rbc

#!/bin/bash

RBC_VERSION="1.3"

PATH=/usr/local/sbin:/usr/local/bin:/sbin:/bin:/usr/sbin:/usr/bin:/root/bin

EMAIL_OWNER=0
QUIET=0
REBOOT_ALLOWED=0
REBOOT_REQUIRED=0
SUMMARY=0
SYSLOG=0
MANUAL_REBOOT=1
REBOOT_DOW="X"

die () {
    printf "Error, exiting: " 1>&2
    echo "$@" 1>&2
    exit 255
}

syslogit () {
    logger -t rbc -p auth.warning -f "${1}"
}

usage () {
    printf "%s" "$0" 1>&2
    printf " usage summary:\n" 1>&2
    printf "   -h: (or --help) print usage summary\n" 1>&2
    printf "   -l: log to SYSLOG facilities\n" 1>&2
    printf "   -m: email reboots@localhost (set the alias to email externally)\n" 1>&2
    printf "   -q: suppress output to STDOUT\n" 1>&2
    printf "   -r: reboot the system in 30 minutes if reboot is required, SYSLOG is forced\n" 1>&2
    printf "   -s: print a one line status summary in CSV, with the following format\n"
    printf "        hostname, OS release, uptime, auto-reboot disabled, RBC status code, reboot day of week\n" 1>&2
    printf "   -V: display version and exit\n" 1>&2
    printf "   EXIT codes:\n" 1>&2
    printf "%14s %-70s\n" "0:" "reboot not required" 1>&2
    printf "%14s %-70s\n" "1:" "reboot required, kernel only" 1>&2
    printf "%14s %-70s\n" "2:" "reboot required, processes only" 1>&2
    printf "%14s %-70s\n" "3:" "reboot required, kernel & processes" 1>&2
    printf "%14s %-70s\n" "254:" "printed help summary and exited" 1>&2
    printf "%14s %-70s\n" "255:" "failed to run due to other error" 1>&2
}

version () {
    printf "RBC version ${RBC_VERSION}\n"
}

email_owner () {
    [ $REBOOT_REQUIRED -eq 0 ] && SUBJECT="${HOSTNAME}: reboot not required"
    [ $REBOOT_REQUIRED -eq 1 ] && SUBJECT="${HOSTNAME}: reboot requried"
    if [[ $REBOOT_ALLOWED -eq 1 && $REBOOT_REQUIRED -eq 1 ]]; then
	SUBJECT="*** ${HOSTNAME} will reboot in 30 minutes ***"
    fi
    
    mail -s "${SUBJECT}" reboots@localhost < "${1}"
}

for X in mail hostname lsof wc column awk rpm sed head uname dd tr cut date logger shutdown; do
    which "${X}" >/dev/null 2>&1 || die "required application ${X} not found"
done

DATEY="$(date +%Y%m%d-%H%M%S)"
HOSTNAME="$(hostname -s)"
TMPFILE=/tmp/"$(dd if=/dev/urandom bs=1 count=256 2>/dev/null |tr -c -d '0-9a-f' |cut -c-7)"

for X in ${1:+"$@"}; do
    case "${X}" in
	(-h|--help)
	    usage
	    exit 254
	    ;;
	(-V|--version)
	    version
	    exit 0
	    ;;
	-l)
	    SYSLOG=1
	    ;;
	-m)
	    EMAIL_OWNER=1
	    ;;
	-q)
	    QUIET=1
	    ;;
	-r)
	    REBOOT_ALLOWED=1
	    SYSLOG=1
	    ;;
	-s)
	    QUIET=1
	    SUMMARY=1
	    ;;
	*)
	    usage
	    die "invalid parameter: ${X}"
	    ;;	   
    esac
done

