[unix-pc.general] UNIXPC uucp problem

becker@ziebmef.uucp (Bruce Becker) (06/28/88)

I've been having a problem with my 7300/3b1, in that it won't
receive uucp logins from anyone at all (BAD LOGIN/MACHINE NAME), but I
can connect to any machine by dial-out... I don't understand why...
If any 3b1 users have had this problem, please email some suggestions
to me...

Thanks, Bruce Becker
UUCP: ...!unicus!becker!bdb, ...!lsuc!humvax!becker, ...!ncrcan!ziebmef!becker
BitNet: BECKER@HUMBER.BITNET

jeff@cjsa.UUCP (C. Jeffery Small) (06/30/88)

In article <1988Jun27.202651.9458@ziebmef.uucp>, becker@ziebmef.uucp (Bruce Becker) writes:
> I've been having a problem with my 7300/3b1, in that it won't
> receive uucp logins from anyone at all (BAD LOGIN/MACHINE NAME), but I
> can connect to any machine by dial-out... I don't understand why...
> If any 3b1 users have had this problem, please email some suggestions
> to me...

I had this very problem when I first set up USENET software.  I believe
the problem has to do with the way uucp, nuucp, etc. entries appear and
are parsed in the passwd file.  I believe that your problem will go away
(and your security increased) if you make seperate password entries with
unique uids for each machine that you wish to allow to dial in.  

I hope this helps.
--
Jeffery Small          (203) 776-2000     UUCP:   uunet!---\
C. Jeffery Small and Associates	                  ihnp4!--- hsi!cjsa!jeff
123 York Street, New Haven, CT  06511          hao!noao!---/

brant@manta.UUCP (Brant Cheikes) (07/03/88)

In article <103@cjsa.UUCP> jeff@cjsa.UUCP (C. Jeffery Small) writes:
>In article <1988Jun27.202651.9458@ziebmef.uucp>, becker@ziebmef.uucp writes:
>> I've been having a problem with my 7300/3b1, in that it won't
>> receive uucp logins from anyone at all (BAD LOGIN/MACHINE NAME), but I
>> can connect to any machine by dial-out... I don't understand why...
[...]
>I had this very problem when I first set up USENET software.  I believe
>the problem has to do with the way uucp, nuucp, etc. entries appear and
>are parsed in the passwd file.

My suggestion would be to check the /etc/inittab file for an entry
that attaches a getty (or uugetty) to the modem port.  You must have a
line of the form:

 ph1:2:respawn:/usr/lib/uucp/uugetty -r -t60 ph1 1200

(leading space essential)
This puts a uugetty on the phone line waiting for an incoming
connection.  I believe the getty initializes the port to answer
incoming calls.  You may need to replace /usr/lib/uucp/uugetty above
with /etc/getty.  Without a getty, your machine won't answer.
-- 
Brant Cheikes
University of Pennsylvania
Department of Computer and Information Science
Internet: manta!brant@bpa.bell-atl.com, UUCP: {bpa,drexel}!manta!brant

mhm@cbterra.ATT.COM (Mike H. Moran) (07/04/88)

In article <382@manta.UUCP>, brant@manta.UUCP (Brant Cheikes) writes:
> In article <103@cjsa.UUCP> jeff@cjsa.UUCP (C. Jeffery Small) writes:
> >In article <1988Jun27.202651.9458@ziebmef.uucp>, becker@ziebmef.uucp writes:

[ much stuff about inittab enteries, etc ....]

> This puts a uugetty on the phone line waiting for an incoming
> connection.  I believe the getty initializes the port to answer
> incoming calls.  You may need to replace /usr/lib/uucp/uugetty above
> with /etc/getty.  Without a getty, your machine won't answer.

Not quite true, uugetty is supposed to be intelligent :-) in that it
will answer incomming calls as well as allow out going calls on the
same port without need of inittab modfication.

Mike Moran 
Contracted to AT&T-BL			UUCP:  att!cbosgd!mhm
Columbus, Ohio				       mhm@cbosgd.att.com  

jcs@tarkus.UUCP (John C. Sucilla) (07/05/88)

In article <382@manta.UUCP> brant@manta.UUCP (Brant Cheikes) writes:
>In article <103@cjsa.UUCP> jeff@cjsa.UUCP (C. Jeffery Small) writes:
>>In article <1988Jun27.202651.9458@ziebmef.uucp>, becker@ziebmef.uucp writes:
>>> I've been having a problem with my 7300/3b1, in that it won't
>>> receive uucp logins from anyone at all (BAD LOGIN/MACHINE NAME), but I
>>I had this very problem when I first set up USENET software.  I believe
>>the problem has to do with the way uucp, nuucp, etc. entries appear and
>>are parsed in the passwd file.
>My suggestion would be to check the /etc/inittab file for an entry
>that attaches a getty (or uugetty) to the modem port.  You must have a
>line of the form:

The fact that he's getting a bad login/machine name in his log files are
proof that he has a getty of some kind running on the port.  I'd say
that /usr/lib/uucp/Permissions is the problem.  Try temporarily setting
it to the lowest possible level of security and see what happens.

