My spam-trap script also sweeps the logs for misbehavior from connecting
hosts - non-SMTP commands, illegal pipelining, etc.

Two "misbehaves" earn an entry in postscreen_blacklist;   and (when I
get around to it)  two BLACKLISTED refusals will generate an IPtables
entry.    The blacklist entries die after a couple of days or so.

Allen C

On 11/02/16 05:25, Noel Jones wrote:
> On 2/10/2016 3:41 PM, Allen Coates wrote:
>> I have very similar problems.
>>
>> I was however, thinking along the lines of a command-line executable (or
>> script), specifically  to rescind a temporary white-list entry.   I am
>> not very good at the "big picture"  :-)
>>
>> Allen C
>>
>> On 10/02/16 17:35, Mike Coddington wrote:
>>> I had a problem with an IP address sneaking into Postscreen's whitelist. I 
>>> added the IP address to postscreen_access.cidr and set it to REJECT, as I 
>>> typically do with problem IP ranges. It seemed as though the temporary 
>>> whitelist was overriding my request to reject the mail though. I ended up 
>>> removing postscreen_cache.db and restarting Postfix. Is there a way to go 
>>> into that database and manually remove specific entries? Most of them are 
>>> fine, which is why I was sad to have to do the brute force technique that I 
>>> used.
>>>
>>> I’m guessing it’s not possible but figured I’d ask.
>
> Since it's fairly disruptive to alter the postscreen cache db -- it
> can only be done with postfix stopped -- a better and more practical
> solution is to add the offending clients to a regular old indexed
> check_client_access table.
>
> Changes to an indexed tables (hash, cdb, btree, *sql) are picked up
> automatically by smtpd with minimal disruption to mail flow, and
> unless the offending client is making dozens (hundreds?) of
> connections per minute the overall performance difference is
> negligible.  In the case of a flood of connections, a firewall block
> is probably a better solution anyway.
>
> So the bottom line is that although it is possible to remove a
> client from the postscreen automatic whitelist cache, it's not worth
> the trouble.
>
>
>
>   -- Noel Jones

Reply via email to