Spam Filter

Thinking Farm's mail servers provide integrated anti-spam and anti-virus filtering mechanisms. While virus detection tends to be quite straight-forward, identifying spam emails is a whole lot more difficult. Different people may have slightly different definitions of spam, or different tolerance of spam-like emails. Furthermore, they may have different preference about what to do with emails that have been identified as spam.

To this end, we provide users the flexibility to fine-tune their spam filter configuration. This can be done on a per-user level (i.e. for each mailbox account), or across the entire virtual host. We provide a global default configuration that works best in most cases, so you can choose not to do any fine-tuning and still minimize spam coming into your mailbox accounts.

All emails sent to your mailbox are automatically analyzed by our mail servers, even if you've configured your mailbox to forward all your emails elsewhere. A spam score is computed and assigned to the emails. Emails that are more likely to be spam are given higher spam scores. You can fine-tune how this scoring is done, and you can also decide what to do with the emails based on the scores.

Your per-user spam filter configuration is found under the Spam Filter tab in your Account Settings.

Configuration

required_score

The required_score is a number that specifies the spam score required before an email is considered spam. This number should be in the range 0.0 to 9999.9 (one place of decimal).

When the spam score of an email reaches or exceeds required_score, our mail servers will consider the email as spam and reject it. Rejecting an email is different from silently deleting or dropping emails (refer to the section "Rejected Email vs Deleted Email" below).

If not specified, the default required_score is 9999.9. This means that no emails are ever considered spam and hence are never rejected. (Spam scores are always calculated and always recorded in the email header.)

Note that emails identified to be virus infected are always rejected rightaway. This is not a configurable setting.

whitelist_from <pattern> ...

This is a list of sender's email addresses or email address patterns that you want to added to your whitelist. This means that if an email is sent from one of these email addresses or an email address matching one of these patterns, the spam score for the email will be reduced somewhat (-100.0) so that it is much less likely to be considered spam.

The email addresses or email address patterns must be specified as a space separated list. Patterns are specified in file-glob style, using the * character (to match any one or more characters) and ? character (to match one single character).

You can use this to specify email addresses which send email that is often tagged incorrectly as spam.

unwhitelist_from <pattern> ...

This is used to remove email addresses or email address patterns from the whitelist. You can use this to specify exceptions to your own whitelist, or the default whitelist specified across the virtual host.

The email address and email address patterns are specified in the same manner as whitelist_from explained above.

blacklist_from <pattern> ...

This is a list of sender's email addresses or email address patterns that you want blacklisted. This means that if an email is sent from one of these email addresses or an email address matching one of these patterns, the spam score of the email will be increased somewhat (+100.0) so that it is much more likely to be considered spam.

The email address and email address patterns are specified in the same manner as whitelist_from explained above.

You can use this to specify email address which send email that is often not caught as spam.

unblacklist_from <pattern> ...

This is used to remove email addresses or email address patterns from the blacklist. You can use this to specify exceptions to your own blacklist, or the default blacklist specified across the virtual host.

The email address and email address patterns are specified in the same manner as whitelist_from explained above.

whitelist_to <pattern> ...

This is similar in principle to whitelist_from, but matches against the email recipient.

The email address and email address patterns are specified in the same manner as whitelist_from explained above.

Rejected Email vs Deleted Email

When Thinking Farm's mail servers receive an email from another mail server, it immediately performs anti-virus scanning and spam analysis online during the mail delivery transaction itself. If a decision is reached to reject the email (e.g. due to detected virus or spam), that mail delivery transaction is made to fail, so that the other mail server will see that the delivery attempt has failed.

The advantages of immediate rejection during delivery are:

• There is a delivery failure, so someone (the sender in this case) will receive a proper error message. The danger with silently deleting emails is that no one can easily know about the problem. This is important when emails are are wrongly identified as spam and rejected.
• The other mail server is responsible for reporting the error to the sender (because its delivery attempt to our mail server failed). If you think about the many virus emails today that forge the sender's email address, this will free us (and you) from having to deal with complaints that our error messages are sent wrongly.