Reboot a Raspberry Pi if LAN is unreachable

lan_watchdog.sh

#!/bin/bash
#
# Step 1) mkdir -p /root/bin
# Step 2) vi/nano/pico /root/bin/lan_watchdog.sh and edit the line WIFI_GW and add your own gateway IP to ping
# Step 3) chmod +x /root/bin/lan_watchdog.sh
# Step 4) Install at screen and arping `apt install screen arping at`
# Step 5) Add to /etc/rc.local BEFORE exit 0
#         # start lan_watchdog to reboot if connection is unavailable
#         cd /root/bin/ && echo 'screen -dmS lan_watchdog /root/bin/lan_watchdog.sh' | at now+10min
# Step 6) Make sure everything is connected and correct and reboot the Pi
#
# NOTE: I used the screen command to launch this from rc.local so I can login from time to time and check the output of the command.
#
# preference: use arping rather than ping since ICMP can be inadvertently firewalled in netfilter
# limitation: host to arping must be on the same subnet
# ping syntax should be very similar - change arping to ping if you want to check host on different subnet

# function gets called to do any cleanup before rebooting
reboot_host()
{
   # add an entry in syslog to keep track of down times
   logger -p0 "WIFI_GW is unreachable - going down for reboot"
   echo "WIFI_GW is unreachable - Rebooting" 
   reboot
}

# host we should always be able to reach when wi-fi is up
WIFI_GW="192.168.0.1"

# number of seconds to wait before checking arping again
SLEEP_SECS=120

# number of times arping is allowed to fail before a reboot
MAX_TIMES_DOWN=10

# number of consecutive times, in loop, that WIFI_GW has been unreachable
down_count=0

# loop forever
while [ 1 ]; do

   # check using two  arp packets - if arping returns non-zero status, assume it's an error
   arping -c 2 ${WIFI_GW}
   result=$?

   # if ping result is non-zero, add to the down counter
   #   if result is zero (successful) reset down_count (temp network hiccup?)
   test ${result} -ne 0 && let down_count="$down_count+1"
   test ${result} -eq 0 && down_count=0

   # if arping fails in the loop MAX_TIMES_DOWN (consecutively) then reboot
   test $down_count -gt ${MAX_TIMES_DOWN} && reboot_host

   # wait a while before checking again
   sleep ${SLEEP_SECS};

done