init.d script for monitoring and adjusting route metrics of 2 WANs in a multi-WAN setup on OpenWRT. Used as a simple alternative to mwan3, where I just want failover and all traffic forwarded through wireguard vpn (mwan3 seems to struggle with this).
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

143 lines
4.2 KiB

#!/bin/bash
#wan failover script. ping multiple hosts and failover from primary to secondary after too many fails on primary
#how often to send pings
INTERVAL=1
#how many packets to send to each host
PACKETS=1
#how often (in seconds) to re-check primary wan while on secondary wan and primary had failed
RECOVERY_CHECK=300
#reliable hosts for checking are cloudflare, quad9, hgoogle
HOSTS=("1.1.1.1" "9.9.9.9" "8.8.8.8" "208.67.220.220")
PING_RESULTS=()
#interface and luci names of primary and secondary wan
WAN1_IFACE_NAME="eth0"
WAN2_IFACE_NAME="wwan0"
WAN1_LUCI_NAME="wan"
WAN2_LUCI_NAME="wanb" #currently unused
#vpn gareway
VPN_GATEWAY="139.99.201.65"
#get the current active wan interface, in case the script is started during failover
function get_active_wan {
local ROUTE=0
local CURRENT_METRIC=""
local ROUTE_IFACE=""
while [ $ROUTE -le 100 ]
do
ROUTE_IFACE=`uci get network.@route[$ROUTE].interface`
ROUTE_TARGET=`uci get network.@route[$ROUTE].target`
if [ "$ROUTE_IFACE" == "uci: Entry not found" ]; then
break
elif [ "$ROUTE_IFACE" == "$WAN1_LUCI_NAME" ] && [ "$ROUTE_TARGET" == "$VPN_GATEWAY" ]; then
CURRENT_METRIC=`uci get network.@route[$ROUTE].metric`
break
fi
let ROUTE++
done
if [ "$CURRENT_METRIC" == "0" ]; then
#current wan1 metric is 0, so it is set as primary
echo $WAN1_IFACE_NAME
else
echo $WAN2_IFACE_NAME
fi
}
#change the primary wan metric
function change_wan_metric() {
CHANGED=$SECONDS
local ROUTE=0
local ROUTE_IFACE=""
while [ $ROUTE -le 100 ]
do
ROUTE_IFACE=`uci get network.@route[$ROUTE].interface`
ROUTE_TARGET=`uci get network.@route[$ROUTE].target`
if [ "$ROUTE_IFACE" == "uci: Entry not found" ]; then
break
elif [ "$ROUTE_IFACE" == "$WAN1_LUCI_NAME" ] && [ "$ROUTE_TARGET" == "$VPN_GATEWAY" ]; then
uci set network.@route[$ROUTE].metric="$1"
uci commit
reload_config
break
fi
let ROUTE++
done
logger -t failover "`date`: Changed active WAN to $USINGWAN"
}
function debug {
local PING_ZIPPER=""
for (( i=0; i<${#PING_RESULTS[@]}; ++i)); do
PING_ZIPPER+="${HOSTS[$i]} ${PING_RESULTS[$i]}|"
done
logger -t failover "`date`: pings:$PINGS ($PING_ZIPPER), counter:$COUNTER, changed:$CHANGED, seconds:$SECONDS, usingwan:$USINGWAN, checkwan:$CHECKWAN"
}
#which wan is currently being used/checked
USINGWAN=$(get_active_wan)
CHECKWAN=$(get_active_wan)
#how many times all pings failed
COUNTER=0
#when wan was last changed
CHANGED=0
logger -t failover "`date`: Failover script started, current wan $USINGWAN"
while sleep $INTERVAL
do
if [ "$USINGWAN" == "$WAN2_IFACE_NAME" ] && [ $(($SECONDS-$CHANGED)) -gt $RECOVERY_CHECK ]; then
debug
CHECKWAN=$WAN1_IFACE_NAME
CHANGED=$SECONDS
#NOTE: we dont do a modulo check because and instead just reset CHANGED because pings can take longer than the interval
#during a failure so it can potentially take multiples of RECOVERY_CHECK even after primary wan has recovered
fi
PINGS=0
for (( c=0; c<${#HOSTS[@]}; c++ ))
do
RET=`ping -I $CHECKWAN -w 1 -W 1 -c $PACKETS ${HOSTS[c]} 2>/dev/null | awk '/received/ {print $4}'`
PING_RESULTS[c]=$RET
if [ "$RET" == "$PACKETS" ]; then
let PINGS++
fi
done
#debug if any pings fail
# if [ $PINGS -lt ${#HOSTS[@]} ]; then
# debug
# fi
# if all pings failed increase counter
if [ $PINGS -eq 0 ]; then
let COUNTER++
else
COUNTER=0
fi
# counter failed pings to all hosts multiple times, switch WAN
# NOTE: if both wan fail somultaneously, it will just switch back and forth. but not like it matters, neither would work anyway... lol
if [ $COUNTER -gt 2 ]; then
debug
if [ "$USINGWAN" == "$WAN1_IFACE_NAME" ]; then
#if failed and currently on wan, switch to wanb
USINGWAN=$WAN2_IFACE_NAME
change_wan_metric 2
elif [ "$USINGWAN" == "$WAN2_IFACE_NAME" ]; then
#if failed and currently on wanb, switch back to wan
USINGWAN=$WAN1_IFACE_NAME
change_wan_metric 0
fi
elif [ $COUNTER -eq 0 ]; then
if [ "$CHECKWAN" == "$WAN1_IFACE_NAME" ] && [ "$USINGWAN" == "$WAN2_IFACE_NAME" ]; then
#ping(s) successful, if on wan2 and wan1 has recovered return to wan1
debug
USINGWAN=$WAN1_IFACE_NAME
change_wan_metric 0
fi
fi
#set the checkwan back to the usingwan, in case it was set to the other wan for recovery check
CHECKWAN=$USINGWAN
done;