xref: /csrg-svn/usr.sbin/sendmail/FAQ (revision 68069)
166282Seric			    Sendmail Version 8
266282Seric			Frequently Asked Questions
3*68069Seric		         Version 8.9 of 12/09/94
466282Seric
566282Seric
667021SericThis FAQ is specific to Version 8 of sendmail.  Other questions,
767021Sericparticularly regarding compilation and configuration, are answered
867892Sericin src/READ_ME and cf/README (found in the V8 sendmail distribution).
966282Seric
10*68069Seric======================================================================
11*68069SericGENERAL QUESTIONS
12*68069Seric======================================================================
13*68069Seric
1467892Seric  * What do you wish everyone would do before sending you mail?
1567892Seric
1667892Seric	Read this FAQ completely.  Read src/READ_ME and cf/README
1767892Seric	completely.  Ask themselves if their question hasn't already
1867892Seric	been answered.
1967892Seric----------------------------------------------------------------------
2067021Seric  * Where can I get Version 8?
2167021Seric
2267021Seric	Via anonymous FTP from FTP.CS.Berkeley.EDU in /ucb/sendmail.
2367021Seric----------------------------------------------------------------------
2466282Seric  * What are the differences between Version 8 and other versions?
2566282Seric
2667021Seric	See doc/changes/changes.me in the sendmail distribution.
2766282Seric----------------------------------------------------------------------
2866282Seric  * What happened to sendmail 6.x and 7.x?
2966282Seric
3067892Seric	When I released a new (Alpha/Beta) version of sendmail, I changed
3167892Seric	it to Release 6.  Development continued in that tree until 4.4BSD
3266282Seric	was released, when everything on the 4.4 tape was set to be
3366282Seric	version 8.1.  Version 7.x never existed.
3466282Seric----------------------------------------------------------------------
35*68069Seric  * What books are available describing sendmail?
36*68069Seric
37*68069Seric	There is one book available devoted to sendmail:
38*68069Seric
39*68069Seric	    Costales, Allman, and Rickert, _Sendmail_.  O'Reilly &
40*68069Seric		Associates.
41*68069Seric
42*68069Seric	Several books have sendmail chapters, for example:
43*68069Seric
44*68069Seric	    Nemeth, Snyder, and Seebass, _Unix System Administration
45*68069Seric		Handbook_.  Prentice-Hall.
46*68069Seric	    Carl-Mitchell and Quarterman, _Practical Internetworking with
47*68069Seric		TCP/IP and UNIX_.  Addison-Wesley.
48*68069Seric	    Hunt, _TCP/IP Network Administration_.  O'Reilly & Associates.
49*68069Seric
50*68069Seric	Another book about sendmail is due out "soon":
51*68069Seric
52*68069Seric	    Avolio & Vixie, _Sendmail Theory and Practice_.  Digital
53*68069Seric		Press (release date unknown).
54*68069Seric
55*68069Seric======================================================================
56*68069SericCOMPILING AND INSTALLING SENDMAIL 8
57*68069Seric======================================================================
58*68069Seric
5966282Seric  * Version 8 requires a new version of "make".  Where can I get this?
6066282Seric
6166282Seric	Actually, Version 8 does not require a new version of "make".
6266282Seric	It includes a collection of Makefiles for different architectures,
6367892Seric	only one or two of which require the new "make".  For a supported
6467892Seric	architecture, use ``sh makesendmail''.  If you are porting to a
6567892Seric	new architecture, start with Makefile.dist.
6666282Seric
6766282Seric	If you really do want the new make, it is available on any of
6867021Seric	the BSD Net2 or 4.4-Lite distribution sites.  These include:
6966282Seric
7066282Seric		ftp.uu.net		/systems/unix/bsd-sources
7166282Seric		gatekeeper.dec.com	/.0/BSD/net2
7266282Seric		ucquais.cba.uc.edu	/pub/net2
7366282Seric		ftp.luth.se		/pub/unix/4.3bsd/net2
7466282Seric
7566282Seric	Diffs and instructions for building this version of make under
7666282Seric	SunOS 4.1.x are available on ftp.css.itd.umich.edu in
7767556Seric	/pub/systems/sun/Net2-make.sun4.diff.Z.  A patchkit for Ultrix
7867556Seric 	is on ftp.vix.com in /pub/patches/pmake-for-ultrix.Z.  Patches
7967556Seric	for AIX 3.2.4 are available on ftp.uni-stuttgart.de in
8067556Seric	/sw/src/patches/bsd-make-rus-patches.
