[comp.bugs.sys5] Uucp forwarding with Honey-DanBer

jw@pan.UUCP (Jamie Watson) (07/31/87)

--------------

The uucp distributed with System V, at least up to release 2.0, contained
specific administration files for control of uucp forwarding - that is, use
of commands such as 'uucp file pi0!pi1!pi2!~/'.  If memory serves, they
were called FWDFILE and ORIGFILE.

I am now running System V release 3.0, with HDB uucp.  After a careful
search of everything I can think of in the documentation, including all the
manual pages for all the various uucp programs, and both bnu articles in
the System Administrator's Guide, I can't find anything equivalent in HDB.

Testing forwarding with two machines running V.3, the logfile entries led
me to believe that my forwarding commands were being turned into something
that looked approximately like 'uux pi1!uucp ...', and I was being denied
permission to execute uucp on the remote machine.  So, I added a line to
the Permissions file granting permission to execute uucp to the test
systems.  This seemed to make things work - the file I was sending showed
up on the machine I was sending it to.  So, the first question is:

    - Is this the correct (and only) way to get forwarding to work?

Next, I tried to forward from a System V release 2 machine through a
release 3 machine, with this command: 'uucp /usr/jw/foo pi0!paninfo!pi1!~/'.
It failed, and mailed me back an error message from pi0 (the V.2 system)
telling me that 'Remote Access to file /usr/jw/foo on system paninfo (the
V.3 machine) was Denied'!  Surprising, since by the time it got to paninfo,
the file should have been some ordinary uucp temp file - there was no
reason for the V.3 machine to be looking in /usr/jw.  Examining the work
files that are created, I see that the V.2 machines create a very different
command file (X.*) than the V.3 machines do.  Second question:

    - Is there perhaps an incompatibility here between V.2 and V.3?
      If so, what can be done about it?  Are we just stuck?

Finally, please don't flame me about the risks involved with allowing uucp
forwarding.  I know the problems.  We have a relatively large number of
System V machines internally, and we need this functionality.  The security
issue is one of the reasons I'm concerned about what I think is the HDB
method of allowing forwarding.

Thanks in advance for any help.

Jamie Watson			...!mcvax!cernvax!paninfo!jw
Pansystem Informatics, Ltd.
Bahnhofstrasse 50		Tel: 1/833.36.44
CH-8305 Dietlikon		Twx: 827 332 pi
Switzerland			Fax: 1/833.19.59

ka@hropus.UUCP (Kenneth Almquist) (08/10/87)

	Is there perhaps an incompatibility here between V.2 and V.3?
	If so, what can be done about it?  Are we just stuck?

The multihop file transfer stuff in System V release 1 and 2 will not
work with older UUCP implementations, including HoneyDanBer.  HDB
*should* have been upgraded to understand the System V multihop file
transfer when it was made the standard UUCP for System V.  If it wasn't,
the only solution is to file a bug report (which *might* not be ignored)
and go back to SVR2 uucp.

	The security issue is one of the reasons I'm concerned about
	what I think is the HDB method of allowing forwarding.

Right.  Another problem with the way that HDB handles multihop file
transfers is that it does not support multihop fetches, i. e. a command
like "uucp a!b!~/ka/file ." will not work.  The advantage of the HDB
approach is that it will let you forward through systems running any
version of the UUCP software.
					Kenneth Almquist