harbo@diku.dk (Klaus Harbo) (01/04/91)
We have several DS3100s running V3.1 and will soon be upgrading to V4.1. However, one of the machines works as mail-platform for a lot of people, and has a somewhat non-standard sendmail.cf, the details of which I know very little. - A lot of people around here depend heavily on e-mail, and absence of mail would be more or less a catastrophe. Has anyone experienced any problems with sendmail, going from V3.1 to ver. 4.0/4.1 that we should be aware of before we upgrade? Any advice? Thanks in advance. Klaus Harbo |--------------------------------------------------------------------------| | Klaus Harbo | | | Euromath Center | e-mail: harbo@euromath.dk | | Universitetsparken 5 | phone: +45 3135 3133 ext. 417 | | DK-2100 Copenhagen | fax: +45 3135 0427 | |--------------------------------------------------------------------------|
DEEJ@MAINE.BITNET (Dj Merrill) (01/05/91)
I have recently upgraded from 3.1 to ultirx 4.0 on Decstation 3100s. I do have a slight weirdness with mail (although it may not be from the upgrade - I only recently assumed responsibility of these machines). When I do a 'reply' from a mail file that comes from a foreign (meaning not one of the other 3100s we have), mail puts the word 'on' in front of the reply address in the To line: Ex. To: on BLANES@MECAN1.maine.edu The 'on' does not appear anywhere in the original mail header that I can tell. Consequently, the mailer tries to send a copy of the mail to a userid 'on', which of course does not exist. I may have munged something up when upgrading, but I can't figure out why it does this. All seems fine if replying to a mail file from one of the other Decstations on our network. +------------------------------+ | Dj Merrill | | Systems and Network Manager | | University of Maine NCGIA | | | | deej@grouse.umesve.maine.edu | | deej@maine.maine.edu | +------------------------------+
avolio@decuac.dec.com (Frederick M. Avolio) (01/05/91)
There is a bug in the /usr/ucb/mail program with ULTRIX 4.0 (at least -- I don't know if it is fixed with 4.1). There is a buffer being used to hold an addressee's hostname to check against your own. Anyway, the buffer is too short for your hostname (and I assume other hostnames as well). If you have sources, patch optim.c -- fakepath needs to be longer than 20. (80 is probably big enough :-). ) I've sent a note along about it to the folks who maintain it. You might check to see if there is a patch available as this will bit anyone who's hostname.domain is longer than 20 characters. Fred