On Mon, 23 Oct 2006, Jo Rhett wrote:
> David B Funk wrote:
> > On Thu, 19 Oct 2006, Jo Rhett wrote:
> >
> >> Richard Frovarp wrote:
> >>> Or for any machine that hosts more domains than has IPs. Even being able
> >>> to edit the reverse doesn't mean it will always be the same.
> >> How many different names does your mailserver use in its HELO?
> >>
> >> And what mailserver is that? That's not possible in qmail, postfix,
> >> sendmail, et al...
> >
> > You're a bit behind the times Jo, check out the 'h' argument to
> > 'ClientPortOptions' or the 'HeloName' variable in sendmail 8.13.
>
> I can find no documentation of either. Googling just gets me lots of
> examples of a script called SendMail()
Some of us pre-date Sir Timothy & his bright idea, had to make our
ones & zeros the hard way by banging two rocks together, had to learn
to find and read documentation. (I first ran into sendmail on a VAX-750
running BSD-4.2 in the early 80's).
In every sendmail release for the last decade there's been a document
"doc/op/op.me" which is the configuration and operations manual. In that
doc for 8.13.* you'll find:
ClientPortOptions=options
[O] Set client SMTP options. The options
are key=value pairs separated by commas.
Known keys are:
Port Name/number of source port for connection
(defaults to any free port)
Addr Address mask (defaults INADDR_ANY)
Family Address family (defaults to INET)
SndBufSizeSize of TCP send buffer
RcvBufSizeSize of TCP receive buffer
Modifier Options (flags) for the client
The Address mask may be a numeric address in
dot notation or a network name. Modifier
can be the following character:
h use name of interface for HELO command
A don't use AUTH when sending e-mail
S don't use STARTTLS when sending e-mail
If ``h'' is set, the name corresponding to
the outgoing interface address (whether cho-
sen via the Connection parameter or the
default) is used for the HELO/EHLO command.
Now if the only way you can relate to things is via a web-page then
look at: http://www.sendmail.org/doc/sendmail-current/doc/op/op.pdf
> Looking at the code, heloname would appear to be statically defined,
I'm not sure what your strong points are Jo, but reading 'c' code doesn't
appear to be one of them. I made no mention of 'heloname' ('c' is case
sensitive). In the sendmail source file readcf.c the variable 'HeloName'
is assigned a value in the case statement:
case O_HELONAME:
HeloName = newstr(val);
break;
where 'val' is the token that has just been parsed out of the config file.
(not a static definition).
> which brings me back to my original point: how many names does his
> mailserver use in helo?
>
> Sure, if it has 3 interfaces (and uses them all) then he'll need three
> names. But he won't need 1000 or however many virtual hosts he has...
If that was your point, then why did you make that bogus assertion that
it wasn't possible for MTAs (at least sendmail) to use different HELO names?
And if he wants to use TLS-SSL then he'll have to have a different
interface and matching name for each virtual host.
My whole point here was not to make you look foolish but to point
out that maybe you should stop and think a bit more before going off
and making unsupportable statements.
For example, a while back you were complaining about a FP from a bank
anti-phisihing rule. It was probably caused by that defective milter you
were using. A bit of digging (rather than ranting) might have shown
you something that other sendmail-2-SA milter authors found out years
ago, the need for that added 'Received:' header.
--
Dave Funk University of Iowa
<dbfunk (at) engineering.uiowa.edu> College of Engineering
319/335-5751 FAX: 319/384-0549 1256 Seamans Center
Sys_admin/Postmaster/cell_admin Iowa City, IA 52242-1527
#include <std_disclaimer.h>
Better is not better, 'standard' is better. B{
