Date: Wed, 25 Oct 1995 10:30:40 -0500 (CDT) From: Guy Helmer <ghelmer@alpha.dsu.edu> To: "Jordan K. Hubbard" <jkh@time.cdrom.com> Cc: Ollivier Robert <roberto@keltia.freenix.fr>, "Jonathan M. Bresler" <jmb@kryten.atinc.com>, gary@palmer.demon.co.uk, chat@freebsd.org Subject: Re: moving some mail. Message-ID: <Pine.OSF.3.91.951025101327.29569B-100000@alpha.dsu.edu> In-Reply-To: <27644.814631999@time.cdrom.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, 25 Oct 1995, Jordan K. Hubbard wrote: > > Like Bill has said, it is the paradigm used Usenet. It works but it is much > > slower than mail (even with the timeouts). The vast majority of sites are > > three or four hops away from the other. Usenet can make it more then > > twenty... > > I think what *really* needs to happen is for sendmail to get a lot > smarter about this.. Sendmail knows who it can't reach, and if it > kept timing statistics for some number of "frequent destinations" then > it could even intuit who was slow and who was fast, reordering its > work queue accordingly. > > Jordan Uh-oh - more code in sendmail! :-) I've got a Perl script written by Paul Pomes (at UIUC or CICnet) that I believe solves this sort of problem nicely. The script moves older messages from one queue directory to another, such as from /var/spool/mqueue to /var/spool/mqueue2; one infrequently runs "sendmail -oQ/var/spool/mqueue2 -q" from root's crontab for the second level queue, and likewise even more levels can be setup for less frequent intervals. Keeping the main queue short seems to help sendmail out quite a bit. I haven't had to implement this yet on DSU's mail hub (some days I have been tempted!), but it wouldn't be hard to install. I could forward the script to anyone who would like it. Guy Helmer, Dakota State University Computing Services - ghelmer@alpha.dsu.edu
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.OSF.3.91.951025101327.29569B-100000>