Does anyone have any thoughts on the following?
I have a sendmail server running as a mail gateway to the internet. All incoming mail is forwarded to a Lotus Notes/Domino server from the sendmail server. The purpose of this server is spam filtering and Antivirus. I currently have real-time blackhole lists configured via sendmail to do spam filtering. MailScanner is installed to provide virus scanning of all email.
I would like to add Spamassassin to the mix to catch the few messages that do make it through the RBLs. Here's what I'm looking for.
1. I have a perl script that runs every 10 minutes to pull email addresses from the sendmail log that are outgoing messages and add them to SendMail's access.db file. It's a sort of auto-whitelisting feature that assumes if a user sends email to some address then it is a legit address and they want to receive email from it. I need to be able to either have spamassassin use this file or be able to add entries to both files.
2. As all my users are on Lotus Notes I need to be able to use the SQL option for storing user preferences so I can turn it on/off per user and have difference Bayesian databases per user.
3. I want to automatically whitelist all email going out from my company. Ideally, I'd like spamassassin not to check it at all.
4. What's the best way to deal with teaching the Bayes filters about incorrectly classified messages? Somebody somewhere mentioned setting up 2 mailboxes on the sendmail server, SPAM and NOTSPAM. Then have the user forward messages back to the appropriate mailbox when it is tagged incorrectly. Will this work with per user Bayes settings? Will this work with the message being forwarded through several different mail servers?
So now, can this be done? If so, how should I go about this.
Thanks in advance,
Bluecrack
I have a sendmail server running as a mail gateway to the internet. All incoming mail is forwarded to a Lotus Notes/Domino server from the sendmail server. The purpose of this server is spam filtering and Antivirus. I currently have real-time blackhole lists configured via sendmail to do spam filtering. MailScanner is installed to provide virus scanning of all email.
I would like to add Spamassassin to the mix to catch the few messages that do make it through the RBLs. Here's what I'm looking for.
1. I have a perl script that runs every 10 minutes to pull email addresses from the sendmail log that are outgoing messages and add them to SendMail's access.db file. It's a sort of auto-whitelisting feature that assumes if a user sends email to some address then it is a legit address and they want to receive email from it. I need to be able to either have spamassassin use this file or be able to add entries to both files.
2. As all my users are on Lotus Notes I need to be able to use the SQL option for storing user preferences so I can turn it on/off per user and have difference Bayesian databases per user.
3. I want to automatically whitelist all email going out from my company. Ideally, I'd like spamassassin not to check it at all.
4. What's the best way to deal with teaching the Bayes filters about incorrectly classified messages? Somebody somewhere mentioned setting up 2 mailboxes on the sendmail server, SPAM and NOTSPAM. Then have the user forward messages back to the appropriate mailbox when it is tagged incorrectly. Will this work with per user Bayes settings? Will this work with the message being forwarded through several different mail servers?
So now, can this be done? If so, how should I go about this.
Thanks in advance,
Bluecrack