Re: where is /etc/hosts supposed to come from?
- To: debian-devel@lists.debian.org
- Subject: Re: where is /etc/hosts supposed to come from?
- From: Osamu Aoki <osamu@debian.org>
- Date: Sat, 2 Jan 2010 22:06:01 +0900
- Message-id: <[🔎] 20100102130601.GB22692@osamu.debian.net>
- Mail-followup-to: debian-devel@lists.debian.org
- In-reply-to: <20091231232951.GA3031@microcomaustralia.com.au>
- References: <87k4w5toe0.fsf@windlord.stanford.edu> <20091230013612.GD7527@xvii.vinc17.org> <87pr5xgke4.fsf@windlord.stanford.edu> <m3bphghpzi.fsf@neo.luffy.cx> <87ws04dhto.fsf@windlord.stanford.edu> <20091230171851.GC18722@khazad-dum.debian.net> <20091231001100.GE3375@microcomaustralia.com.au> <20091231154814.GA22383@osamu.debian.net> <slrnhjpkj9.etb.trash@kelgar.0x539.de> <20091231232951.GA3031@microcomaustralia.com.au>
Hi,
On Fri, Jan 01, 2010 at 10:29:51AM +1100, Brian May wrote:
> On Thu, Dec 31, 2009 at 04:33:44PM +0000, Philipp Kern wrote:
> > Using .local in own environments sadly clashes with the (I think now
> > installed by default) mDNS namespace for locally-resolvable hostnames.
> > But it somehow makes sense for a laptop; it would even resolve. ;-)
>
> Yes, that is what I was getting at too. It is also a problem for Mac OS X
> boxes, at least without fiddling.
I was naively thinking if hostname part being unique, it does not clash.
Am I incorrect?
> This can be a problem even if you don't intend to use zeroconf/mdns on your
> network. Typically what happens is DNS queries time out without resolving
> and packet traces show that there are no requests happening.
I see this. But this is general problem whatever bogus domain name is
used including ones like .invalid or .pri.
> I don't know of any good choice for a TLD that is guaranteed not to be used on
> the Internet, although currently I am using .pri (short for .private) here.
I see your point. It is less likely to crash since very few people have
chosen .pri than .local. But this doesn not seem to be something as a
widely accepted gneral practice as you described.
> Even .local is only mentioned in a experimental RFC that had expired (at least
> last I checked which was a while ago now).
>
> In case any is confused of the relationship between mdns and zeroconf:
...
> of the GNU C Library (glibc) providing host name resolution via Multicast
> DNS (using Zeroconf, aka Apple Bonjour / Apple Rendezvous ), effectively
> allowing name resolution by common Unix/Linux programs in the ad-hoc mDNS
> domain .local.
I thought FQDN under such environment is <hostname>.local . (Chosing
overlapping <hostname> itself on a single LAN is not good idea and very
rare.)
It seems these is no agreed practice to minimize negative impact. But
there are many good-enough solutions used widely by picking bogus domain
like .pri .lan .???
Osamu
Reply to: