> -----Original Message----- > From: Jean Delvare [mailto:khali@xxxxxxxxxxxx] > Sent: Saturday, November 17, 2012 10:35 AM > To: Leslie Rhorer > Cc: 'Guenter Roeck'; lm-sensors@xxxxxxxxxxxxxx > Subject: Re: Getting alarms using lm-sensors > > On Sat, 17 Nov 2012 10:10:31 -0600, Leslie Rhorer wrote: > > OK, this works: > > > > mail -a "Content-Type: text/plain; charset=UTF-8" -a From:sensor_monitor > -s > > "RAID-Server Sensor Event Notification" ... > > > > Would you guys be interested in me posting the script for future > reference? > > Sure, why not. Our reference script is: > http://www.lm-sensors.org/browser/lm-sensors/trunk/prog/daemon/healthd.sh > but it assumes the chip and driver report alarms on out-of-bounds > conditions. > > -- > Jean Delvare OK, well here is one that supports a chip or drive that does not report OOB conditions. Comments and bug reports most welcome. First, there must exist a colon-delimited table named "sensetab" unique to the output of `sensors` on the system in question. Here is an example from this system: V01:.85:1.6:Vcore Voltage:Vcore:alarm:boot:is low:is critical. Shutting down... V02:10.2:13.8:+12V Voltage:+12V:alarm:boot:is out of bounds:is critical. Shutting down... V03:4.5:5.5:+5V Voltage:+5V:alarm:boot:is out of bounds:is critical. Shutting down... V04:2.97:3.63:+3.3V Voltage:+3.3V:alarm:boot:is out of bounds:is critical. Shutting down... V05:1.71:2.09:DDR2:Memory Voltage:alarm:boot:is out of bounds:is critical. Shutting down... V06:1.08:1.32:HT:HyperThreading Voltage:alarm:boot:is out of bounds:is critical. Shutting down... V07:.99:1.61:SB:SouthBridge Voltage:alarm:boot:is out of bounds:is critical. Shutting down... V08:1.08:1.32:BR:BR Voltage:alarm:boot:is out of bounds:is critical. Shutting down... V09:2.25:2.85:VDDA:VDDA Voltage:alarm:boot:is out of bounds:is critical. Shutting down... V10:.85:1.04:DDR2 TERM.:Memory Termination:alarm:boot:is out of bounds:is critical. Shutting down... V11:1.14:1.43:VDDNB:NorthBridge Voltage:alarm:boot:is out of bounds:is critical. Shutting down... F01:800::CPU_FAN FAN Speed:CPU Fan:alarm::is failing: F02:800::CHA_FAN1 FAN Speed:Rear Fan #1::alarm:is failing: F03:800::CHA_FAN2 FAN Speed:Rear Fan #2::alarm:is failing: F04:::OPT_FAN1 FAN Speed::alarm::is failing F05:::OPT_FAN2 FAN Speed::alarm::is failing F06:::OPT_FAN3 FAN Speed:Water Cooler Fan:alarm::is failing F07:800::PWR_FAN FAN Speed:Power Supply Fan:alarm::is failing F08:::CHA_FAN3 FAN Speed:Water Cooler Pump:alarm::is failing T01:50:85:CPU Temperature:CPU Temperature:alarm:boot:is too high.:is critical. Shutting down... T02:53:85:MB Temperature:Motherboard Temperature:alarm:boot:is too high.:is critical. Shutting down... T03:::OPT1:Coolant Level:alarm::is too low. Add coolant.: T04:::OPT2::::: T05:::OPT3::::: T06:::temp1::::: In this version, the fields are: F1 - SensorID F2 - First boundary value F3 - Second boundary value F4 - Name of sensor output by `sesnors` F5 - Name of sensor sent in e-mail F6 - Action to take if first boundary is violated ( alarm or boot ) F7 - Action to take if second boundary is violated ( alarm or boot ) F8 - Failure text to send in e-mail if first boundary is violated F9 - Failure text to send in e-mail if second boundary is violated These fields can easily be moved, added, or deleted by assigning the -f values at the top of the script. Blank fields are ignored, but all fields should exist on every line. Now the script itself: #! /bin/bash export LC_ALL=en_US.UTF-8 export LANG=en_US.UTF-8 export LANGUAGE=en_US.UTF-8 SenseDir=/usr/share/sensors # Location of files SenseTab=$SenseDir/sensetab # Lookup table SenseStat=$SenseDir/alarm_stat # Status file containing active alarms, when the last e-mail was sent in epoch time, and the action taken. NOTIFY=1200 # Number of seconds between e-mail notifications for each event # Field number position identifiers in $SenseTab. Modify to add, delete, or move fields in the table PosSID="-f1" PosLVAL="-f2" PosHVAL="-f3" PosAID="-f4" PosTID="-f5" PosBRL1="-f6" PosBRH1="-f7" PosBRL2="-f8" PosBRH2="-f9" cd $SenseDir # Check the lower boundary for violations. lo_check() { if [[ $MINMAX -eq 0 ]] then (( $(echo "$CVAL > $LVAL" | bc -l) )) && BERRL=1 else (( $(echo "$CVAL < $LVAL" | bc -l) )) && BERRL=2 fi [[ $BERRL -gt 0 ]] && action_check $PosBRL1 $PosBRL2 } # Check the upper boundary for violations. hi_check() { (( $(echo "$CVAL > $HVAL" | bc -l) )) && BERRH=1 [[ $BERRH -gt 0 ]] && action_check $PosBRH1 $PosBRH2 } # Set the ACTION directive. 