8167489Seric
8267489Seric	There is also a Linux version available on the main Linux
8367489Seric	distribution sites as pmake; this version is included as
8467489Seric	standard with the current Slackware distributions.
8566282Seric----------------------------------------------------------------------
8666282Seric  * What macro package do I use to format the V8 man pages?
8766282Seric
8866282Seric	The BSD group switched over the the ``mandoc'' macros for
8966282Seric	the 4.4 release.  These include more hooks designed for
9066282Seric	hypertext handling.  However, new man pages won't format
9166282Seric	under the old man macros.  Fortunately, old man pages will
9266282Seric	format under the new mandoc macros.
9366282Seric
9467892Seric	Get the new macros with the BSD Net2 or 4.4-Lite release
9567892Seric	(see above).
9666282Seric
9766282Seric	This macro set is also available with newer versions of groff.
9866282Seric
99*68069Seric======================================================================
100*68069SericCONFIGURATION QUESTIONS
101*68069Seric======================================================================
10266282Seric
10366282Seric  * How do I make all my addresses appear to be from a single host?
10466282Seric
10566282Seric	Using the V8 configuration macros, use:
10666282Seric
10766282Seric		MASQUERADE_AS(my.dom.ain)
10866282Seric
10966282Seric	This will cause all addresses to be sent out as being from
11066282Seric	the indicated domain.
11166282Seric----------------------------------------------------------------------
11266282Seric  * How do I rewrite my From: lines to read ``First_Last@My.Domain''?
11366282Seric
11466282Seric	There are a couple of ways of doing this.  This describes using
11566282Seric	the "user database" code.  This is still experimental, and was
11666282Seric	intended for a different purpose -- however, it does work
11766282Seric	with a bit of care.  It does require that you have the Berkeley
11866282Seric	"db" package installed (it won't work with DBM).
11966282Seric
12066282Seric	First, create your input file.  This should have lines like:
12166282Seric
12266282Seric		loginname:mailname	First_Last
12366282Seric		First_Last:maildrop	loginname
12466282Seric
12566282Seric	Install it in (say) /etc/userdb.  Create the database:
12666282Seric
12766282Seric		makemap btree /etc/userdb.db < /etc/userdb
12866282Seric
12966282Seric	You can then create a config file that uses this.  You will
13066282Seric	have to include the following in your .mc file:
13166282Seric
13266282Seric		define(confUSERDB_SPEC, /etc/userdb.db)
13366282Seric		FEATURE(notsticky)
13466282Seric----------------------------------------------------------------------
13566282Seric  * So what was the user database feature intended for?
13666282Seric
13766282Seric	The intent was to have all information for a given user (where
13866282Seric	the user is the unique login name, not an inherently non-unique
13966282Seric	full name) in one place.  This would include phone numbers,
14066282Seric	addresses, and so forth.  The "maildrop" feature is because
14166282Seric	Berkeley does not use a centralized mail server (there are a
14266282Seric	number of reasons for this that are mostly historic), and so
14366282Seric	we need to know where each user gets his or her mail delivered --
14466282Seric	i.e., the mail drop.
14566282Seric
14666282Seric	We are in the process of setting up our environment so that
14766282Seric	mail sent to an unqualified "name" goes to that person's
14866282Seric	preferred maildrop; mail sent to "name@host" goes to that
14966282Seric	host.  The purpose of "FEATURE(notsticky)" is to cause
15066282Seric	"name@host" to be looked up in the user database for delivery
15166282Seric	to the maildrop.
15266282Seric----------------------------------------------------------------------
15366282Seric  * Why are you so hostile to using full names for e-mail addresses?
15466282Seric
15566282Seric	Because full names are not unique.  For example, the computer
15666282Seric	community has two Andy Tannenbaums and two Peter Deutsches.
