Linux Server Monitoring with IPMI

Listing 1 ipmi init script for Linux

#!/bin/sh
#
# chkconfig: 2345 11 60
# description: start, stop, or query ipmi system monitoring tools
# config: /etc/sysconfig/ipmi

# Source function library.
. /etc/init.d/functions

# Exit silently if we don't have a sysconfig file,
# and read IPMI setting from it to determine whether or
# not to continue.
# The only current setting is ipmi={YES|NO}, whether or not
# to enable IPMI.
[ -f /etc/sysconfig/ipmi ] || exit 0
. /etc/sysconfig/ipmi
[ "${IPMI}" = "yes" ] || exit 0

RETVAL=0

start() {
    echo -n $"Starting ipmi: "

    # If ipmidev isn't listed in /proc/devices, try
    # loading the modules.
    if ! grep -q ipmidev /proc/devices
    then
        /sbin/modprobe ipmi_msghandler || RETVAL=1
        /sbin/modprobe ipmi_devintf || RETVAL=1
        # Try loading new driver module, fall back to old
        # module if that fails.
        if ! /sbin/modprobe ipmi_si >/dev/null 2>&1
        then
        /sbin/modprobe ipmi_kcs_drv || RETVAL=1
        fi
    fi


    # If ipmidev still isn't listed in /proc/devices after we load
    # modules, this just isn't going to work.  Set RETVAL to mark
    # this failure.
    grep -q ipmidev /proc/devices || RETVAL=1

    # Check if the device file exists and create if not.
    if [ ! -c /dev/ipmi0 ] && [ $RETVAL -eq 0 ]
    then
        major=$(awk '/ ipmidev$/{print $1}' /proc/devices)
        /bin/mknod -m 0600 /dev/ipmi0 c $major 0 || RETVAL=1
    fi

    if [ $RETVAL -eq 0 ] && touch /var/lock/subsys/ipmi ; then
        echo_success
        echo
    else
        echo_failure
        echo
    fi
}

stop() {
    echo -n $"Shutting down ipmi: "

    # Stop doesn't actually do anything because we currently don't
    # unload ipmi modules on stop.  That might change in the future
    # if we decide unloading the ipmi modules is safe.
    RETVAL=0

    if [ $RETVAL -eq 0 ] && rm -f /var/lock/subsys/ipmi ; then
        echo_success
        echo
    else
        echo_failure
        echo
    fi
}

dostatus() {
    # Extract cpu temperatures from ipmitool output.

    # Abort if we don't have the ipmitool program.
    if ! /usr/bin/ipmitool -V >/dev/null
    then
        echo "/usr/bin/ipmitool not found!" >&2
        exit 1
    fi

    # Abort if ipmi isn't loaded.
    if ! grep -q ipmidev /proc/devices
    then
        echo "ipmi not listed in /proc/devices!" >&2
        exit 1
    fi

    # Check if we are running on a v1.0 IPMI system, and
    # change our processor search string appropriately.
    if /usr/bin/ipmitool -I open bmc info | \
        grep -q "IPMI Version.*1.0"
    then
        IpmiVersion="1.0"
    fi

    # Determine # of running processors
    NumProcs=$(grep -c processor /proc/cpuinfo)
    if [ $NumProcs -eq 0 ]
    then
        echo "Can't determine number of processors!" >&2
        exit 1
    fi

    # Now build the query string.  Concatenate it into
    # one string because that's more efficient on 2.4 systems.
    Count=1
    TempString=""
    while [ $Count -le $NumProcs ]
    do
        if [ x$IpmiVersion  = x"1.0" ]
        then
        TempString="$TempString CPU\ $Count"
        else
        TempString="$TempString Processor$Count\ Temp"
        fi
        Count=$((Count + 1))
    done
    # building TempString like this and eval'ing it is ugly, but
    # it's the only way I could make the quoting work.  Sorry.
    TempString="/usr/bin/ipmitool -I open sensor get $TempString"
    eval $TempString | awk -v "c=$Count" '
BEGIN {
   n = 1
}
/Sensor Reading/ {
   printf "CPU%s Temp: %s\n",n,$4
   n++
}
END {
   if ( n != c) {
    printf "Error: found %s CPUs, but got temps for %s\n",--c,--n >"/dev/stderr"
    exit 1
   }
   exit 0
}'
    RETVAL=$((RETVAL + $?))
    return $RETVAL
}

restart() {
    stop
    start
    RETVAL=$?
}

condrestart() {
    [ -e /var/lock/subsys/ipmi ] && restart || :
}

remove () {
    # Actually remove the drivers.  Don't do during stop in case
    # this causes system to become unstable (a la lm_sensors)
    if /sbin/lsmod | awk '{print $1}' | grep -q ipmi_
    then
        # Try removing both 2.4 and 2.6 modules.
        /sbin/rmmod ipmi_si 2>/dev/null
        /sbin/rmmod ipmi_kcs_drv 2>/dev/null
        /sbin/rmmod ipmi_devintf
        /sbin/rmmod ipmi_msghandler
    else
        echo "No ipmi modules loaded!" >&2
        RETVAL=1
        return $RETVAL
    fi

    # Wait a sec to give modules time to unload.
    sleep 1

    # Check if we failed to remove any modules, and complain if so.
    if /sbin/lsmod | awk '{print $1}' | grep -q ipmi_
    then
        echo "ipmi modules still loaded!" >&2
        RETVAL=1
        return $RETVAL
    fi
}

# See how we were called.
case "$1" in
  start)
    start
    ;;
  stop)
    stop
    ;;
  status)
    dostatus
    ;;
  restart|reload)
    restart
    ;;
  condrestart)
    condrestart
    ;;
  remove)
    remove
    ;;
  *)
    echo "Usage: ipmi {start|stop|status|restart|condrestart|remove}"
    exit 1
esac

exit $RETVAL