bob@MorningStar.Com (Bob Sutterfield) (11/22/90)
In article <4279@awdprime.UUCP> dcheney@dcheney.austin.ibm.com (David J. Cheney) writes:
An important point needs to be made about mail: many people currently have
<name>@<machine>.austin.ibm.com
in their ~/.signature files. If <machine>.austin.ibm.com is not
pingable, you CANNOT successfully deliver mail to <name> at or via
that machine,
The best way to find out mail paths to a specific user is to call
the person and ask. We are evaluating alternative approaches to
solving this problem. IBM Austin employees without approved nodes
have been asked to correct their signature files.
The problem isn't .signature files (that users control), it's their
mail and news headers (that the system administrators control). If my
friend sends me mail specifying "From:
whoever@machine.austin.ibm.com" and there's no Reply-To: line in the
headers, then my mailer *must* attempt delivery to the machine named
in the From: line. If I receive mail from a user, it's reasonable
(and normal practice in the rest of the Internet) to assume that I can
reply to the mail without bothering to call him on the telephone.
If you're going to have a policy that selectively isolates machines,
then please completely implement the policy. Don't allow your
machines to generate unreplyable message headers. If a machine cannot
accept messages, than don't let messages from that machine leak into
the world outside your wall.
See RFC1123 (Requirements for Internet Hosts -- Application and
Support), section 5.3.7(D) on mail gatewaying (which is what you're
doing, selectively, between IBM's internal environment and the
Internet):
(D) The gateway MUST ensure that all header fields of a
message that it forwards into the Internet meet the
requirements for Internet mail. In particular, all
addresses in "From:", "To:", "Cc:", etc., fields must be
transformed (if necessary) to satisfy RFC-822 syntax, and
they must be effective and useful for sending replies.
Mail leaking from non-approved hosts at austin.ibm.com violates the
last phrase in the last sentence of that paragraph.
I don't want to flame IBM and wave RFCs at you, but if you're going to
bring up the subject of your nonconformant mailers in a public forum,
then you've set yourself up as fair game. I've redirected followups
to comp.mail.misc.
And, lacking a way to get private mail to rangoon.austin.ibm.com,
would you please convey my thanks to Win Bo for his congratulations on
the birth of my son? Lauri, Andy, and I hope that he, Than, and Ryan
are also doing well, but I have been unable to respond privately to
his gracious note, conveyed to me via private mail. See how silly the
effects of selective isolation policies can be?fletcher@cs.utexas.edu (Fletcher Mattox) (11/22/90)
In article <4279@awdprime.UUCP> dcheney@dcheney.austin.ibm.com (David J. Cheney) writes: >The best way to find out mail paths to a specific user (AWD has for some >time had the uucp mail path: > @cs.utexas.edu:ibmchs!auschs!<machine>.austin.ibm.com!<user> >) is to call the person and ask. We are evaluating alternative approaches >to solving this problem. Can someone please explain to me why IBM doesn't simply advertise MX records which point to cs.utexas.edu for all these "internal" hosts? I will happily forward all such mail via the !-path above. This would go a long way toward fixing the wretched state of email to IBM Austin. All those mail and news headers would suddenly start working! Heck, I could save several hours a month by not having to explain to dozens of people why their mail to IBM bounces. (David's suggestion to phone the person notwithstanding, my mailbox tells me that many folks simply ask <postmaster@cs.utexas.edu>). I make this suggestion annually, by the way. Who knows, maybe I'll get lucky one of these years.
kls@ditka.UUCP (Karl Swartz) (11/24/90)
In article <14941@cs.utexas.edu> fletcher@cs.utexas.edu (Fletcher Mattox) writes: >In article <4279@awdprime.UUCP> dcheney@dcheney.austin.ibm.com (David J. Cheney) writes: >>The best way to find out mail paths to a specific user (AWD has for some >>time had the uucp mail path: >> @cs.utexas.edu:ibmchs!auschs!<machine>.austin.ibm.com!<user> >>) is to call the person and ask. >Can someone please explain to me why IBM doesn't simply advertise MX >records which point to cs.utexas.edu for all these "internal" hosts? Perhaps they haven't yet heard about MX records and mail exchangers? When last I looked at sendmail on AIX (on an RT a bit over a year ago) it hadn't a clue about MX records (in addition to being rife with severe brain damage of various sorts). I haven't gotten around to looking on the RS/6000 yet but so far /bin/cat is the only utility I've found so far that isn't broken on this "new and improved" RT so I don't hold out much hope. -- Karl Swartz |UUCP {uunet,decwrl}!daver!ditka!kls 1-408/223-1308 |INet kls@ditka.chicago.com "I never let my schooling get in | the way of my education."(Twain) |Snail 1738 Deer Creek Ct., San Jose CA 95148
jackv@turnkey.tcc.com (Jack F. Vogel) (11/25/90)
In article <31914@ditka.UUCP> kls@ditka.UUCP (Karl Swartz) writes: >Perhaps they haven't yet heard about MX records and mail exchangers? >When last I looked at sendmail on AIX (on an RT a bit over a year ago) >it hadn't a clue about MX records (in addition to being rife with >severe brain damage of various sorts). You are correct about the RT, its sendmail is very old and does not have MX support. Sendmail on the 6000, however definitely does have MX, even though it is still based on the old code. AIX 1.2 (that's 370 and PS/2) as of a couple of updates back has a pure 5.61 port that I did. Don't know if the 6000 folk will pick that up or not. Basically I think its more an administrative lack of experience than not having the code to do it. Disclaimer: I speak for myself, not LCC nor IBM. -- Jack F. Vogel jackv@locus.com AIX370 Technical Support - or - Locus Computing Corp. jackv@turnkey.TCC.COM
smoot@cs.utexas.edu (Smoot Carl-Mitchell) (11/25/90)
In article <1990Nov24.175319.26238@turnkey.tcc.com> jackv@turnkey.TCC.COM (Jack F. Vogel) writes: >You are correct about the RT, its sendmail is very old and does not have >MX support. Sendmail on the 6000, however definitely does have MX, even >though it is still based on the old code. AIX 1.2 (that's 370 and PS/2) >as of a couple of updates back has a pure 5.61 port that I did. Don't >know if the 6000 folk will pick that up or not. Which is silly, since I ported a reasonably up to date version of sendmail with MX support to an old RT and used it successfully as a mail gateway, until it was replaced by a Sun. Really was not that hard to do. Maybe someone will do the same with the IDA Sendmail code, so RS6000 users will have a reasonable implementation. -- Smoot Carl-Mitchell, Texas Internet Consulting smoot@tic.com, smoot@cs.utexas.edu
paul@uxc.cso.uiuc.edu (Paul Pomes - UofIllinois CSO) (11/25/90)
smoot@cs.utexas.edu (Smoot Carl-Mitchell) writes: >Maybe someone will do the same with the IDA Sendmail code, so RS6000 users >will have a reasonable implementation. 'Tis done except that it won't work with sendmail.fc files. It will work if the system-supplied malloc.o and getpwent.o functions are replaced (the latter has the entry point for the kernel malloc linked in). A new release of the 5.65+IDA is due out next week. It has a raft of portability improvements and now compiles with gcc requiring only "-fpcc-struct-return". /pbp -- Paul Pomes UUCP: {att,iuvax,uunet}!uiucuxc!paul Internet, BITNET: paul@uxc.cso.uiuc.edu US Mail: UofIllinois, CSO, 1304 W Springfield Ave, Urbana, IL 61801-2910