1 = Send Alarm 2 = Shutdown action_check () { ACTION=1 [[ $(echo $line | cut -d ":" $1 ) == "boot" ]] && ACTION=2 ATEXT=$(echo $line | cut -d ":" $2 ) # $ATEXT = Text action for e-mail } # Check if the alarm is already active and if so how long ago the e-mail was sent. # Do nothing if the alarm is already active and $NOTIFY seconds or less have passed. # Send e-mail and update $SenseStat if more than $NOTIFY seconds have passed. alarm_email () { WRITE=1 touch $SenseStat # Make sure the file exists. STATL=$( grep $SID $SenseStat ) # See if the alarm exists. if [[ -n $STATL ]] # If so, check how old it is. then ESEC=$( date +%s ) # Get epoch time. SSEC=$( echo $STATL | cut -d ":" -f2 ) # Get elapsed time. TSEC=$((ESEC - SSEC)) # Compute the difference. if [[ $TSEC -gt $NOTIFY ]] # If more than $NOTIFY seconds have passed, upodate the file then WRITE=0 # Set up for e-mail notification grep -v $SID $SenseStat > $SenseStat.tmp mv $SenseStat.tmp $SenseStat fi else WRITE=0 # Alarm is not yet active, so set up to send the e-mail and update $SenseStat fi if [[ $WRITE -eq 0 ]] # If $WRITE is reset, send the e-mail and update $SenseStat then echo "$TID $ATEXT Value: $F2" | mail -a "Content-Type: text/plain; charset=UTF-8" -a From:sensor_monitor -s "Server Sensor Event Notification" xxxxxxxxxxxxx@xxxxxxxxxxxx yyyyyyyyyyyy@xxxxxxxxxxxx echo "$SID:$ESEC:$ACTION" >> $SenseStat sleep 1 fi } while read line # Read and process one line of $SenseTab at a time do SCAN=0 SID=$( echo $line | cut -d ":" $PosSID ) # $SID = Sensor ID AID=$(echo $line | cut -d ":" $PosAID ) # $AID = Sensor Name TID=$(echo $line | cut -d ":" $PosTID ) # $TID = Text Name for e-mail FIELDS=$( sensors | grep "$AID:" ) # $FIELDS = Text from `sensors` F2=$( echo $FIELDS | cut -d ":" -f2 ) # $F2 = Description from `sensors` CVAL=${F2% (*} # $CVAL = Current Sensor Value CVAL=${CVAL//[A-Z°+]/ } LVAL=$(grep $SID $SenseTab | cut -d ":" $PosLVAL ) # $LVAL = Lower Configured Value HVAL=$(grep $SID $SenseTab | cut -d ":" $PosHVAL ) # $HVAL = Higher Configured Value AVAL=${FIELDS#* =} # $AVAL = Lower Default Value AVAL=${AVAL%,*} AVAL=${AVAL//[A-Z°+) ]/ } BVAL=${FIELDS##* = } # $BVAL = Higher Default Value BVAL=${BVAL//[A-Z°+) ]/ } [[ -n $LVAL ]] && SCAN=1 # $SCAN: 1 - Lower value only [[ -n $HVAL ]] && SCAN=2 # $SCAN: 2 - Higher value only [[ -n $LVAL && -n $HVAL ]] && SCAN=3 # $SCAN: 3 - Lower & Higher value MINMAX=1 # $MINMAX: 1 - Minimum & Maximum echo $F2 | grep -q crit && MINMAX=0 # $MINMAX: 0 - High & Critical BERRL=0 # $BERRL: 0 - No Alarm, 1 - Above High, 2 - Below Minimum BERRH=0 # $BERRH: 0 - No Alarm, 1 - Above Maximum / Critical ACTION=0 # $ACTION: 0 - Do Nothing, 1 - Send e-mail, 2 - Shutdown case "$SCAN" in # Check low boundary, high boundary, or both based upon whether they are defined in $SenseStat. 1) lo_check # Check low boundary. ;; 2) hi_check # Check high boundary. ;; 3) lo_check # Check both. hi_check ;; esac case "$ACTION" in # Perform action based upon values in $SenseTab. alarm = send e-mail boot = shutdown 0) # Take no action, except to clear the alarm in $SenseStat, if it is active. grep -q $SID $SenseStat if [[ $? -eq 0 ]] then grep -v $SID $SenseStat > $SenseStat.tmp mv $SenseStat.tmp $SenseStat fi ;; 2) # Shutdown, allowing enough time for the e-mail to be sent shutdown -py 30 & ;& 1) # Send the e-mail alarm_email ;; esac done < $SenseTab _______________________________________________ lm-sensors mailing list lm-sensors@xxxxxxxxxxxxxx http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
