ISP Redundancy - monitored hosts questions

[D_TK]

Good day everyone.

Setting up ISP redundancy in primary/backup mode for a few clusters and i'm wondering what the correct approach is for the advanced\monitored hosts config.  

primary link is a DIA circuit from AT&T , and backup link is a cable modem from comcast.  

For the monitored hosts, are these just universally reachable targets like 8.8.8.8??  Or more targeted like the AT&T DNS server for link 1 and comcast DNS server for link 2?

Appreciate any insight.

thanks.

[the_rock]

Just use google dns, I had few customers do it and works fine. Fortinet does same thing. But yes, you could use targeted ones as well.

[D_TK]

Thanks, but maybe i'm not understanding how this works.  Currently eth4 is the primary DIA link, and eth5 is the backup cable modem.  I added the cogent DNS server as the primary monitored host, and the comcast DNS server as the backup monitored host.  After pushing policy, i'm seeing the attached - ICMP requests to both of the monitored hosts sourced only from eth4.

I assumed i would see eth4 polling its monitored host, and eth5 polling comcast??   Does this seem correct?

thanks

[the_rock]

Im pretty positive what you got is right...you wont see anything on backup link, thats totally normal. Its sort of like if you configured say bgp on a cluster, show bgp peers would only show established on master, never on backup.

[Wolfgang]

If you run ISP redundancy in LoadSharing mode both links are using their configured monitored hosts for probing. With HA it works like mentioned @the_rock .

[D_TK]

Gents, thanks for the help, it seems like it's working normally.  Next time i'm at that location i'll pull the cable and see if it does what it should.

[Wolfgang]

@D_TK  With "fw isp_link <Name of ISP link> {up | down}" you can change the state of one of the ISP links to test failover.