Milter-greylist

I realize that this isn't exactly what you were thinking when talking 
about using SQL, but we have all of the whitelist/blacklist rules dumped 
from SQL on a regular basis onto the tail end of the conf.  (everytime a 
rule is changed...)  So in our case, we would win instant changes rather 
than having end-users waiting 10 mins for their changes to take effect.

More directly related to where you were going with the memory/db dumps, 
being able to dump/sync/update ram from a central db would be a solution 
for the fact that even with the sync feature, the servers seem to get 
out of wack with regard to each other.  Think 6 servers, acting as one, 
where periodically a server is taken down for a bit for maintainence, 
etc.  If it could populate the tables in RAM from SQL on startup, then 
sync changes with the SQL server at given intervals, it could be pretty 
handy.

Just throwing it out there.  If I actually thought it was worth the time 
to code it, I would have done it already.  ;-)

Cheers,

~Ethan B.



Emmanuel Dreyfus wrote:

> On Thu, May 19, 2005 at 10:29:16AM -0400, Steven W. Orr wrote:
> 
>>I am about to start using milter-greylist and I'm already using 
>>spamassassin with MySQL for the bayes and auto-whitelist and preferences. 
>>So I was wondering if anyone had this in the works.
>>
>>Also, what sort of database does milter-greylist currently use?
> 
> 
> It does not use a database, it stores the state in memory and regularly
> dumps it do disk to resume operation after a crash.
> 
> There is no plan to use an external database (what do you win?).
>