Very Computer

In a perl cgi (or another language if necessary) is there a way to sense the
host-name the user's browser requested.  

For example, if www.company1.com is a primary name for a server and
www.company2.com is one alias, is there a way to sense (environment
variable, etc.) that the host www.company2.com was requested instead of
www.company1.com.

If not through perl, etc. what might be able to be done through the DNS.

Any help would be greatly appreciated.

Please CC: me on any posts.

Greg Keene

***************************************************************
Greg Keene                            The Watertown Group, Inc.

phone - 541/753-7551                         fax - 541/753-5548

***************************************************************

No. There is no way to tell what DNS alias was used

Quote:>If not through perl, etc. what might be able to be done through the DNS.

No. There is no way to do this via DNS.

The _only_ way to differentiate between multiple host/domain names on a single
machine, under the current version of the HTTP specification, is to use
a different IP address for each domain, and either execute a separate HTTPd
for each name, or use a virtual host server that can manage many addresses
at once (e.g. NCSA 1.5)

writes:

Quote:>In a perl cgi (or another language if necessary) is there a way to
>sense the
>host-name the user's browser requested.  

No.  In current HTTP, the only information that the server gets
is that there was an incoming call on a particular interface.
Appropriate servers can indicate which IP _number_ that corresponds
to.  But that number could correspond to several different DNS
names, and the server cannot know which.

Quote:>For example, if www.company1.com is a primary name for a server and
>www.company2.com is one alias, is there a way to sense (environment
>variable, etc.) that the host www.company2.com was requested instead of
>www.company1.com.

The key word here is "alias", and that makes the answer "no".

A host can support several IP numbers with different names, and
if each number has only one name, then it can deduce which name
it was called by.  But that wasn't what you asked.

I set followups to *.cgi please, I can't see any advantage in
dragging along all the others.

best regards

What kind of internet connection does the machine need to be able to do this?

I'm thinking about getting a static IP addresses from an ISP via an
ISDN PPP connection.  Is it possible for an ISP to assign more than
one IP address over a single PPP connection so I can use a virtual host
setup?

Thanks!

--
***********************************************************************

***********************************************************************

Yes- this is possible. Instead of a single IP address, you'll need to have
your ISP route a small subnet to you, giving you many addresses. Some ISP's
are reluctant to do this, and most will charge extra.

If your ISP will not give you a subnet out of their address block (which
is not an unreasonable position for them to take) see if they will arrange
for additional address space from their provider.

If all else fails, it is still possible (but not easy) to request that
Internic assign you a Class C address block. You will still need the
cooperation of your provider in routing the addresses to your system over the
PPP connection. The biggest advantage of an Internic-assigned Class C is that
it is portable- you can take it with you when you change providers.

The disadvantages are that it requires the cooperation of your provider and
your provider's provider so the address space is routed properly, and that
they are very scarce now.

Quote:

> If all else fails, it is still possible (but not easy) to request that
> Internic assign you a Class C address block. You will still need the

For one system? Well, if you're serving a whole lot of virtual servers...

Quote:> cooperation of your provider in routing the addresses to your system over the
> PPP connection. The biggest advantage of an Internic-assigned Class C is that
> it is portable- you can take it with you when you change providers.

These days, with the routing tables growing and growing, the chance of
actually pulling this off is getting worse.

Besides, what's the problem with renumbering ONE HOST?

IMHO, virtual servers (yes, we're running a bunch of them too) are _stupid_
and a costly mistake which could have been avoided if only the HTTP specs
had mandated passing the full URL from day one.  :-(

--
VAXCENTUATE - To underline the importance of the VAX.
--

Schleiermacherstra?e 12  \  Unix+Linux+Mac    | Phone: ...please use email.
90491 Nrnberg (Germany)  \   Consulting+Networking+Programming+etc'ing     42
          PGP: 1B 89 E2 1C 43 EA 80 44  15 D2 29 CF C6 C7 E0 DE
       Click <A HREF="http://smurf.noris.de/~smurf/finger">here</A>.

Well, the URI. What are the plans, if any, for migration to HTTP/1.1
once ratified?  That allows (I guess it's a MUST) a browser to
transmit a Host: header, or make a request with the URI.

It's an understandable omission, though, considering that one wouldn't
expect URLs to have become so "important" and image-creating, as the
URL wasn't intended to become so quotable or required to be memorable,
(and to some extent, the directories like Yahoo! and Yell! mean
that we don't need to remember them).

-- jrg.

-----BEGIN PGP SIGNED MESSAGE-----

: IMHO, virtual servers (yes, we're running a bunch of them too) are _stupid_
: and a costly mistake which could have been avoided if only the HTTP specs
: had mandated passing the full URL from day one.  :-(

I agree entirely.  I've got almost 30 running on our webserver here,
and it is a Very Bad Thing to have each one (some of which only have
maybe 2 or 3 pages on them!) take up its own IP address; this whole
vanity top-level domain thing is a rotten trend.  

I wonder what percentage of the latest rash of registrants will be around
when their initial two-year payment runs out?  Will we run out of
namespace before some kind of stasis is reached?   These things
keep me up at night.

- --
     - Eric Sorenson, Satanic SysAdmins Inc. - http://satanic.org/ -
        SATANIX 1.0: An Operating System for the Next Millenium.

[c.p.t.d added to ng's/f'ups]

-----BEGIN PGP SIGNATURE-----
Version: 2.6.1

iQBVAwUBMTJd4J2vLqbVwdjRAQFQ/QIAhyX4fx9DdLVtJwzR5OVuCLkzQvAdCbi5
crjC92VJpf3TQHGfSfg5sgnsSO0nph0I7DYU8YEheW/0uSkYa69SbA==
=G/LD
-----END PGP SIGNATURE-----