What are the implications of not using interfaces when defining policies on fortigate firewalls?


Policy would be

edit 1
    set srcintf "any"
    set dstintf "any"
    set srcaddr "ip-"
    set dstaddr "ip-
    set status enable
    set schedule "always"
    set service "TCP-8080"
    set logtraffic disable

Is there any benefits of specifying the interface?

My understanding is the fortigate firewalls apply RPF checks against traffic when it enters the firewall.

For example if packet from via port1, It would expect a route in the routing table to exist for that prefix if not RPF would block it.

We are looking at automating how we populate firewall rules and removing the interface logic would make it a a lot simpler to implement.

Any thoughts or experiences would be good.

  • Did any answer help you? If so, you should accept the answer so that the question doesn't keep popping up forever, looking for an answer. Alternatively, you could provide and accept your own answer.
    – Ron Maupin
    Feb 21, 2018 at 16:32

2 Answers 2


The interface is part of the policy. When you use any, the rule applies to all interfaces, ie. it doesn't matter where the packet enters or is bound to exit the firewall for the policy to be applied.

FGs apply RPF - in order to accept a source address on an interface there must be an appropriate route out of that interface. So, essentially you're not required to use interfaces in the policies at all times.

RPF can be disabled by turning on asymmetric routing (config system setting, set asymmetric enable), disabling stateful inspection on the way. A better way may be to set a route with a high metric that won't ever get used.


There is hardly ever a good or mandatory reason not to specify the interfaces in a policy. There is always a price you pay for convenience: loss of control and time while maintaining and debugging.

Try to determine the flow of traffic through, say, 80 firewall policies most of which use the 'any' interface. Which one is 'shadowing' others, and are explicit policies hit at all, if not, why? This is near to impossible. Of course you can sniff but that's in realtime. Firewalls are a lot about planning and foreseeing.

To put it differently: policies reflect the flow of data (and it's protection) in the way you have planned and foreseen it. Firewalls are built to prevent deviations from that plan. That's why there are implicit DENY policies for each interface pair in FortiOS, as an example. You do have the information about which network exists (or is routed to) behind each interface. In not using this information you prevent the firewall from attaining it's full potential.

The loss of RPF is IMHO only a side effect of the 'any' interface. Usually, if the protected networks are connected to the internet you need a default route. This route is used all the time if you use an 'any' interface, even for traffic of unknown origin. Effectively, this eliminates RPF.

  • You don't lose RPF because of any interfaces in the policies.RPF is based on the present routes and it needs to be explicitly deactivated.
    – Zac67
    Nov 4, 2017 at 18:45
  • As you've posted, RPF checks if there is a route via the ingress interface to the source address. If you use the 'any' interface it will only check if there is a route - which for unknown traffic will be the default route. It's not disabled but ineffective. Nov 4, 2017 at 18:48
  • Policies are not the basis for RPF but the routes are. Please check the manual.
    – Zac67
    Nov 4, 2017 at 18:54

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.