Russ Allbery | 1 Jul 01:05 1999
Picon

Re: X-trace considered harmful

Brad Templeton <brad <at> templetons.com> writes:

> Of course.  But I wasn't talking about spool directories.  Do you mount
> your log files or history databases over NFS?  If so, I think that's a
> very poor design.  Multiple news transports simultaneously writing to
> the same log file and history file over NFS?

Oh, right, sorry.  Yeah, I doubt anyone's distributing their logs over
NFS.

--

-- 
Russ Allbery (rra <at> stanford.edu)         <URL:http://www.eyrie.org/~eagle/>

John Moreno | 1 Jul 04:44 1999

Re: Followups (was Section_4.02.01)

chl <at> clw.cs.man.ac.uk (Charles Lindsey) wrote:

> In <772747DB6A9.AAAB05 <at> isnmmp01.interpath.net> John Moreno <phenix <at> interpath.com> writes:
> 
> >That sounds right to me.  I'd like to point out that this is a real
> >world preference - I've actually encountered included message/rfc822
> >where I'd have wanted the MIME aware client to provide a way to
> >consider it the precursor instead of the enclosing message (which
> >frequently wasn't much more than an attribution).  So whatever we
> >recommend (if anything) it needs to be clear that it is the /default/
> >and is subject to the user overriding it.
> 
> >(The only question I'd have is why you think it wouldn't be appropriate
> >if the last part was message/rfc822?)
> 
> I can think of arguments for and against.
> 
> Case A.
> ------- 
> 
> --boundary
> 
> I am forwarding this mail to the group at the reqest of its author:
> 
> --boundary
> Content-Type: message/rfc822
> 
> From: some author
> 
> Blah! Blah! Blah!
(Continue reading)

Clive D.W. Feather | 1 Jul 09:49 1999
Picon

Re: Complaints-To

Bill Davidsen said:
> I have to disagree with this. While having a standard address is all
> well and good, in many cases it gets so much traffic that it suffers in
> time for detail. By standardising on a header line, each injector in an
> organization has the option to provide an abuse address for just the
> users of their small part of the organisation. I don't think it should
> be required, but I do think most large organizations would find
> legitimate uses for it.

And there are other cases:

    From: idiot <at> sample.demon.co.uk

No address at sample.demon.co.uk is likely to help you in the slightest;
the correct address is <abuse <at> demon.net>. I'm sure we're not the only ISP
like that. [I suspect, but haven't verified, that <abuse <at> demon.co.uk> goes
to the same place.]

--

-- 
Clive D.W. Feather  | Work:  <clive <at> demon.net>   | Tel: +44 20 8371 1138
Internet Expert     | Home:  <clive <at> davros.org>  | Fax: +44 20 8371 1037
Demon Internet Ltd. | WWW: http://www.davros.org | Mobile: +44 973 377646

Clive D.W. Feather | 1 Jul 10:19 1999
Picon

Re: X-trace considered harmful

Russ Allbery said:
>> Of course.  But I wasn't talking about spool directories.  Do you mount
>> your log files or history databases over NFS?  If so, I think that's a
>> very poor design.  Multiple news transports simultaneously writing to
>> the same log file and history file over NFS?
> Oh, right, sorry.  Yeah, I doubt anyone's distributing their logs over
> NFS.

You'd be surprised. I believe we do something of the sort though, if so,
I'm sure the contention issues have been worked out.

Please don't just jump in and say that NFS is automatically bad. Far from
it these days.

--

-- 
Clive D.W. Feather  | Work:  <clive <at> demon.net>   | Tel: +44 20 8371 1138
Internet Expert     | Home:  <clive <at> davros.org>  | Fax: +44 20 8371 1037
Demon Internet Ltd. | WWW: http://www.davros.org | Mobile: +44 973 377646

Brad Templeton | 1 Jul 10:33 1999
Picon

Re: Complaints-To

On Thu, Jul 01, 1999 at 08:49:34AM +0100, Clive D.W. Feather wrote:
> Bill Davidsen said:
> > I have to disagree with this. While having a standard address is all
> > well and good, in many cases it gets so much traffic that it suffers in
> > time for detail. By standardising on a header line, each injector in an
> > organization has the option to provide an abuse address for just the
> > users of their small part of the organisation. I don't think it should
> > be required, but I do think most large organizations would find
> > legitimate uses for it.
> 
> And there are other cases:
> 
>     From: idiot <at> sample.demon.co.uk
> 
> No address at sample.demon.co.uk is likely to help you in the slightest;
> the correct address is <abuse <at> demon.net>. I'm sure we're not the only ISP
> like that. [I suspect, but haven't verified, that <abuse <at> demon.co.uk> goes
> to the same place.]