15766282Seric	At one time, Bell Labs had two Stephen R. Bournes with offices
15866282Seric	a few doors apart.  You can create alternative addresses
15966282Seric	(e.g., Stephen_R_Bourne_2), but that's even worse -- which
16066282Seric	one of them has to have their name desecrated in this way?
16167892Seric	And you can bet that one of them will get most of the other
16267892Seric	person's e-mail.
16366282Seric
16467892Seric	So called "full names" are just an attempt to create longer
16567892Seric	versions of unique names.  Rather that lulling people into a
16667892Seric	sense of security, I'd rather that it be clear that these
16767892Seric	handles are arbitrary.  People should use good user agents
16867892Seric	that have alias mappings so that they can attach arbitrary
16967892Seric	names for their personal use to those with whom they correspond
17067892Seric	(such as the MH alias file).
17166282Seric
17266282Seric	Even worse is fuzzy matching in e-mail -- this can make good
17366282Seric	addresses turn bad.  For example, I'm currently (to the best
17466282Seric	of my knowledge) the only ``Allman'' at Berkeley, so mail
17566282Seric	sent to "Allman@Berkeley.EDU" should get to me.  But if
17666282Seric	another Allman ever appears, this address could suddenly
17766282Seric	become ambiguous.  I've been the only Allman at Berkeley for
17866282Seric	over fifteen years -- to suddenly have this "good address"
17966282Seric	bounce mail because it is ambiguous would be a heinous wrong.
18066282Seric
18167892Seric	Finger services should be as fuzzy as possible (within
18267892Seric	reason, of course).  Mail services should be unique.
18366282Seric----------------------------------------------------------------------
184*68069Seric  * Should I use a wildcard MX for my domain?
185*68069Seric
186*68069Seric	If at all possible, no.
187*68069Seric
188*68069Seric	Wildcard MX records have lots of semantic "gotcha"s.  For
189*68069Seric	example, they will match a host "unknown.your.domain" -- if
190*68069Seric	you don't explicitly test for unknown hosts in your domain,
191*68069Seric	you will get "config error: mail loops back to myself"
192*68069Seric	errors.
193*68069Seric----------------------------------------------------------------------
194*68069Seric  * How can I get sendmail to deliver local mail to $HOME/.mail
195*68069Seric    instead of into /usr/spool/mail (or /usr/mail)?
196*68069Seric
197*68069Seric	This is a local mailer issue, not a sendmail issue.  Either
198*68069Seric	modify your local mailer (source code will be required) or
199*68069Seric	change the program called in the "local" mailer configuration
200*68069Seric	description to be a new program that does this local delivery.
201*68069Seric	I understand that "procmail" works well, although I haven't
202*68069Seric	used it myself.
203*68069Seric
204*68069Seric	You might be interested in reading the paper ``HLFSD: Delivering
205*68069Seric	Email to your $HOME'' available in the Proceedings of the
206*68069Seric	USENIX System Administration (LISA VII) Conference (November
207*68069Seric	1993).  This is also available via public FTP from
208*68069Seric	ftp.cs.columbia.edu:/pub/hlfsd/{README.hlfsd,hlfsd.ps}.
209*68069Seric----------------------------------------------------------------------
210*68069Seric  * I'm trying to to get my mail to go into queue only mode, and it
211*68069Seric    delivers the mail interactively anyway.  (Or, I'm trying to use
212*68069Seric    the "don't deliver to expensive mailer" flag, and it doesn't
213*68069Seric    delivers the mail interactively anyway.)  I can see it does it:
214*68069Seric    here's the output of "sendmail -v foo@somehost" (or Mail -v or
215*68069Seric    equivalent).
216*68069Seric
217*68069Seric	The -v flag to sendmail (which is implied by the -v flag to
218*68069Seric	Mail and other programs in that family) tells sendmail to
219*68069Seric	watch the transaction.  Since you have explicitly asked to
220*68069Seric	see what's going on, it assumes that you do not want to to
221*68069Seric	auto-queue, and turns that feature off.  Remove the -v flag
222*68069Seric	and use a "tail -f" of the log instead to see what's going on.
223*68069Seric
224*68069Seric	If you are trying to use the "don't deliver to expensive mailer"
225*68069Seric	flag (mailer flag "e"), be sure you also turn on global option
226*68069Seric	"c" -- otherwise it ignores the mailer flag.
