[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: /etc/hosts and system entries



Perhaps the best course of action is then to modify the local mail server (sendmail/postfix) to send out mail as the fully qualified domain name?

Then again, what to do in the situation of DHCP.

I guess that if mail is expected to be delivered locally then it's not a big deal.

but if you're running a bunch of servers you are going to want to send the emails to a central location.

I guess it sounds like a problem to solve outside of the hosts file


(sorry for the thinking out loud/rambling)


David Cantrell wrote:
On Thu, 27 Sep 2007 10:10:42 -0400
Simo Sorce <ssorce redhat com> wrote:

On Thu, 2007-09-27 at 10:02 -0400, David Cantrell wrote:
On Thu, 27 Sep 2007 09:59:16 -0400
Harry Hoffman <hhoffman ip-solutions net> wrote:

So, /etc/hosts comes setup by default (i.e. after kickstart install)

# Do not remove the following line, or various programs
# that require network functionality will fail.
127.0.0.1               localhost.localdomain localhost

I'm fairly certain to not too long ago (redhat-9 perhaps) the hostname of the system was also added to the localhost entry:

127.0.0.1  my.host.com my localhost.localdomain localhost


This had the distinct advantage that when apps (i.e. yum-updatesd) sent mail from the system via a mail host then address would appear as:
root my host com  instead of root localhost com

Am I remembering correctly, in terms of how I believe it used to be? If so, anyone know why it changed?
https://bugzilla.redhat.com/show_bug.cgi?id=253979

Fixed in rawhide.

Why it changed...don't know, but I'll take the blame since I'm responsible for a lot of the network gutting and rewriting in anaconda.  Most likely a mistake on my part.

Please, PLEASE, reconsider.
I've long hated this thing of assigning the hostname to 127.0.0.1, it
always breaks when using kerberos/winbindd as the hostname needs to
reflect the public facing ip.

I personally think that Gnome is at fault here, is there any smarter way
to at least change the hostname mappingi hosts when the main network
interface gets an IP?

OK, now this is making sense as to why it changed before.  It's incorrect.

So here we are where a certain group of people want the hostname added to the 127.0.0.1 line and another group that doesn't.  I tend to agree with the latter, but I would rather explore this post F-8 than now.  Removing it, yet again, in rawhide after F-8 is released and any bugs that get opened we determine the program that's at fault and reassign the bug to that package.  It'll be annoying to users, but I think it's best to not have the hostname on the 127.0.0.1 line.




[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]