[Mailman-Users] sendmail error.

Jeme A Brelin jeme at brelin.net
Mon Aug 14 02:01:53 CEST 2000


On Sun, 13 Aug 2000, Christopher P. Lindsey wrote:
> > The DefaultUser line is commented out of sendmail.cf.
> > 
> > I assume this means it's using mail:mail (uid 8, gid 8).
> 
> Not necessarily.  There's a whole bunch of stuff that sendmail goes
> through to determine which uid/gid to run things under.  If you have
> the sendmail "bat" book, it's in section 34.8.15 and covers a few
> pages.
> 
> But usually, it defaults to this:
> 
>    "If an entire u or g option is missing, the default value becomes
>    1 (usually 'daemon')."

Wait wait... then why does /var/log/mail.err say:
Mailman mail-wrapper: Failure to exec script. WANTED gid 8, GOT gid 1.  (Reconfigure to take 1?)

This doesn't make sense to me... if it's getting 1 and wants 8, then why
doesn't it get 8 when I configure with --with-mail-gid=8 ?

> Quickest and easiest test to see what uid sendmail uses for its g and
> u options is to add an alias like this:
>    testuid: /tmp/testuid
> Run '/usr/lib/sendmail -bi', then send mail to testuid.  It will create
> /tmp/testuid with permissions for the user that sendmail is running
> as.

So it goes...

I did what is described above and got this:
[root at shaft]~$ ls -al /tmp/testuid
-rw-------    1 daemon   daemon        353 Aug 13 15:57 /tmp/testuid


So this says to me that sendmail is running as daemon:daemon... but then
why didn't mailman work when I used not --with-mail-gid= line?  Should it
not have defaulted to daemon?

And STILL why does the maillog say WANTED 8, GOT 1?

Oi vey.

I'm going to reconfigure mailman specifically with --with-mail-gid=1 and
see what happens.

OK... that worked.

So now my questions is simply, why does sendmail's log seem to indicate
the reverse condition of what is really happening?

That is all.
J.
-- 
   -----------------
     Jeme A Brelin
    jeme at brelin.net
   -----------------
 [cc] counter-copyright
 http://www.openlaw.org







More information about the Mailman-Users mailing list