[Greylist-users] Why and how often greylisting fails against ordinal servers?

[TSS Support]

>Yes, I know the need of whitelisting. But losing mails from mailing
>list servers is acceptable because mailing list service usually
>provides an archive also.
>
>The problem is that I may be in danger of losing my customers' mail.
>This is unacceptable because I can't get it back anyway. And my
>customers are open so I can't predict the server from which their
>mails come, that is to say, whitelisting is useless in this case.
>
>I hope RFC non-compliant servers are very rare, otherwise I should
>consider modifying greylisting method, like "receive body for safe
>and temporary reject".

Well, there are probably bigger mail sites than me who can chime in, but our
experience with customers asking for whitelisting suggests that overall,
less than one-tenth of a percent of mailservers are so broken that the
customer doesn't at least get a bounce back. If you are concerned about
that, consider setting the expiration on the record to 25 hours. You will
see a little more spam get through due to repeat spam runs, but also get
servers that only retry every 24 hours, and senders who retry the next day.
Other than that, you've described exactly the problem with greylisting.

The real thorn in our side so far seems to be hideously misconfigured
Exchange servers (don't know how people get them that messed up) and
Groupwise servers (they report a 4xx error back as a 500 to the client).
Other people have probably found different cases that bother them the most.


On a complete tangent, my greylisting daemon seems to die randomly. At
6:25am. Not every morning, just some mornings. The cron job for
db_maintinence runs at midnight, so it doesn't seem related to that.
However, that's when all of my logs are rotated. Has anyone seen relaydelay
die when the logs are rotated? Is there a fix in CVS? I'm running 0.04.
Thanks.

-Keith, TSS