On 7/11/2013 9:01 PM, Mike Brown wrote:
> W T Riker wrote:
>> I suspect someone has already done this somewhere but I can't seem to
>> come up with the right key words in my search. I'd like to set up spam
>> and ham mailboxes to which all my users can redirect/bounce errors for
>> Bayes training for the site. Then I can run sa-learn via cron against a
>> single mailbox. Can someone point me to some info on this? Thanks.
> You want to know how to configure "aliases" in your MTA, i.e. sendmail,
> postfix, exim, or qmail. They all support files as the destination.
> So I suggest doing a web search for the name of your MTA plus "aliases".
>
> See also https://wiki.apache.org/spamassassin/SiteWideBayesFeedback -
> at the end it points to a postfix-specific recipe.
>
Thanks for the reply. Sorry if my question is not clear but I know how
to set up aliases (although I don't understand why that is needed) and
how to have users redirect misclassified messages. What I am struggling
with is the other side of the equation. Most of my users access mail via
Squirrelmail or a remote mail reader and don't log on to the OS. I THINK
I can create ham and spam user mboxes then use sa-learn to process each
via cron. Users would redirect misclassified mail to the appropriate
ham/spam user. The question is, is that sufficient such that headers are
preserved as long a it is done using redirect/bounce processing as
opposed to forwarding regardless of the type of access (web or remote
mail readers)?
