[comp.mail.uucp] Too many hops... Why?

berger@datacube.UUCP (01/29/87)

Could anyone tell me why I got the following mail returned and how I
can fix this?

--------------------------------------
From MAILER-DAEMON@ATHENA.MIT.EDU Thu Jan 29 08:15:49 1987
Received: by datacube.UUCP (4.12/UUCP-Project/rel-1.0/12-12-86)
	id AA01810; Thu, 29 Jan 87 08:15:46 est
Received: by ATHENA (5.45/4.7)
	id AA07416; Thu, 29 Jan 87 04:11:34 EST
Received: by M4-035-T (5.45/4.7)
	id AB03378; Thu, 29 Jan 87 04:11:28 EST
Date: Thu, 29 Jan 87 04:11:28 EST
From: Mail Delivery Subsystem <MAILER-DAEMON@ATHENA.MIT.EDU>
Subject: Returned mail: Unable to deliver mail
Message-Id: <8701290911.AB03378@M4-035-T>
To: <berger@datacube.UUCP>
Status: R

   ----- Transcript of session follows -----
554 sendall: too many hops (17 max)

   ----- Unsent message follows -----
Received: by M4-035-T (5.45/4.7)
	id AA03378; Thu, 29 Jan 87 04:11:28 EST
Received: by ATHENA (5.45/4.7)
	id AA07412; Thu, 29 Jan 87 04:11:25 EST
Received: by M4-035-T (5.45/4.7)
	id AA03374; Thu, 29 Jan 87 04:11:20 EST
Received: by ATHENA (5.45/4.7)
	id AA07407; Thu, 29 Jan 87 04:11:17 EST
Received: by M4-035-T (5.45/4.7)
	id AA03370; Thu, 29 Jan 87 04:11:10 EST
Received: by ATHENA (5.45/4.7)
	id AA07403; Thu, 29 Jan 87 04:11:08 EST
Received: by M4-035-T (5.45/4.7)
	id AA03366; Thu, 29 Jan 87 04:11:02 EST
Received: by ATHENA (5.45/4.7)
	id AA07399; Thu, 29 Jan 87 04:11:00 EST
Received: by M4-035-T (5.45/4.7)
	id AA03362; Thu, 29 Jan 87 04:10:55 EST
Received: by ATHENA (5.45/4.7)
	id AA07395; Thu, 29 Jan 87 04:10:52 EST
Received: by M4-035-T (5.45/4.7)
	id AA03358; Thu, 29 Jan 87 04:10:47 EST
Received: by ATHENA (5.45/4.7)
	id AA07391; Thu, 29 Jan 87 04:10:44 EST
Received: by M4-035-T (5.45/4.7)
	id AA03354; Thu, 29 Jan 87 04:10:38 EST
Received: by ATHENA (5.45/4.7)
	id AA07387; Thu, 29 Jan 87 04:10:36 EST
Received: by M4-035-T (5.45/4.7)
	id AA03350; Thu, 29 Jan 87 04:10:31 EST
Received: by ATHENA (5.45/4.7)
	id AA06589; Thu, 29 Jan 87 02:51:07 EST
Received: by CCA.CCA.COM; Thu, 29 Jan 87 02:45:33 EST
Received: by datacube.UUCP (4.12/UUCP-Project/rel-1.0/12-12-86)
	id AA28917; Thu, 29 Jan 87 00:19:26 est
Date: Thu, 29 Jan 87 00:19:26 est
From: berger%datacube.UUCP@CCA.CCA.COM (Bob Berger)
Message-Id: <8701290519.AA28917@datacube.UUCP>
To: xpert@athena.mit.edu

**************** Body of Mail ***********************
-----------------------------------------------------------
				Bob Berger 

Datacube Inc. Systems / Software Group	4 Dearborn Rd. Peabody, Ma 01960
VOICE:	617-535-6644;	FAX: (617) 535-5643;  TWX: (710) 347-0125
UUCP:	ihnp4!datacube!berger
	{seismo,cbosgd,cuae2,mit-eddie}!mirror!datacube!berger

berger@datacube.UUCP (01/29/87)

