im sure this horse has been beaten to death but i cannot for the life of me figure out how people are making this work. using standard failover using ping it works fine, as soon as i try to do any kind of recursive anything it fails to work properly at all.
I have two DHCP interface ports on my RB5009 router and im running the latest firmware of 7.14.2. Port 1 is my fiber internet modem which is DHCP reserved to 192.168.1.218, port 3 is my failover 5g Tmobile router that is “locked” so it must be DHCP only, no static IP address.
I have seen so many different posts on using netwatch, mangle, scripts of all sorts and none of them really worked. The netwatch didnt work because it pinged the same DNS server for both the failover and the main, which caused it to flip flop constantly. Setting up recursive routes didnt seem to work properly either (not sure why?).
Suggestion #1 is to post the export of the actual configuration that “does not work”. A single typo or misunderstanding can break it all.
Suggestion #2 is to elaborate on what exactly “does not work” means, although since you say that “using ping” (in a periodically scheduled script?) it works, I guess you are aware of things like existing connections not failing over etc.
Using the same “canary address” for both WANs is indeed a bad idea, regardless whether you combine it with a handmade script, netwatch, or recursive next-hop search. Plus, as you call the 5G WAN uplink a failover one (i.e. you do not plan on load distribution), it only makes sense to monitor its availability if you have got a notification channel that will warn you if that uplink goes down. For the failover itself, knowledge of the status of the backup link is useless, as if the primary link fails and the backup one is also unavailable, there is nothing left to do anyway.