[ "$TMPFILE" == "/tmp/" ] && die
echo "### ${HOSTNAME}: Reboot automation notice: ${DATEY} ###" > "${TMPFILE}"
lsof +c0 -b -d DEL 2>/dev/null | \
    grep -v '\s/SYSV\|/\[aio\]$\|/dev/shm/.\+\|DEL.\+/tmp/.\+\|/var/lib/sss/mc/\(initgroups\|passwd\)' | \
    grep -v 'DEL.*\.config/dconf/user$\|DEL.*/run/user/.*/dconf/user$\|pulseaudio.*DEL.*/run/user' | \
    grep -v 'Xwayland.*DEL.*/run/user/[0-9]\|java.*DEL.*/var/crashplan/jna-.*tmp$\|caja.*DEL.*gvfs-metadata' | \
    grep -v 'tuned.*DEL.*/ffi\|DEL.*/memfd:pulseaudio\|DEL.*/var/lib/sss/mc/.' | \
    grep -v 'firewalld.*DEL.*/ffi\|rhsmd.*DEL.*/ffi\|DEL.*/memfd:xorg\|DEL.*/memfd:dbus' | \
    grep -v 'DEL.*/memfd:xshmfence\|DEL.*/#[0-9]*\|DEL.*/i915\|memfd:\(mozilla-ipc\|pipewire\)' | \
    grep -v 'DEL.*/memfd:libffi\|DEL.*/home/.*/.local/share/gvfs-metadata|DEL.*/jenkins/.cache/' | \
    grep -v '/.cache/JNA/temp/\|/usr/lib/fontconfig/cache/\|/usr/share/.*\.cache' | \
    awk 'NR==1 || !/dev\/zero/ {print $2,$1,$4,$NF}' | column -t >> "${TMPFILE}"

if [ "$(wc -l < ${TMPFILE})" -gt 2 ]; then
    (( REBOOT_REQUIRED+=2 ))
else
    echo "### ${HOSTNAME}: Reboot automation notice: ${DATEY} ###" > "${TMPFILE}"
fi

C_KERN="kernel-$(uname -r)"
A_KERN="$(rpm -q --last kernel |head -1 |sed 's|\s\+.*$||')"

if [ "$A_KERN" != "$C_KERN" ]; then
    (( REBOOT_REQUIRED++ ))
    printf "Running %s, but %s available\n" "$C_KERN" "$A_KERN" >> "${TMPFILE}"
fi

echo "### ${HOSTNAME}: End of reboot automation notice: ${DATEY} ###" >> "${TMPFILE}"

[ $REBOOT_REQUIRED -eq 0 ] && printf "\t %s: Reboot is not required.\n" "${HOSTNAME}" > "${TMPFILE}"
[ $SYSLOG -eq 1 ] && syslogit "${TMPFILE}"
[ $EMAIL_OWNER -eq 1 ] && email_owner "${TMPFILE}"
[ $QUIET -eq 0 ] && cat "${TMPFILE}"

[ -f "${TMPFILE}" ] && rm -f "${TMPFILE}"

if [ $REBOOT_REQUIRED -ne 0 ]; then
    [ $REBOOT_ALLOWED -eq 1 ] && shutdown -r +30 "System reboot is required.  To cancel use shutdown -c.  But don't because this system needs to be rebooted"
fi

if [ $SUMMARY -eq 1 ] ; then
    if [ -f /etc/cron.d/auto-reboot ] ; then 
	if grep -q '^\s*[0-9].*/usr/.*/\?sbin/rbc.*-r.*$' /etc/cron.d/auto-reboot 2>&1 ; then MANUAL_REBOOT=0 ; fi
	if [ $MANUAL_REBOOT -eq 0 ] ; then REBOOT_DOW="$( awk '{ print $5 }' < /etc/cron.d/auto-reboot)" ; fi
    fi
    # get the following the hard way since RHEL 6 uptime does not have the -s flag :(
    UPSINCE=$(date -d @$(( $(date +%s) - $(sed 's|\..*$||' /proc/uptime) )) +%Y-%m-%d\ %H:%M:%S)
    printf "%s,%s,%s,%s,%s,%s,%s\n" \
	   "${HOSTNAME}" \
	   "$(sed 's|[Rr]ed.*inux|RHEL|;s|\s*[Ss]erver||;s|release\s*||' /etc/redhat-release)" \
	   "$(uname -r |sed 's|\.[^.]*$||;s|\.[^.]*$||')" \
	   "${UPSINCE}" \
	   "${MANUAL_REBOOT}" \
	   "${REBOOT_REQUIRED}" \
	   "${REBOOT_DOW}"
fi

exit $REBOOT_REQUIRED