I then got this mail today as well. It seems related:

From MAILER-DAEMON%ocean@harvard.HARVARD.EDU Thu Jan 29 11:18:55 1987
Received: by datacube.UUCP (4.12/UUCP-Project/rel-1.0/12-12-86)
	id AA04511; Thu, 29 Jan 87 11:18:52 est
Message-Id: <8701291353.AA07332@CCA.CCA.COM>
Received: by harvard.HARVARD.EDU; Thu, 29 Jan 87 08:53:48 EST
Date: Thu, 29 Jan 87 08:53:48 EST
Received: by ocean.harvard.edu; Thu, 29 Jan 87 08:53:17 EST
From: MAILER-DAEMON%ocean@harvard.HARVARD.EDU (Mail Delivery Subsystem)
Subject: Returned mail: Service unavailable
To: <berger@datacube.UUCP>
Status: R

   ----- Transcript of session follows -----
>>> HELO ocean.harvard.edu
<<< 553 ocean.harvard.edu I refuse to talk to myself
554 <leslie@ocean>... Service unavailable: Bad file number

   ----- Unsent message follows -----
Received: from harvard.HARVARD.EDU by endor.HARVARD.EDU; Thu, 29 Jan 87 03:30:53 EST
Received: by harvard.HARVARD.EDU; Thu, 29 Jan 87 03:39:35 EST
Received: by ATHENA (5.45/4.7)
	id AA06589; Thu, 29 Jan 87 02:51:07 EST
Received: by CCA.CCA.COM; Thu, 29 Jan 87 02:45:33 EST
Received: by datacube.UUCP (4.12/UUCP-Project/rel-1.0/12-12-86)
	id AA28917; Thu, 29 Jan 87 00:19:26 est
Date: Thu, 29 Jan 87 00:19:26 est
From: berger%datacube.UUCP@CCA.CCA.COM (Bob Berger)
Message-Id: <8701290519.AA28917@datacube.UUCP>
To: xpert@athena.mit.edu
Subject: Please add to mailing list

Please add me to your mailing list.  Please send mail to
datacube!xpert. This is an alias to automatically post it  to
our news/notes system.

				Thanks,
				Bob Berger 

Datacube Inc. Systems / Software Group	4 Dearborn Rd. Peabody, Ma 01960
VOICE:	617-535-6644;	FAX: (617) 535-5643;  TWX: (710) 347-0125
UUCP:	ihnp4!datacube!berger
	{seismo,cbosgd,cuae2,mit-eddie}!mirror!datacube!berger

dudek@endor.UUCP (01/31/87)

Ahhh, that last bounce: 'ocean.harvard.edu - I refuse to talk to
myself' was due to one of my sendmail configuration files which
failed to function correctly when said site ('ocean.harvard.edu')
upgraded to 4.3BSD.  My fault, sorry.

What follows is a complaint about 4.3BSD sendmail and domain servers -
those of you lucky enough not to have either can ignore this...

What actually caused the problem was a 4.3BSD version of sendmail which
supports the domain servers, and sets the '$w' (hostname) macro to
'ocean.harvard.edu', instead of just 'ocean'.  My sendmail.cf tried to match
mail addressed to 'user@ocean' by using the '$w' macro, which of course
failed.  The sendmail.cf then decided that this was non-local mail, which
it should deliver to that other host 'ocean', which it tried
to do.  Sendmail cleverly realized that it was talking to itself, and
bounced the message with the 'I refuse to talk to myself' error, thus saving
a costly mail loop (and also avoiding the 'too many hops' problem).

