAL=`$CMD_SNMPGET -t 2 -r 2 -v 1 -c $COMMUNITY -OvqU $HOSTNAME ${OID_TOTAL}.${SWAP_USED_ID}`
MEMORY_UNIT=`$CMD_SNMPGET -t 2 -r 2 -v 1 -c $COMMUNITY -OvqU $HOSTNAME ${OID_UNIT}.${MEMORY_USED_ID}`
SWAP_UNIT=`$CMD_SNMPGET -t 2 -r 2 -v 1 -c $COMMUNITY -OvqU $HOSTNAME ${OID_UNIT}.${SWAP_USED_ID}`
MEMORY_USED=`$CMD_SNMPGET -t 2 -r 2 -v 1 -c $COMMUNITY -OvqU $HOSTNAME ${OID_USED}.${MEMORY_USED_ID}`
SWAP_USED=`$CMD_SNMPGET -t 2 -r 2 -v 1 -c $COMMUNITY -OvqU $HOSTNAME ${OID_USED}.${SWAP_USED_ID}`
else
# Windows 2000
MEMORY_TOTAL=`$CMD_SNMPGET -t 2 -r 2 -v 1 -c $COMMUNITY -OvqU $HOSTNAME ${OID_TOTAL_W2K}.0`
SWAP_TOTAL=`$CMD_SNMPGET -t 2 -r 2 -v 1 -c $COMMUNITY -OvqU $HOSTNAME ${OID_TOTAL}.${SWAP_USED_ID}`
MEMORY_UNIT=1024
SWAP_UNIT=`$CMD_SNMPGET -t 2 -r 2 -v 1 -c $COMMUNITY -OvqU $HOSTNAME ${OID_UNIT}.${SWAP_USED_ID}`
MEMORY_USED=`$CMD_SNMPWALK -t 2 -r 2 -v 1 -c $COMMUNITY $HOSTNAME ${OID_USED_W2K} | cut -f 4 -d " " |paste -sd+|bc`
SWAP_USED=`$CMD_SNMPGET -t 2 -r 2 -v 1 -c $COMMUNITY -OvqU $HOSTNAME ${OID_USED}.${SWAP_USED_ID}`

fi


Maybe isn't best way, but t works.
Regards
Cristiano

It needs only one change when deciding whether to notify for WARNING or CRITICAL status:

------------------------
instead of:

if [ $MEMORY_USED_REAL_POURCENT -gt $WARNING ] && [ $WARNING != 0 ]; then
STATE=$STATE_WARNING

this line shall be like this:

if [ $MEMORY_USED_REAL_POURCENT -gt $WARNING ] && [ $MEMORY_USED_REAL_POURCENT -lt $CRITICAL ] && [ $WARNING != 0 ]; then
STATE=$STATE_WARNING
----------------------------

Also, the value for size should be changed from Go to GB.

on a 12 GB memory produces an error:
./check_snmp_memory.txt -H 172.16.1.100 -C snmp -w 80 -c 90
./check_snmp_memory.txt: line 123: /usr/bin/bc: No such file or directory
./check_snmp_memory.txt: line 123: /usr/bin/bc: No such file or directory
./check_snmp_memory.txt: line 123: /usr/bin/bc: No such file or directory
./check_snmp_memory.txt: line 123: /usr/bin/bc: No such file or directory
Memory usage : Go used for a total of Go (92%), SWAP usage : Go used for a total of Go (45%)| total=12873891840B;10299113472;11586502656;0 used=11957960704B;0;0;0 swap=12575440896B;0;0;0 buffer=0B;0;0;0 cache=0B;0;0;0

This plugins is very useful.
I support procotol version.
https://gist.github.com/1097533

Thanks.