> If syslog client sends a message over TCP/UDP/IP, it by definition has
> an IP address.  If it send the message locally, then I think 127.0.0.1
> (local IP) is more appropriate.

I think the case we are describing is a very unusual one. It only
happens when the device (better said: software running) has no way of
*obtaining knowledge* of its IP address. I am not sure if we will find
this in reality at all. So it does not matter (at least in my view) that
it has an IP address - it can not get hold of it). So I am using a
special value to say "I'm clueless".

On the other hand, 127.0.0.1 is a valid value - it may be used when a
sender talks to a receiver on the same machine.

I would like to differentiate these two cases.

> But I am assuming that we won't use syslog-protocol through say local
> UNIX pipes like syslog does now locally on say Solaris.  If we do want
> to support that, than we need the case of an unspecified IP.  But if
> syslog is used over TCP/IP, "unspecified IP" is an oxymoron.

As I said... depends. I am talking about a device *having* an IP address
but the software not knowing it. If we all agree this is so unlikely
that we should not support it, then I think we should fully drop this
"last resort rule" - because we are saying it can never apply ;)

Rainer
>
> Anton.
>
> > -----Original Message-----
> > From: Rainer Gerhards [mailto:[EMAIL PROTECTED]
> > Sent: Friday, April 23, 2004 9:01 AM
> > To: [EMAIL PROTECTED]
> > Cc: Anton Okmianski
> > Subject: RE: Issue 14: allow unqualified hostname
> >
> >
> > Anton & all,
> >
> > in IPv6, we have the "unspecified address", which I think is
> > exactly what we should use in the case an device does
> > actually know nothing about itself (last case in Anton's
> > messsage below) it is "0:0:0:0:0:0:0:0".
> >
> > Some links:
> > http://www.iana.org/assignments/ipv6-address-space
> > http://www.cisco.com/en/US/products/sw/iosswrel/ios_abcs_ios_t
> > he_abcs_ip
> > _version_60900aecd800c111d.html
> >
> > According to RFC 3330
> > (http://www.rfc-editor.org/rfc/rfc3330.txt), I think we can
> > also use "0.0.0.0" for IPv4 addressing in this case.
> >
> > Comments are highly appreciated.
> >
> > Rainer
> >
> > > -----Original Message-----
> > > From: Anton Okmianski [mailto:[EMAIL PROTECTED]
> > > Sent: Monday, April 19, 2004 10:19 PM
> > > To: Rainer Gerhards
> > > Cc: [EMAIL PROTECTED]
> > > Subject: RE: Issue 14: allow unqualified hostname
> > >
> > > Rainer:
> > >
> > > I like Devin's suggestion of recommending a specific
> > preference order:
> > >
> > > FQDN
> > > Static IPv4/IPv6
> > > Hostname
> > > Dynamic IPv4/IPv6
> > > "127.0.0.1" (when everything is unknown)
> > >
> > > Maybe the language should be a bit more restrictive than
> > just SHOULDs
> > > and MAYs here.  Maybe: "MUST provide FQDN if it is known.
> > If unknown -
> > > static IP. If unknown -- hostname. If unknown - dynamic IP.  If
> > > unknown -- (a) can't use syslog or (b) we explain what they should
>
> > > use."
> > >
> > > I don't know if we decided on the last one. If syslog is to be
> used
> > > for remote logging only, then requiring knowledge of at
> > least an IP is
> > > acceptable.  If, however, we expect it to be used in host-local
> > > scenarios as well, then we need to clarify what they should
> > there when
> > > nothing I known. Devin suggested 127.0.0.1. I like it. Maybe also
> > > allow an IPv6 equivalent of that as well if it exists.
> > >
> > > Thanks,
> > > Anton.
> > >
> > > > -----Original Message-----
> > > > From: [EMAIL PROTECTED]
> > > > [mailto:[EMAIL PROTECTED] On Behalf Of
> > Rainer Gerhards
> > > > Sent: Monday, April 19, 2004 11:14 AM
> > > > To: [EMAIL PROTECTED]
> > > > Subject: Issue 14: allow unqualified hostname
> > > >
> > > >
> > > > Hi WG,
> > > >
> > > > this is in regard to issue 14, which talks about allowing the
> > > > unqualified hostname. Based on previous feedback, I think
> > > > this is concensus in the WG (see
> > > > http://www.syslog.cc/ietf/protocol/issue14.htm> l
> > > > for a short
> > > > list).
> > > >
> > > > If nobody objects, I will go ahead and
> > > > edit it in the following way:
> > > >
> > > > Hostname & FQDN SHOULD  be used, IP (v4/6) address or "bare"
> > > > hostname MAY be used.
> > > >
> > > > Rainer
> > > >
> > > >
> > >
> > >
> >
>
>


Reply via email to