Question

I undertand that the higher the interfaces are in the list, the more priority they have.

But how does the failover feature work with this?

Without failover, I'd assume the higher interface has the priority and if it goes offline, the next interface below will come online?

If so, what is the point of the failover feature?

As an example, we have:
Wireless Client = Primary connection when in our premises
Mobile connection = Secondary connection for when the RUTX11 isn't in the building.

So when the vehicle, leaves the building, it loses the wireless connection (Building SSID) and then the mobile should then become the default wan? Does the RUTX11 wireless client then keep scanning for the building SSID until it is detected and re-associates itself to become the primary connection again? If so, why have the failover function? Or am I missing something here?

Answer 1

Hello,

   

The main difference between Failover and the interface metric selection (Network → Interfaces) is that interfaces themselves cannot detect when there is no internet connection on the link. It can only detect whether the connection is up or down (cable connected or disconnected). What failover does, is it pings the specified host and checks if the pings go through.

So for example, when the router is connected to an operator, but the operator is not actually able to provide an internet connection, the interface will not do anything, but failover will try to switch to another interface (if available). Failover is commonly used in places where downtime can be catastrophic. For example, if a business has a fiber connection and a mobile connection for a backup, and the fiber connection does not have internet access but is still plugged into the router and the router uplink is active, failover will be needed to use to switch to a mobile connection, as setting an interface metric here will not help.

In your use case, the only advantage of using failover would be if the router is connected to your WiFi, but the WiFi network does not have internet access, the failover will see that the pings are not reaching the destination and the router will switch to a mobile interface.

  

Hope this clears it up!

Best regards,
DaumantasG

Comments

Sorry, this is definitely wrong. Recent mwan3 can use ping to detect problems with internet access. Just a question of configuration. Thus, it detects connections problems, like you decribe for "failover".

Modern mwan3 switches to other interface, in case it detects "disconnected", which means, configurable number of failed pings on actual highest prio interface. Switchover (also) done using packet marks, and special routing.

I have to admit, that I am running official openwrt on a large fleet of RU955, having 3 interfaces for failover (wan, wifiwan, wwan), managed by mwan3. May be, the Teltonika stuff uses an outdated, very old version of mwan3.

---

I have edited the last post, mwan3 is indeed what we call failover. We just have its configuration in a different menu from the interfaces.

---

Ok, that makes sense.... in my scenario above, how does the connection switch back to the wireless client after it has failed over to the mobile 4G network? Does the client continue to poll for the building SSID it should connect to and once connected, it becomes the primary connection again?

So interface set up order is:
1. Primary connection - RUTX11 Wireless client set to failover and ping 2 internal hosts (when on premises)
2. Secondary connection - RUT11X mobile 4G (no failover) when out and about

---

Hello,

  

Exactly, as both of these connections can be active at the same time, it will simply switch back to WLAN as soon as it is available.

As for the failover options, both interfaces must be enabled with the failover in order to work properly back and forth.

  

Best regards,
DaumantasG

---

As for the failover options, both interfaces must be enabled with the failover in order to work properly back and forth 

Now that bit does make me think with regards to this setup. I'm wondering if I should set the mobile side up with failover and ping some internet host eg 8.8.8.8 and then block that on our internal firewall?

So when the vehicle is:

1. On premise - Wireless client is active with failover enabled and pinging internally. Mobile client is down as it can't ping public ip address eg 8.8.8.8

2. When not on premise - Wireless client is not active due to losing connectivity with internal client. Mobile client is active as it can ping public ip address eg 8.8.8.8

And if so, do I need to have 2 seperate failover policies for each of the above with each of them opposite to each other?

---

Hello,

  

I think you misunderstood the failover service. I've re-created your setup to explain it a little better.

Let's say we have two WAN interfaces - WiFi and mobile. WiFi has a metric (priority) of 1, while mobile connection will have a metric of 2 (the lower - the bigger preference):

It should also be noted, that failover needs to be selected in the top right corner.

These are the settings used inside both of the interfaces:

So, let's say you're leaving the building with the RUTX11. The router will send pings every 3 seconds. After the WiFi disconnects, pings will stop going through, thus the router will assume that the connection was lost and will switch to the mobile interface.

Since both of these interfaces can be active at the same time (unlike the SIM cards), as the router reconnects to the building's WiFi, it will automatically start using the WiFi, as it has a higher priority.

More information can be found here.

  

Best regards,
DaumantasG