Multi-WAN Load Balancing: How Policy Rules and WAN Failover Actually Work

Three internet lines, one router, and still everyone complains. That is the moment most network admins finally open the Multi-WAN module. The idea is simple: define a few line solutions, then tell the router which traffic should use which one. But the part that trips people up is not the balancing itself — it is the policy layer and the failover setting. Let’s look at both.

Multi-WAN solution and policy configuration screen

1. Line Solutions Only Work When Policies Point Traffic At Them

You start by creating line solutions. Maybe you want one dedicated WAN for business servers, a second solution that balances two lines for office users, and a third that sends specific websites out through a single link. Each solution holds its own load balancing rank per interface. So far, so good.

Here is the catch: a solution does nothing on its own. Traffic only lands in it because a policy matched. A policy answers three questions — where the traffic comes from, where it is going, and which ports it uses.

  • Source — choose Any, Single IP, or IP Range. Single IP accepts both a host address like 192.168.1.10 and a subnet like 192.168.1.0/24, which is handy when you want a whole floor or department on one line.
  • Target — same syntax options for an IP or domain, plus a Customize mode that lets you pick by application type, web categories, countries and regions, or custom IPs and domains.
  • Ports — the destination ports the rule applies to.

The single most important behaviour to internalise is ordering. Rules are processed from the top of the list downward, and the first match wins. That means a broad “any source, any target” rule sitting at the top will quietly swallow every rule beneath it. When a policy appears to be ignored, nine times out of ten it is simply being outranked. Drag the re-order icon to move rules up or down, and re-test after every change.

A practical habit: put the narrow, high-priority rules first — the business server that must stay on its own line, the site that must always exit through a specific WAN — and let the general “everyone else, balanced” rule sit at the very bottom as the catch-all.

2. WAN Failover Is Off By Default — Turn It On

Load balancing distributes traffic, but it does not protect you when a line dies. That is a separate feature, and it is disabled unless you enable it.

When WAN Failover is active, the router continuously checks line availability. If an interface stops responding, the load weighting is re-balanced across the remaining lines so users keep working instead of staring at a spinner.

  • Enable reconnect and the WAN interface will be automatically restarted after a failure — useful for flaky PPPoE lines that recover on their own.
  • For a static IP WAN interface, the system automatically pings the WAN gateway to verify the link.
  • Prefer IP addresses over domain names for connection testing. A DNS-based check can fail for reasons unrelated to the line itself, which triggers unnecessary failovers.

WAN failover detection settings

The same advice applies to the detection targets: a public, stable IP that you control or trust is a far better health signal than a hostname that depends on a resolver working.

Get these two pieces right — ordered policies that send the right traffic down the right solution, and failover that catches the line that drops at 2 a.m. — and the multi-WAN setup stops being a source of tickets and starts being something you forget about.

Source: http://wiki.wfilterngf.com/Mwan