Check IO stats of one or all disks

There is a better way to get the modify timestamp of a file - check the -c option of stat: OLDDISKSTAT_EPOCH=$(stat $HISTFILE -c %Y) Furthermore the perfdata documentation states, that you should add the warning and crit levels: echo "${OUTPUT}summary: $TPS io/s, read $SECTORS_READ sectors (${KBYTES_READ_PER_SEC}kB/s), write $SECTORS_WRITE sectors (${KBYTES_WRITTEN_PER_SEC}kB/s) in $TIME seconds | tps=${TPS};${WARN_TPS};${CRIT_TPS};; read=${BYTES_READ_PER_SEC};${WARN_READ};${CRIT_READ};; write=${BYTES_WRITTEN_PER_SEC};${WARN_WRITE};${CRIT_WRITE};;"

Plugin is reading /sys/block/DEVICE/stat file, so does not depend on output of other programs (like iostat). Also keeps continuous measurement of system by keeping previous read in temporary file. One thing to point out though: Plugin checks read/write tresholds against bytes read/written per second, not against sectors per second as help states.

It's very useful check which must be turned on for every server! PS: but small changes for check_all_diskstat.sh can be applied as soon it not returning a correct exit status for nagios notification. I make some changes which can help somebody else: ################### #!/bin/bash EXITCODE=0 CHK=/usr/lib/nagios/plugins/check_diskstat.sh WARN=${1:-"300,10000,10000"} CRIT=${2:-"400,20000,20000"} for DEVICE in `ls /sys/block`; do if [ -L /sys/block/$DEVICE/device ]; then DEVNAME=$(echo /dev/$DEVICE | sed 's#!#/#g') echo -n "$DEVNAME: " OUTPUT="`$CHK -d $DEVICE -w $WARN -c $CRIT`" STATUS=$? if [ "$EXITCODE" -le "$STATUS" ]; then EXITCODE=$STATUS; fi echo $OUTPUT | sed "s#=#_$DEVNAME=#g" fi done exit $EXITCODE ###################