Appreciate the response. My trick is due to the internet routing, not really local redundancy--the IP's in question are routed on different backbones until we get the BGP-4 online--IF we can ever get everything coordinated between my backbones. Sigh... Things shouldn't be this hard.
Erik Brandsberg CIO, The Link On Sat, 4 Oct 1997 [EMAIL PROTECTED] wrote: > Synopsis: VirtualHost directives don't respond to all the IP's that a name > might resolve to. > > State-Changed-From-To: open-closed > State-Changed-By: dgaudet > State-Changed-When: Sat Oct 4 19:39:45 PDT 1997 > State-Changed-Why: > Now that I understand the entire problem ... this should work, > but even if it is broken it won't be fixed in 1.2. The vhost code > has been changed again in 1.3 and it should support this ... maybe > not in 1.3a1, but in 1.3b1 whenever we release it. > > Note that if redundancy is your concern there are other ways to > achieve the same thing. You only need one IP address, unless you're > running two servers -- in that case you could use m4 or perl to > generate two configs for them. If you want to use just one server > with two network interfaces then I suggest (and I suggest this for > all servers, even with one network interface) that you put all the > vhost ip addresses on the *loopback* interface rather than on > the ethernet interface(s). If you slap an entire net of addresses > on your loopback then you only need to tell your router where > to find that net ... and if one host should go down then you can > move to another host by changing the route on the router. > > In any event, sorry for the hassle. The vhost code sucks when > it comes to HTTP/1.1-Host based hosts. > > Dean > >
