Secondary WAN is up and running in MT, but it’s not failing over from the non-working WAN, despite “check-gateway” being enabled.
Then I tried to ping the dead WAN’s gateway, and sure enough it’s responding. So basically the gateway is up, but it’s further down the line that the link is down. However, MT doesn’t know since it’s only pinging the gateway.
For now I just disabled the primary WAN so I can be up and running.
So my question is, is it possible to make “check-gateway” ping a public, outside IP instead of the gateway to detect a down link more reliably?
To answer your question: YES!
I have a similar setup (except my backup WAN is also my email provider and thus the additional line for the email server IP).
Basically a recursive method that checks to see if TWO public IPs are available, first it looks to determine if google1 is available and if thats down it checks if DYNDNS1 is available and vice versa.
To do this check it routes through the Primary WAN and if the Primary WAN is down it uses (Distance=10) the backup WAN to do the recursive looks.
As soon as the primary is backup (ping works to the DNS public locations from the Primary WAN), the routing switches back to the Primary).
(I will add now my email involved line as I didnt want it distracting from the above)
add comment=Email_bypass distance=1 dst-address=aa.bbb.e.ff/32 gateway=
aa.ccc.dd.1
Thanks a lot for sharing this! I just wonder, will there be any performance penalties with this type of setup? Like, how often does MT evaluate the gateways? Is it doing it in the background on a set interval?
I dont have a clue but I dont think it will be much though. You could always monitor the load with no other traffic going on I suppose. Its a pretty efficient way of doing this as far as I know.