Hi all, I made a recent switch to have my F5s be BGP peers with my upstream routers in order to leverage ecmp and have proper load sharing among the units. This has a drawback in that if pools go down the virtual address is withdrawn from routing and as a result fallback host in the http profile (or even irules) is no longer an option. Has anyone done a similar setup and found a workaround? I'm running through a few options in my head but curious to hear other ideas.
Ironically there is/was a bug where routes do not get withdrawn when the pool is down and that's what i'm looking for an option.
Can't you just add a route to your BGP configuration that sends your application traffic to the fallback host, but with a weight that means the route won't be used if any of the routes pointing to your virtual server(s) aren't available?