kris_h:
>
> Hey Ansgar,
>
> thank you for your quick reply.
>
> For sure the quasi-static tables should be managed via deploymant-systems or
> simplier rsync and that alike.
>
> We distribute the more dynamic tables - e.g. cidr-tables with self-harvested
> current spammer's IPs - actually by simply distributing those files with
> rsync.
>
> I hesitate to use distributed filesystems, because the multiple r/w-access
> results in big overhead/load.
> Also the 'permanent' inbound MTA-avaibility goes miles before applying
> rules/filtering.
>
> I searching for pros/cons for:
> 1. distributing new entries to all servers with special program/interfaces
> to named pipes or memcache
> 2. DRBD or GlusterFS
> 3. REDIS - which allows triggers
>
> Maybe Wietse has a redis-table-module in his drawer? ;-)
Sorry, I don't write code and then sadistically keep it for myself.
Wietse
