Weird problem: related to load balancer?

[dzeanah]

I'm not sure if this is pfSense related or not, but here goes:

If I'm within my network and I take down the web servers that are configured as part of the load balancing pool, I'll eventually get sent to https://mywebsite, which points to pfSense.

Outsiders seem to get the same issue, where they're directed to the https: version of my domain (which I don't support), and that times out.

Once the web servers are responding again the users' web browsers sometimes still point to the nonexistent https:// version, at least until they restart, clear their cache, or wait for the problem to go away.

Is there something I can do to eliminate this problem?  Right now the fall back pool points to the same pool that failed, so this may be part of the problem.  Previously it directed to a page that said "the server is down, come back in a few minutes" but that offered a similar problem: when the server was actually back up, many users would still see the old page for a day or so.

Thoughts on how best to get around this, other than "don't let the servers drop?"  Sometimes I need to bring my cluster down…

Thanks.

[marcelloc]

Change pfsense gui port and disable gui redirect rule