But is there some great difficulty in saying that, if you select an
injector name, it should have an MX (or CNAME pointer) to the machine
where the usenet-abuse || abuse E-mail addresses will be valid?

Now my feeling is that really there is still only one item that needs
to go into an article for outside consumption in almost all cases, and
that is the user identity.   Being only one, it seems natural to me that
the place to put it is the already existing, but currently unused tail
entry on the Path line that used to hold the user identity years ago.

But I'll also concede that in the future we might think of other stuff
(Continue reading)

Brad Templeton | 1 Jul 10:37 1999
Picon

Re: X-trace considered harmful

On Thu, Jul 01, 1999 at 09:19:43AM +0100, Clive D.W. Feather wrote:
> Russ Allbery said:
> >> Of course.  But I wasn't talking about spool directories.  Do you mount
> >> your log files or history databases over NFS?  If so, I think that's a
> >> very poor design.  Multiple news transports simultaneously writing to
> >> the same log file and history file over NFS?
> > Oh, right, sorry.  Yeah, I doubt anyone's distributing their logs over
> > NFS.
> 
> You'd be surprised. I believe we do something of the sort though, if so,
> I'm sure the contention issues have been worked out.
> 
> Please don't just jump in and say that NFS is automatically bad. Far from
> it these days.

I'm sure NFS is just fine for some apps.   However, what somebody said was
they thought appending a log entry to a file was some sort of difficult
task.  Even with an NFS unable to do an atomic write to a log file, it's
not exactly hard to simply have, on this NFS server, one injection log
file per machine.

It makes searching the logs easy.  "grep message-id /usr/lib/news/log/inject/*"
(or whatever place you put these non-aged logs.)

The difficulty of logging, and finding logs remains unconvincing to me as
a reason to put your audit trail log information into every article.

Charles Lindsey | 1 Jul 11:43 1999
Picon
Picon

Re: X-trace considered harmful

In <ylpv2d5tp9.fsf <at> windlord.stanford.edu> Russ Allbery <rra <at> stanford.edu> writes:

>Charles Lindsey <chl <at> clw.cs.man.ac.uk> writes:

>> 	%demon.net![193.195.224.128]!romana.davros.org!clive

>I believe that would be:

>        demon.net%[193.195.224.128]!romana.davros.org!clive

Yes. You are correct.

>I agree that NNTP-Posting-Host is potentially made obsolete by the new
>Path semantics, and I don't have much of a problem with that (although I'm
>somewhat reluctant to toss out all of the work that has been put into
>enforcing NNTP-Posting-Host headers and not allowing users to add them in
>most cases).

No harm will be done if those currently using it continue to do so.
Eventually, it will likely wither away.

--

-- 
Charles H. Lindsey ---------At Home, doing my own thing------------------------
Email:     chl <at> clw.cs.man.ac.uk  Web:   http://www.cs.man.ac.uk/~chl
Voice/Fax: +44 161 437 4506      Snail: 5 Clerewood Ave, CHEADLE, SK8 3JU, U.K.
PGP: 2C15F1A9     Fingerprint: 73 6D C2 51 93 A0 01 E7  65 E8 64 7E 14 A4 AB A5

Charles Lindsey | 1 Jul 11:47 1999
Picon
Picon

Re: X-trace considered harmful

In <yln1xh5tfg.fsf <at> windlord.stanford.edu> Russ Allbery <rra <at> stanford.edu> writes:

>Charles Lindsey <chl <at> clw.cs.man.ac.uk> writes:

>> The poster himself is reasonably treated as a site. All the concepts you
>> listed can (mostly) fit into this "site" model.