227*68069Seric----------------------------------------------------------------------
228*68069Seric  * There are four UUCP mailers listed in the configuration files.
229*68069Seric    Which one should I use?
230*68069Seric
231*68069Seric	The choice is partly a matter of local preferences and what is
232*68069Seric	running at the other end of your UUCP connection.  Unlike good
233*68069Seric	protocols that define what will go over the wire, UUCP uses
234*68069Seric	the policy that you should do what is right for the other end;
235*68069Seric	if they change, you have to change.  This makes it hard to
236*68069Seric	do the right thing, and discourages people from updating their
237*68069Seric	software.  In general, if you can avoid UUCP, please do.
238*68069Seric
239*68069Seric	If you can't avoid it, you'll have to find the version that is
240*68069Seric	closest to what the other end accepts.  Following is a summary
241*68069Seric	of the UUCP mailers available.
242*68069Seric
243*68069Seric	uucp-old (obsolete name: "uucp")
244*68069Seric	  This is the oldest, the worst (but the closest to UUCP) way of
245*68069Seric	  sending messages accros UUCP connections.  It does bangify
246*68069Seric	  everything and prepends $U (your UUCP name) to the sender's
247*68069Seric	  address (which can already be a bang path itself).  It can
248*68069Seric	  only send to one address at a time, so it spends a lot of
249*68069Seric	  time copying duplicates of messages.  Avoid this if at all
250*68069Seric	  possible.
251*68069Seric
252*68069Seric	uucp-new (obsolete name: "suucp")
253*68069Seric	  The same as above, except that it assumes that in one rmail
254*68069Seric	  command you can specify several recipients.  It still has a
255*68069Seric	  lot of other problems.
256*68069Seric
257*68069Seric	uucp-dom
258*68069Seric	  This UUCP mailer keeps everything as domain addresses.
259*68069Seric	  Basically, it uses the SMTP mailer rewriting rules.
260*68069Seric
261*68069Seric	  Unfortunately, a lot of UUCP mailer transport agents require
262*68069Seric	  bangified addresses in the envelope, although you can use
263*68069Seric	  domain-based addresses in the message header.  (The envelope
264*68069Seric	  shows up as the From_ line on UNIX mail.)  So....
265*68069Seric
266*68069Seric	uucp-uudom
267*68069Seric	  This is a cross between uucp-new (for the envelope addresses)
268*68069Seric	  and uucp-dom (for the header addresses).  It bangifies the
269*68069Seric	  envelope sender (From_ line in messages) without adding the
270*68069Seric	  local hostname, unless there is no host name on the address
271*68069Seric	  at all (e.g., "wolf") or the host component is a UUCP host name
272*68069Seric	  instead of a domain name ("somehost!wolf" instead of
273*68069Seric	  "some.dom.ain!wolf").
274*68069Seric
275*68069Seric	Examples:
276*68069Seric
277*68069Seric	We are on host grasp.insa-lyon.fr (UUCP host name "grasp").  The
278*68069Seric	following summarizes the sender rewriting for various mailers.
279*68069Seric
280*68069Seric	Mailer          sender		rewriting in the envelope
281*68069Seric	------		------		-------------------------
282*68069Seric	uucp-{old,new}	wolf		grasp!wolf
283*68069Seric	uucp-dom	wolf		wolf@grasp.insa-lyon.fr
284*68069Seric	uucp-uudom	wolf		grasp.insa-lyon.fr!wolf
285*68069Seric
286*68069Seric	uucp-{old,new}	wolf@fr.net	grasp!fr.net!wolf
287*68069Seric	uucp-dom	wolf@fr.net	wolf@fr.net
288*68069Seric	uucp-uudom	wolf@fr.net	fr.net!wolf
289*68069Seric
290*68069Seric	uucp-{old,new}	somehost!wolf	grasp!somehost!wolf
291*68069Seric	uucp-dom	somehost!wolf	somehost!wolf@grasp.insa-lyon.fr
292*68069Seric	uucp-uudom	somehost!wolf	grasp.insa-lyon.fr!somehost!wolf
293*68069Seric
294*68069Seric======================================================================
295*68069SericRESOLVING PROBLEMS
296*68069Seric======================================================================
297*68069Seric
298*68069Seric  * When I compile, I get "undefined symbol inet_aton" messages.
