How to configure failover in edgerouter x 5 ports?

Question

I have connected to my edgerouter x 5 ports, 2 WAN (WAN1 to eth0 and WAN2 to eth1 in failover)

Problem: WAN1 has 10 Mbps and WAN2 Failover 8 Mbps. When WAN1 drops, but not enough (example: 1 or 0.5 Mpbs), WAN2 failover does not enter.

What I want is that when WAN1 drops to XX Mbps, WAN2 goes into operation in failover mode and returns to WAN1 when it exceeds XX Mbps, in a recovery time of XX minutes

Where XX are values that I determine manually

My edgerouter x:

I did not find the information in the ubiquiti user manuals for this router HERE, HERE or HERE

I appreciate the help.

PS: I apologize if this is not the right forum to ask this question. I searched on stackexchange but didn't find any hardware to post this question

*What* parameter do you want to trigger on? When available bandwidth is under 2Mb/s? How can you know if available BW is <2Mb/s? — vidarlo, May 09 '23 at 16:07
The question describes the scenario. What I want to activate the failover when I want, according to my parameters and not when the router decides. — acgbox, May 09 '23 at 16:28
I suppose that it will be necessary to take them from the RX/TX (although these values do not represent reality, but it is close). But I really wouldn't know — acgbox, May 09 '23 at 20:35
You can't do that. You don't know why you have low speed. It can be a video feed that just happens to consume 400Kb/s, or it can be a download that is slow. The problem is simply that the router has ***no way of knowing*** the parameter you want to act upon. This is not a trivial problem; it's fundamental. — vidarlo, May 09 '23 at 20:50

Peter Zhabin · Answer 1 · 2023-05-09T18:53:05.987

There's no such thing as Ethernet running at 2Mbps. If you have some other means of getting a bandwidth report reliably (e.g. DSL or 3G modem status page) you can incorporate this check in a bash script, save it somewhere in /config/scripts so it will survive firmware upgrades and then call it in your load-balance stanzas like in the example below.

In this example WAN2 is on eth2 and is a backup, while WAN1 is on eth1 and its the primary link. You would also need to create routing tables 10 and 20 with the appropriate defaults in each and reference them in firewall modify rules:

load-balance {
group WAN2 {
    exclude-local-dns disable
    flush-on-active disable
    gateway-update-interval 20
    interface eth1 {
        failover-only
        route {
            table 10
        }
        route-test {
            initial-delay 180
            interval 60
            type {
                script /config/scripts/pinger
            }
        }
    }
    interface eth2 {
        route {
            table 20
        }
        route-test {
            initial-delay 180
            interval 60
            type {
                script /config/scripts/pinger
            }
        }
    }
    lb-local disable
    lb-local-metric-change disable
}
group WAN1 {
    exclude-local-dns disable
    flush-on-active disable
    gateway-update-interval 20
    interface eth1 {
        route {
            table 10
        }
        route-test {
            initial-delay 180
            interval 60
            type {
                script /config/scripts/pinger
            }
        }
    }
    interface eth2 {
        failover-only
        route {
            table 20
        }
        route-test {
            initial-delay 180
            interval 60
            type {
                script /config/scripts/pinger
            }
        }
    }
    lb-local disable
    lb-local-metric-change disable
}
}

The script is called with three parameters:

#!/bin/bash
targets=(
    '192.168.10.1'
    '192.168.20.1'
    '192.168.30.1' )
if [ $# != 3 ]
then
  echo "Usages: $0 <group> <intf> <status>"
  exit 1
fi

group=$1
intf=$2
status=$3

for host in "${targets[@]}"
do
  /bin/ping -n -c 1 -W 1 -w1 -I $intf $host
  if [ $? == 0 ]
    then
      exit 0
  fi
done

# fail

exit 1

This script uses ping via interface provided by EdgeOS upon the script startup to check availability of three uplink hosts and returns 0 (ok) if at least one of them responds and 1 (failure) when no hosts respond.

You could use the same approach to implement some logic so that when the $intf has a good status according to your checks, return 0 with exit 0, otherwise return 1 with exit 1.

So, according to your answer, the "Trigger Level" and "Recovery Time" failover options do not exist in the Edgarouter-5-ports configuration. — acgbox, May 09 '23 at 14:43
There's no such thing in EdgeOs in general, as far as I'm aware. Built-in functionality for load balancing is `route-test type ping` with success and failure counters triggering failure and recovery. But you can alter this with `route-test type script` as I described in my answer and implement your own logic in this script. — Peter Zhabin, May 09 '23 at 15:33
I think you should explain how your script works and why you set those values — acgbox, May 09 '23 at 16:09
I have included a complete script we use here to monitor dual-wan networks on EdgeOS together with a brief explanation of the logic behind it. — Peter Zhabin, May 09 '23 at 18:54

How to configure failover in edgerouter x 5 ports?

1 Answers1