>How can you meaningfully treat the poster as a site?  We don't seem to be
>using that information for any of the things we use a site for; we aren't
>using it to control feeding, we aren't using it as a way to derive an
>e-mail address, we're not using it to determine the responsible party...
>we're using it for spam filtering and for information passed along in
>abuse reports.

The "poster" is conceptually a "site" insofar as it represents the seat
where the man sat in front of his machine. Thus it is in one sense the
originating "site". I agree that, technically speaking, it is not a "site"
insofar as software does not look at it for purposes like avoiding cycles.

--

-- 
Charles H. Lindsey ---------At Home, doing my own thing------------------------
Email:     chl <at> clw.cs.man.ac.uk  Web:   http://www.cs.man.ac.uk/~chl
Voice/Fax: +44 161 437 4506      Snail: 5 Clerewood Ave, CHEADLE, SK8 3JU, U.K.
PGP: 2C15F1A9     Fingerprint: 73 6D C2 51 93 A0 01 E7  65 E8 64 7E 14 A4 AB A5

Charles Lindsey | 1 Jul 12:17 1999
Picon
Picon

Re: Uppercase

In <ylwvwl7ar9.fsf <at> windlord.stanford.edu> Russ Allbery <rra <at> stanford.edu> writes:

>And then INN came along, and looked at what C News was doing, and tried to
>basically do the same thing....

>I think it's time to finally put this one to rest, though.  There's no
>longer any real reason for it, I'll bet a lot of other servers just do
>case-sensitive comparisons, and I don't think anyone's actually using the
>case insensitivity any more.

OK. I now have the following wording in the Message-ID section:

...
an agent generating an article's message identifier MUST ensure that it is
unique and that it is NEVER reused. Moreover, even though commonly derived
from the domain name of the originating site (and domain names are
case-insensitive), a message identifier MUST NOT be altered in any way during
transport, or when copied (as into a References header), and thus a simple 
(case-sensitive) comparison of octets will always suffice to recognise that
same message identifier wherever it subsequently reappears. 

Will that do?

And would Pete Resnick like to say the same thing for Mail?

--

-- 
Charles H. Lindsey ---------At Home, doing my own thing------------------------
Email:     chl <at> clw.cs.man.ac.uk  Web:   http://www.cs.man.ac.uk/~chl
Voice/Fax: +44 161 437 4506      Snail: 5 Clerewood Ave, CHEADLE, SK8 3JU, U.K.
PGP: 2C15F1A9     Fingerprint: 73 6D C2 51 93 A0 01 E7  65 E8 64 7E 14 A4 AB A5
(Continue reading)

Charles Lindsey | 1 Jul 11:57 1999
Picon
Picon

Re: X-trace considered harmful

In <19990630140958.22885 <at> main.templetons.com> Brad Templeton <brad <at> templetons.com> writes:

>Actually, you must have misread what I said.  I said there were two options,
>one of which is to throw it away, the other of which is to integrate it.
>If integrated, the other options to consider are where to put the
>generated user-identity, either in the last entry, or right after the %.

>Both are valid.

I disagree. Anything added to the Path is _always_ prepended. An injector
either does that, or it throws it away and starts afresh (I prefer the
former as a general rule).

>The only thing not valid is accepting a path line with a '%' in it.  That
>is a double injection, which should be an error.   POST should not be used
>for articles being relayed.   If you want to run a news server using
>suck and POST, you can do so, but you shold not relay for other people.
>Find somebody who will take your ihave feeds.

I agree with that bit about suck feeds, but I do not regard it as so bad
to refuse a '%' in the proto-path. It is a cause for suspicion, but is
likely harmless, and is always obvious. As I said before, it should not
arise in a "well-ordered net", but Real Nets are not always well-ordered
:-( .

>Double inject would be fine if all injectors were minimal, adding only the
>path, but until they are so simple you don't want two sents of injectors
>trying to do the things injectors do.

Certainly a second injector should not be altering important headers. But
(Continue reading)


Gmane