299*68069Seric
300*68069Seric	You've probably replaced your resolver with the version from
301*68069Seric	BIND 4.9.3.  You need to cmpile with -l44bsd in order to get
302*68069Seric	the additional routines.
303*68069Seric----------------------------------------------------------------------
30467892Seric  * I'm getting "Local configuration error" messages, such as:
30567892Seric
30667892Seric	553 relay.domain.net config error: mail loops back to myself
30767892Seric	554 <user@domain.net>... Local configuration error
30867892Seric
30967892Seric    How can I solve this problem?
31067892Seric
31167892Seric	You have asked mail to the domain (e.g., domain.net) to be
31267892Seric	forwarded to a specific host (in this case, relay.domain.net)
31367892Seric	by using an MX record, but the relay machine doesn't recognize
31467892Seric	itself as domain.net.  Add domain.net to /etc/sendmail.cw
31567892Seric	(if you are using FEATURE(use_cw_file)) or add "Cw domain.net"
31667892Seric	to your configuration file.
31767898Seric
31867898Seric	IMPORTANT:  Be sure you kill and restart the sendmail daemon
31967898Seric	after you change the configuration file (for ANY change in
32067898Seric	the configuration, not just this one):
32167898Seric
32267898Seric		kill `head -1 /etc/sendmail.pid`
32367898Seric		sh -c "`tail -1 /etc/sendmail.pid`"
32467898Seric
32567898Seric	NOTA BENE:  kill -1 does not work!
32667892Seric----------------------------------------------------------------------
32766282Seric  * When I use sendmail V8 with a Sun config file I get lines like:
32866282Seric
32966282Seric	/etc/sendmail.cf: line 273: replacement $3 out of bounds
33066282Seric
33166282Seric    the line in question reads:
33266282Seric
33366282Seric	R$*<@$%y>$*		$1<@$2.LOCAL>$3			user@ether
33466282Seric
33566282Seric    what does this mean?  How do I fix it?
33666282Seric
33766282Seric	V8 doesn't recognize the Sun "$%y" syntax, so as far as it
33866282Seric	is concerned, there is only a $1 and a $2 (but no $3) in this
33966282Seric	line.  Read Rick McCarty's paper on "Converting Standard Sun
34066282Seric	Config Files to Sendmail Version 8", in the contrib directory
34166282Seric	(file "converting.sun.configs") on the sendmail distribution
34266282Seric	for a full discussion of how to do this.
34366282Seric----------------------------------------------------------------------
34466282Seric  * I'm connected to the network via a SLIP link.  Sometimes my sendmail
34566282Seric    process hangs (although it looks like part of the message has been
34666282Seric    transfered).  Everything else works.  What's wrong?
34766282Seric
34866282Seric	Most likely, the problem isn't sendmail at all, but the low
34966282Seric	level network connection.  It's important that the MTU (Maximum
35066282Seric	Transfer Unit) for the SLIP connection be set properly at both
35166282Seric	ends.  If they disagree, large packets will be trashed and
35266282Seric	the connection will hang.
35366282Seric----------------------------------------------------------------------
35466282Seric  * I just upgraded to 8.x and suddenly I'm getting messages in my
35566282Seric    syslog of the form "collect: I/O error on connection".  What is
35666282Seric    going wrong?
35766282Seric
35866282Seric    	Nothing.  This is just a diagnosis of a condition that had
35966282Seric    	not been diagnosed before.  If you are getting a lot of these
36066282Seric    	from a single host, there is probably some incompatibility
36166282Seric    	between 8.x and that host.  If you get a lot of them in general,
36266282Seric    	you may have network problems that are causing connections to
36366282Seric    	get reset.
36466282Seric----------------------------------------------------------------------
36567892Seric  * I just upgraded to 8.x and suddenly connections to the SMTP port
36667892Seric    take a long time.  What is going wrong?
