On Wed, Jan 21, 2004 at 11:52:39PM -0500, Paul G. Weiss wrote: > failure: > Sorry._Although_I'm_listed_as_a_best-preference_MX_or_A_for_that_host, > /it_isn't_in_my_control/locals_file,_so_I_don't_treat_it_as_local._(#5.4.6)/ > > I've been seeing a bunch of these in the log, and I finally understand why > they're there, but I'm not sure that they should be. > > Consider the email domain xyz.com. Suppose that the DNS for this domain > is: > > xyz.com. mx 10 server.abc.com. > xyz.com. mx 20 server.pqr.com. > > Suppose I have a qmail server at server.pqr.com. > > I have xyz.com in my rcpthosts but not my locals. In other words, I'm > running backup for the server.abc.com server. > > Normally everything works fine. > > But if the domain server for the abc.com domain goes down so that the > server.abc.com does not resolve, I get the message above. > > Shouldn't qmail return a temporary failure so that the message is held in > the queue until it can resolve the name? > > If instead of relying on DNS for the MX record, I add the line: > > xyz.com:server.abc.com > > to smtproutes then if the abc.com name server is down I get: > > failure: Sorry,_I_couldn't_find_any_host_named_server.abc.com._(#5.1.2)/ > > and even if I have an entry for server.abc.com in the /etc/hosts file I > continue to get the message. > > The only way I can reliably queue the message is if I put the IP address > of server.abc.com in the smtproutes file, which I think is a little > draconian. > > What do you think? >
This is why you should have two distinct and redundant name servers for a domain. So that if one is down the other still resolves your domain. We can't change anything here. The only real solution is to add a secondary name server for your setup. -- :wq Claudio
