MX68W Failover

[Axeolito]

Hello guys,

I have a few question about how the failover of a MX68W works. 

- When I turn off the WAN1 (primary for us) router or disconnecter its fiber, the failover takes up to 5 min to work. Which is the "normal" behavior. (And its way too long btw). 

- But when I disconnect the RJ45 from the router to the Meraki, it looks like the failover is immediate to WAN2. 

Is there any difference in the meraki between those two actions ?

Thanks for helping and sorry for my possible english mistakes. 🙂

[ConnorL · Accepted solution]

You're describing the difference between a soft failover and a hard-failover.

• Soft Failover:  When both the HTTP and ICMP tests have been unsuccessful for a period of time that exceeds 300 seconds, the uplink will be failed over. Therefore, it can take approximately five minutes for failover to occur in the event of a soft failure (where the physical link is still up but provides no internet access).
  • MX <==> Modem <== dead link ==> Internet
  • In this example, the modem is still powered on and the physical connection between the MX and modem is still up, however the internet connectivity tests fail and the 5 minute timer begins
• Hard Failover: When the actual WAN link itself goes fully down, i.e. it's unplugged or the modem/device upstream is removed/turned off. As we detect the link has physically gone down, an immediate fail-over occurs
  • MX < ///// removed link ///////> Modem <==> Internet
  • In this example, the modem in unplugged from WAN1 of the MX, as we detect the link is down, we do an immediate failover

For more info, see: https://documentation.meraki.com/MX/Firewall_and_Traffic_Shaping/Connection_Monitoring_for_WAN_Failo...

[Axeolito]

Hello,

Thank you! It's very clear.

In the meantime, is it possible to lower the 300 seconds to something like 5 seconds ? 
300 seconds is waaaay to long for us... 😞 

[Brash]

I don't know the answer to this but if it can be done, I would suggest that 5 seconds is too short. The soft failover timer is designed to prevent small temporary issues, blips in connectivity etc. from causing constant failovers which may be more impactful to the end user experience.

I would look to aim for something like 30 seconds to 1 minute.

[Axeolito]

You are probably right. Maybe 30 seconds is fine. 
I don't have the answer at the moment anyway. 😛