36767892Seric
36867892Seric	It's probably something wierd in your TCP implementation that
36967892Seric	makes the IDENT code act oddly.  On most systems V8 tries to
37067892Seric	do a ``callback'' to the connecting host to get a validated
37167892Seric	user name (see RFC 1413 for details).  If the connecting
37267892Seric	host does not support such a service it will normally fail
37367892Seric	quickly with "Connection refused", but certain kinds of
37467892Seric	packet filters and certain TCP implementations just time out.
37567892Seric
37667892Seric	To test this, set the IDENT timeout to zero using
37767892Seric	``OrIdent=0'' in the configuration file.  This will
37867892Seric	completely disable all use of the IDENT protocol.
37967892Seric
38067892Seric	Another possible problem is that you have your name server
38167892Seric	and/or resolver configured improperly.  Make sure that
38267892Seric	all "nameserver" entries in /etc/resolv.conf point to functional
38367892Seric	servers.  If you are running your own server make certain that
38467892Seric	all the servers listed in your root cache (usually called
38567892Seric	something like "/var/namedb/root.cache"; see your
38667892Seric	/etc/named.boot file to get your value) are up to date.
38767892Seric	Either of these can cause long delays.
38867892Seric----------------------------------------------------------------------
38967892Seric  * I just upgraded to 8.x and suddenly I get errors such as ``mail:
39067892Seric    options must follow recipients.''  What is going wrong?
39167892Seric
39267892Seric	You need OSTYPE(systype) in your .mc file -- otherwise the
39367892Seric	configurations use a default that probably disagrees with
39467892Seric	your local mail system.  See cf/README for details.
39567892Seric----------------------------------------------------------------------
39666282Seric  * Under V8, the "From " header gets mysteriously munged when I send
39766282Seric    to an alias.
39866282Seric
39966282Seric	``It's not a bug, it's a feature.''  This happens when you have
40066282Seric	a "owner-list" alias and you send to "list".  V8 propogates the
40166282Seric	owner information into the envelope sender field (which appears
40266282Seric	as the "From " header on UNIX mail or as the Return-Path: header)
40366282Seric	so that downstream errors are properly returned to the mailing
40466282Seric	list owner instead of to the sender.  In order to make this
40566282Seric	appear as sensible as possible to end users, I recommend making
40666282Seric	the owner point to a "request" address -- for example:
40766282Seric
40866282Seric		list:		:include:/path/name/list.list
40966282Seric		owner-list:	list-request
41066282Seric		list-request:	eric
41166282Seric
41266282Seric	This will make message sent to "list" come out as being
41366282Seric	"From list-request" instead of "From eric".
41466282Seric----------------------------------------------------------------------
41566285Seric  * I want to run Sendmail version 8 on my DEC system, but you don't
41666285Seric    have MAIL11V3 support in sendmail.  How do I handle this?
41766285Seric
41866285Seric	Get Paul Vixie's reimplementation of the mail11 protocol
41966285Seric	from gatekeeper.dec.com in /pub/DEC/gwtools.
42066285Seric----------------------------------------------------------------------
421*68069Seric  * Messages seem to disappear from my queue unsent.  When I look in
422*68069Seric    the queue directory I see that they have been renamed from qf*
423*68069Seric    to Qf*, and sendmail doesn't see these.
424*68069Seric
425*68069Seric	If you look closely you should find that the Qf files are
426*68069Seric	owned by users other than root.  Since sendmail runs as root
427*68069Seric	it refuses to believe information in non-root-owned qf files,
428*68069Seric	and it renames them to Qf to get them out of the way and
429*68069Seric	make it easy for you to find.  The usual cause of this is
430*68069Seric	twofold:  first, you have the queue directory world writable
431*68069Seric	(which is probably a mistake -- this opens up other security
432*68069Seric	problems) and someone is calling sendmail with an "unsafe"
433*68069Seric	flag, usually a -o flag that sets an option that could
434*68069Seric	compromise security.  When sendmail sees this it gives up
435*68069Seric	setuid root permissions.
436*68069Seric
437*68069Seric	The usual solution is to not use the problematic flags.
438*68069Seric	If you must use them, you have to write a special queue
439*68069Seric	directory and have them processed by the same uid that
440*68069Seric	submitted the job in the first place.
441*68069Seric----------------------------------------------------------------------
442