2CNL,

If you were to have Outlook's junk filter receive all of your emails rather than SpamFilter, you would see that much more than 6% of spam would slip thru. As SpamFilter will never be 100% accurate, some spam will go undetected. It is almost a certainty that another application can further stop some of this remaining spam.

The main issue here is that you have another SMTP server which is receiving and processing your incoming emails in addition to SpamFilter. SpamFilter *must* see the original IP of the sender to stop spam effectively. All of our most efficient filters require to see that IP in order to do their job and stop the spam. If your secondary server processes emails first, and then passes them on to SpamFilter, the only filters that can then check emails for spam are the Bayesian filter, the SURBL filter and your keyword (if you specified any). These filters will only stop a very small percentage of emails, and thus will not be able to noticeably stop spam being forwarded by your secondary SMTP server.

In regards to the numbers above, please do note that many connection attempts are just "probes" that don't result in emails to be sent. Furthermore, SpamFilter caches for a few minutes IPs that sent large amounts of spam in a certain timeframe, and further connection attempts from them are rejected without any emails being transferred. All these factors mean that the statistics are to be taken with a grain of salt, as the numbers will never add up, and in some cases there will be noticeable discrepancies.

StevenJohns,

There's two main issues. The first is ours, and is caused by how SpamFilter applies its filters. All the IP-based filters are checked before the email is actually received, and are thus applied super-fast. If we were to check the IP in the headers as well, we'd have to receive the email as well and then go back and re-apply the IP filters. That will involve quite a bit of work... but as I said that's an internal matter
The second issue is that we've always made it a point since SpamFilter v1.0 six years ago of *not* checking the headers, as they can always be faked. For example, if SpamFilter were to check the IP in the last header, a spammer could add a fake header listing gmail's IP at the top of the email, and send it thru a host not yet IP-blacklisted. If the email is determined to be spam by the other filters, we risk blocking gmail's IP as well. There would have to be a lot of confusing if/then logic to determine what IPs are then reported as spammers and which not. An option would be to only check the last received header if the email has been received by a specific IP (the secondary MX server)...
We're going to do some brainstorming to see what can be done, as this subject is appearing more and more often recently.

I never realy read the license agreement but is it legal to install SpamFilterISP enterprise on a second machine to be used as fall-back server or do i have to purchase a second license for a server that is online there in case something goes wrong ?

Robert,

Sounds like a good solution to me, PLEASE do NOT limit it to one IP though.....