round-robin and multi MX practicalities
kcd at daimlerchrysler.com
Thu Mar 23 22:41:16 UTC 2000
Barry Margolin wrote:
> In article <8989mn$bpm$1 at nyheter.crt.se>,
> <peter at icke-reklam.manet.dot..nu> wrote:
> >David Zade <dzade at rcn.com> wrote:
> >> Beware, I do not believe roundrobin will work for failing over to a backup
> >> server. It just hands out the addresses in rotating order. It will still
> >> handout the address of the dead server.
> >Correct. Any decent application however will detect that this is a dead
> >server and try the next address given in the reply.
> >(remember that all addresses is returned in the dns reply, only the order that
> >changes with round-robin)
> Well, most applications are probably not "decent", by your definition.
> >Also correct. However even here you can create a 'A' record that consist
> >of several addresses. Again, any decent application will detect dead
> >servers and go on with the next.
> Do Netscape, IE, and/or Eudora do this? I haven't checked, but my guess is
> that they aren't all decent, and I wouldn't be surprised if they were all
Netscape 4.5 appears to do failover. Don't know about the others. In fact,
Netscape 4.5 seems to *remember* the failing address -- after an initial pause,
web accesses to the same round-robin name seem to occur at normal speeds (yes,
I cleared my local cache while testing this, and I wasn't going through any
proxies). Or maybe it's just my imagination...
More information about the bind-users