The fix was to hardwire the hostname 'ocean' in the sendmail configuration
instead of using the '$w' macro.  This seems like a loss to me, because I
can no longer run a generic, identical sendmail.cf on multiple machines.
It would be okay if sendmail allowed me to match a multiple token macro
(that is, 'user@$w' should match 'user@ocean.harvard.edu'), but it doesn't.
Am I missing something, is there a way around this without hardwiring
hostnames?  Fortunately, SUN 3.2 seems to do work in the presence of
domain servers (that is, sets '$w' to the hostname without the domain),
but I am stuck with either patching sendmail once on all my 4.3BSD machines,
or patching multiple sendmail.cf's whenever I make a change to the generic
version.  Yuk.

	Glen Dudek
	postmaster@harvard.harvard.edu

chris@columbia.UUCP (Chris Maio) (01/31/87)

In article <1138@husc6.UUCP> dudek@harvard.UUCP (Glen Dudek) writes:
>What follows is a complaint about 4.3BSD sendmail and domain servers -
>those of you lucky enough not to have either can ignore this...
>
>What actually caused the problem was a 4.3BSD version of sendmail which
>supports the domain servers, and sets the '$w' (hostname) macro to
>'ocean.harvard.edu', instead of just 'ocean'.

Glen,

  I got bitten the same way, but decided that it would be better to put the
local-host name in $=w instead of $w; here's a patch for the 4.3 BSD sendmail
that will include both "ocean" and "ocean.harvard.edu" in $=w.

						Chris

RCS file: main.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -c -r1.2 -r1.3
*** /tmp/,RCSt1026201	Sat Jan 31 00:41:08 1987
--- /tmp/,RCSt2026201	Sat Jan 31 00:41:10 1987
***************
*** 109,115 ****
  	bool queuemode = FALSE;		/* process queue requests */
  	bool nothaw;
  	static bool reenter = FALSE;
! 	char jbuf[30];			/* holds MyHostName */
  	extern bool safefile();
  	extern time_t convtime();
  	extern putheader(), putbody();
--- 109,115 ----
  	bool queuemode = FALSE;		/* process queue requests */
  	bool nothaw;
  	static bool reenter = FALSE;
! 	char jbuf[128];			/* holds MyHostName */
  	extern bool safefile();
  	extern time_t convtime();
  	extern putheader(), putbody();
***************
*** 263,268 ****
--- 263,283 ----
  				printf("\ta.k.a.: %s\n", *av);
  #endif DEBUG
  			setclass('w', *av++);
+ 		}
+ 
+ 
+ 		/* if the nameserver is being used, make sure that
+ 		   $=w contains our hostname without the domain name */
+ 		if ((p = index(jbuf, '.')) != NULL) {
+ 		    *p = '\0';		/* strip off the domain name */
+ 		    st = stab(jbuf, ST_CLASS, ST_FIND);
+ 		    if (st == NULL || !bitnset('w', st->s_class)) {
+ 			setclass('w', jbuf);
+ #ifdef DEBUG
+ 			if (tTd(0, 4))
+ 			    printf("\ta.k.a.: %s (no domain)\n", jbuf);
+ #endif DEBUG
+ 		    }
  		}
  
  		/* version */

diamant@hpfclp.HP.COM (John Diamant) (02/01/87)

>    ----- Transcript of session follows -----
> >>> HELO ocean.harvard.edu
> <<< 553 ocean.harvard.edu I refuse to talk to myself
> 554 <leslie@ocean>... Service unavailable: Bad file number

This indicates that ocean.harvard.edu has a bug in its sendmail configuration.
Show this message to the mail adminstrator on that system (you might be
able to mail to postmaster@ocean.harvard.edu unless their mailer is too sick
to handle that address).  In a correct configuration, this should never happen.
The problem is that the logic in the configuration file told it to establish
a network connection with itself (an SMTP link where source and destination
are the same machine).


John Diamant
SCO				UUCP:  {hplabs,hpfcla}!hpfclp!diamant
Hewlett Packard Co.		ARPA Internet: diamant%hpfclp@hplabs.HP.COM
Fort Collins, CO