home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.admin.policy
- Path: sparky!uunet!ferkel.ucsb.edu!ferkel!jim
- From: jim@ferkel.ucsb.edu (Jim Lick)
- Subject: Re: Full mail spool policy?
- Message-ID: <jim.722134280@ferkel>
- Organization: University of California, Santa Barbara
- References: <1992Nov5.194200.3673@news.columbia.edu> <BxwzoA.325@gabriel.keele.ac.uk>
- Date: Thu, 19 Nov 1992 00:51:20 GMT
- Lines: 71
-
- In <BxwzoA.325@gabriel.keele.ac.uk> cca09@keele.ac.uk (Norman Bridges) writes:
- >That's getting a real problem here too. I feel that a responsible
- >automatic mail server should be auditing its readership once in a
- >while and asking readers to specifically re-subscribe.
-
- Sorry, but that's just not going to happen. Even with an automated
- system, this is going to create a collosal headache for the admin
- and the subscribers.
-
- > I also feel
- >that not enough is done to give sysadmins facilities to unsubscribe to
- >mailing lists on behalf of dead accounts. [My apologies if such
- >facilities do exist, its just that it can be difficult to locate the
- >address for unsubscription without prying into a user's mailbox].
-
- The usual mechanism is to remove the account. Most decent lists will
- have bounces directed to an admin address. Others will bounce to the
- poster. Sometimes the poster will forward these to the admin. Otherwise
- the list admin will post a message periodically to catch bounces.
-
- Apropos to the original subject, many lists employ a system where mail
- to a user bounces if their mail file is over some limit. Some mail
- systems (Sendmail+IDA for example) allow you to setup customized
- per user bounce messages. Examples include "joeuser has not logged
- in in over 2 weeks; mail rejected" or "joeuser account has been
- disabled; mail rejected".
-
- I've received mail from some admins asking to please remove so-and-so
- from your list, so obviously some admins will go through the users mail
- file. I don't particularly like this approach though, as it may be in
- violation of the user's privacy. I would recommend the following
- approaches to these problems:
-
- Problem: users often have multi-megabyte mail spool files, and the
- spool area is in danger of filling up.
-
- Solution: Impose a mail spool size limit. When a user reaches the
- limit, mail the user a note along the lines of "Your incoming mail
- file has exceeded out xxxKB limit. Incoming mail will be disabled
- until you trim your file to less than this limit." then bounce all
- additional mail to that user back with the message "joeuser has
- exceeded the local mail file size limit." until the file size shrinks.
-
- Problem: users subscribe to mailing lists but then don't log in for
- several weeks.
-
- Solution: If you are only concerned with the file sizes, the above
- size limit should be sufficient. If user en abstentia is your main
- problem, you could start bouncing mail if a user hasn't logged in in
- a while. This especially seems to be a problem over the summer
- vacation.
-
- Problem: user's account has been removed or disabled
-
- Solution: It is usually sufficient to just delete the account. However,
- no such user bounces don't always mean that the user is gone. Sometimes
- it means oops our NIS just broke or oops our system is going bonkers.
- ocf.berkeley.edu is notorious for periodically bouncing mail with the
- 'no such user' message when really it was just having a bad day. I'll
- usually allow bounces to continue for a few days in case the problem
- was temporary, but if it was bounced back with the specific message that
- the user no longer has an active account, then I'd delete the address
- immediately.
-
- The advantages of these systems is that once they are set up, if your
- software allows it, then everything is basically on auto-pilot. You
- also don't have to rifle through private files either.
-
- --- jim@ferkel.ucsb.edu --- Jim Lick --- jim@tcp.com --- jIngOrO@CaveMUCK ---
- ------- perfect little dream the kind that hurts the most - |\| | |/| -------
- --- CaveMUCK is back! Telnet to cave.tcp.com (128.95.10.106) port 2283 ---
-