Re: /etc/hosts format WAS [Re: host hostname not found]
On Fri 12 Sep 2014 at 22:29:22 +0400, Reco wrote:
> On Fri, 12 Sep 2014 11:23:56 +0100
> Brian <ad44@cityscape.co.uk> wrote:
>
> > > But the most interesting is why there's a need for the whole library to
> > > do the job if a couple of lines in /etc/hosts would do the job just
> > > fine.
> >
> > A few posts back in this thread there is
> >
> > [🔎] 20140906161207.GN4985@copernicus.demon.co.uk">https://lists.debian.org/[🔎] 20140906161207.GN4985@copernicus.demon.co.uk
> >
> > The link it contains leads to another with a recent -devel discussion.
> > There is a 10 year history with the issue; a search with "/etc/hosts"
> > and "Thomas Hood" should bring up some of it.
>
> I assume you're referring to:
>
> https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=247734
>
> An interesting reading, thanks, but it looks on the problem from
> somewhat different angle that I do. What I meant was:
>
> If one as perfectly valid (which seem to be a direct consequence of
> #247734) /etc/hosts with the following values:
>
> 127.0.0.1 localhost
> 127.0.1.1 <fqdn> <hostname>
I would tend to agree with this, if only because exim uses <fqdn> for a
HELO/EHLO and it would require some thought to fit it into a framework
using libnss-myhostname. It took long enough for us to get to something
which worked reliably so my motivation for changing isn't great.
For the moment libnss-myhostname on Jessie has only 3 rdepends (2 are
Recommends:), but there has been talk of d-i installing it by default.
That might see a change to the d-i generated /etc/hosts.
> Why would one need libnss-myhostname in Debian? Especially if Debian
> Installer generates such /etc/hosts for the last 10 years give or take?
>
> I acknowledge that there're other distributions than Debian, and they
> used to do things differently (for example, comment 125 of #247734
> implies that RedHat did not generate such /etc/hosts back then), so to
> address those other distributions' problem libnss-myhostname was
> created.
>
> But using this library in Debian seem to be redundant at best.
I think its ability to accomodate dynamic changes in the hostname is
seen as an advantage. Never having experienced that, I'm unable to
comment further.
Reply to: