Counts vs. Many

Vernon Schryver vjs@calcite.rhyolite.com
Fri Sep 27 15:17:57 UTC 2002


> From: <skahil@iclassfactory.com>

> The thought occurred to me that there may be a limit to the usefulness of
> reporting the number of times an email was received as an absolute number.
> Specifically, it may have problems scaling.
>
> For example: If a small server starts exchanging checksums with a very large
> server, the user's thresholds as absolute numbers would no longer have the
> same affect.

How so?  I assume the users of your DCC small server chose low thresholds
because the small DCC was getting a small sample.  If there are only
a half a dozen mailboxes using the small server, then a bulk threshold
of 6 would make more sense than 20.
When the small server starts exchanging checksums with the large server,
the small server will hear reports of checksums with 20 or more recipients.
That's still bulk mail when it arrives at the small server.

Please also consider `dccd -t` in the dccd man page and notice the
automatic adjustment of the dccd flooding threshold in /var/dcc/dcc_conf
and /var/dcc/libexec/start-dccd.


> Would it be possible for the server to return both the count, as well as
> some sort of metric indicating the significance of the count?
>
> The simplest thing would be to just report the number of checksums in the
> database.
> The user could then make a decision based on a ratio.
>
> Alternatively, it could also incorporate the average age of the checksums in
> the database or whatever.

The number of checksums in a DCC server's database can be obtained with
`cdcc stats`.


Vernon Schryver    vjs@rhyolite.com



More information about the DCC mailing list

Contact vjs@rhyolite.com by mail or use the form.