This ought work:

MACHINE=OTHER REQUEST=yes READ=/ WRITE=/ SENDFILES=yes COMMANDS=rmail:rnews:uux

Later, if you find that the Permissions file  (I think this was the users file
in pre-HoneyDanber uucp) was to blame you can start getting fancy with it
by creating seperate entries for each machine you talk to.


-- 
John "C". Sucilla,  A silicon based life form.
       {ihnp4,chinet,ddsw1}!tarkus!jcs
  You have a better idea? Now's the time..

cks@ziebmef.uucp (Chris Siebenmann) (07/05/88)

 Bruce and I managed to track down and solve (sort of) this problem;
it was apparently a bad USERFILE. However, I looked at it and it
looked fine, and very similar to the one I use here. Are there any
known limitations of the 3.51 USERFILE? 

 Also, are there any guides to 'troubleshooting UUCP connections'?
I've got quite good at finding problems (having stubbed my toes on
most of them by now :-(), but there are undoubtedly other traps just
lurking out there just waiting for me. I have the Nutshell handbook
_Managing UUCP and USENET_ and have found it quite valuable; is there
anything else people would recommend? Better yet, is there some
documentation on the bugs ... ahem, idiosyncracies of the stock 3B1
uucp?

 While I'm here ... Hint #38: Have an external modem you need to dial
with a chat script (to, say, dial something with pulse and then switch
to tone-dialing to get through a PBX), but don't want to always assert
CD on the modem (uucico requires DCD before it will go into the chat
script on a direct connection)? Just define a really dumb modem that
does nothing in modemcap, and point the appropriate L-devices entry at
it.  Unfortunately, this means you have to dial via chat scripts for
all connections over that modem at that speed, but it's better than
nothing...

-- 
	You're a prisoner of the dark sky/The propeller blades are still
	And the evil eye of the hurricane's/Coming in for the kill
Chris Siebenmann		uunet!utgpu!{ontmoh!moore,ncrcan}!ziebmef!cks
cks@ziebmef.UUCP	     or	.....!utgpu!{,ontmoh!,ncrcan!brambo!}cks

cks@ziebmef.uucp (Chris Siebenmann) (07/05/88)

In article <103@cjsa.UUCP> jeff@cjsa.UUCP (C. Jeffery Small) writes:
...
>I had this very problem when I first set up USENET software.  I believe
>the problem has to do with the way uucp, nuucp, etc. entries appear and
>are parsed in the passwd file.  I believe that your problem will go away
>(and your security increased) if you make seperate password entries with
>unique uids for each machine that you wish to allow to dial in.  

 Actually, all uucp dialins have to share the same UID as 'uucp'. If
they don't, connections fail with a very obscure error message (I
believe 'BAD LOGIN/MACHINE NAME' again) -- I got bitten by this when I
first set up the uucp logins on ziebmef.

 In hindsight, this makes sense; it means that people can't just
invoke uucico from their own accounts and pretend to be some random
machine. Instead you somehow have to set your uid to the 'uucp' UID.
Could have been worse, I suppose; they could have hardcoded the uucp
UID into uucico instead of looking it up in /etc/passwd.

-- 
	You're a prisoner of the dark sky/The propeller blades are still
	And the evil eye of the hurricane's/Coming in for the kill
Chris Siebenmann		uunet!utgpu!{ontmoh!moore,ncrcan}!ziebmef!cks
cks@ziebmef.UUCP	     or	.....!utgpu!{,ontmoh!,ncrcan!brambo!}cks

jeff@cjsa.UUCP (C. Jeffery Small) (07/06/88)

In article <1988Jul4.173733.7088@ziebmef.uucp>, cks@ziebmef.uucp (Chris Siebenmann) writes:
> 
>  Actually, all uucp dialins have to share the same UID as 'uucp'. If
> they don't, connections fail with a very obscure error message (I
> believe 'BAD LOGIN/MACHINE NAME' again) -- I got bitten by this when I
> first set up the uucp logins on ziebmef.
> 
>  In hindsight, this makes sense; it means that people can't just
> invoke uucico from their own accounts and pretend to be some random
> machine. Instead you somehow have to set your uid to the 'uucp' UID.
> Could have been worse, I suppose; they could have hardcoded the uucp
> UID into uucico instead of looking it up in /etc/passwd.

This is certainly not true here.  I have fifteen or so accounts set up
for uucico login - all with different UIDs.  However, they each share the
uucp (ie. mail) group ID.  Since uucico doesn't run setgid, this doesn't
seem to matter one way or the other.

Possibly someone with source could take a look and see just what sequence
of events occur when a uucico session starts up.  I would be interested in
being enlightened further.
--
Jeffery Small          (203) 776-2000     UUCP:   uunet!---\
C. Jeffery Small and Associates	                  ihnp4!--- hsi!cjsa!jeff
123 York Street, New Haven, CT  06511          hao!noao!---/

becker@ziebmef.uucp (Bruce Becker) (07/07/88)

In article <3915@cbterra.ATT.COM> mhm@cbterra.ATT.COM (Mike H. Moran) writes:
>In article <382@manta.UUCP>, brant@manta.UUCP (Brant Cheikes) writes:
>> In article <103@cjsa.UUCP> jeff@cjsa.UUCP (C. Jeffery Small) writes:
>> >In article <1988Jun27.202651.9458@ziebmef.uucp>, becker@ziebmef.uucp writes:
>
>[ much stuff about inittab enteries, etc ....]
>
>>[ much stuff about uugetty, etc ....]
>
>Not quite true, uugetty is supposed to be intelligent :-) in that it
>will answer incomming calls as well as allow out going calls on the
>same port without need of inittab modfication.

So where can I get a copy of uugetty for the 3b1?

Cheers, Bruce Becker
UUCP: ...!unicus!becker!bdb, ...!lsuc!humvx!becker, ...!ncrcan!ziebmef!becker
BitNet: BECKER@HUMBER.BITNET

rjg@sialis.mn.org (Robert J. Granvin) (07/08/88)

> Bruce and I managed to track down and solve (sort of) this problem;
>it was apparently a bad USERFILE. However, I looked at it and it
>looked fine, and very similar to the one I use here. Are there any
>known limitations of the 3.51 USERFILE? 

One.  If you add your 101st entry, it'll "die."

Oh, it's even better than that.  No system will ever be able to make a
UUCP connection with your machine until you bring to to 100 entries or
under.  Neat?  

> Also, are there any guides to 'troubleshooting UUCP connections'?
>I've got quite good at finding problems (having stubbed my toes on
>most of them by now :-(), but there are undoubtedly other traps just
>lurking out there just waiting for me. I have the Nutshell handbook
>_Managing UUCP and USENET_ and have found it quite valuable; is there
>anything else people would recommend? Better yet, is there some
>documentation on the bugs ... ahem, idiosyncracies of the stock 3B1
>uucp?

The 3b1 has a goofy OBM.  (Also, please refer to some immediately
previous replies to other messages).

Asides from handshaking an OBM connection and then not knowing what to
do from there, it's also something that is apparently out of spec from
certain areas.  Some modems, including Telebit's under certain
connections will always hang up after attempting to establish a
connection with the given carrier.  The "unusual" thing about it also,
is that if you dial _out_ using the OBM into one of these modems, the
same connection will be rejected as well.


Otherwise, the 3b1 generally operates similarly or as well as other
System V machines/implementations (not taking the bug list into
account, of course :-)

-- 
"I've been trying for some time to                           Robert J. Granvin
 develop a life-style that doesn't          National Information Systems, Inc.
 require my presence."                                       rjg@sialis.mn.org
    -Garry Trudeau                ...{{amdahl,hpda}!bungia,rosevax}!sialis!rjg

james@bigtex.uucp (James Van Artsdalen) (07/16/88)

IN article <1988Jul4.173733.7088@ziebmef.uucp>, cks@ziebmef.UUCP (Chris Siebenmann) wrote:
>  Actually, all uucp dialins have to share the same UID as 'uucp'. If

uucico should be suid to the uucp user.  uucico doesn't care what the
real uid is.  It does use the login name to reference the Permissions
file (assuming we're talking HDB uucp).

>  In hindsight, this makes sense; it means that people can't just
> invoke uucico from their own accounts and pretend to be some random
> machine. Instead you somehow have to set your uid to the 'uucp' UID.
> Could have been worse, I suppose; they could have hardcoded the uucp

Actually, I understand that the uucp uid is indeed a configurable
constant in the uucp package.  I can't imagine what it's used for
though: if someone forgets to make uucico suid, that will be obvious
soon enough, and file read/write permissions are based on the
Permissions file and whether or not the file is readable/writable by
"other".

The security measures to prevent someone from spoofing uucp this way
revolve around the use of the Permissions file and the VALIDATE
keyword (if you don't use VALIDATE, do so).

When uucico is started in master mode, the transfer permissions are
granted based on the machine name.  But when uucico is started in
slave mode (as it would be if a user ran it), permissions are granted
based on the login name, *not* the machine name.  This is why VALIDATE
is so important.  Note that uucico takes the login name from /etc/utmp
to prevent spoofing (or at least it should).
-- 
James R. Van Artsdalen   ...!ut-sally!utastro!bigtex!james   "Live Free or Die"
Home: 512-346-2444 Work: 328-0282; 110 Wild Basin Rd. Ste #230, Austin TX 78746