Wietse Venema:
> Mueller, Martin (Messaging):
> > Thanks for all the work making memcache support available, any
> > thoughts on the potential use of this for anvil?
> 
> As documented, anvil blocks a client that overwhelms the server.
> For that, there is no need to maintain state over more than a few
> minutes. The program can be kept small and stupid, and it is not
> the end of the world when it loses state after "postfix reload".

Perhaps a better question would be "using memcache instead of anvil",
and using the same mechanism for other Postfix daemons to dump some
counters.  That would use a different interface than (lookup, update,
delete and first/next), more something like (increment, decrement,
or reset).

        Wietse

> Enforcing traffic quota requires longer-term memory. For this
> a better solution is to use policyd, postfwd and the likes. These
> can enforce limits on the number of messages, the total number of
> bytes sent, different limits for different users, and other stuff
> that anvil was not meant to do. These programs may benefit from
> using memcache for storage for several hours or so.
> 
>       Wietse
> 
> > Martin
> > 
> > -----Original Message-----
> > From: owner-postfix-us...@postfix.org 
> > [mailto:owner-postfix-us...@postfix.org] On Behalf Of Wietse Venema
> > Sent: Tuesday, December 13, 2011 8:08 AM
> > To: Wietse Venema
> > Cc: Postfix users
> > Subject: Re: memcache client for Postfix
> > 
> > > Combining memcache and proxymap
> > > -------------------------------
> > > 
> > > It seems that memcache is best for (surprise) doing what it was 
> > > designed for: a cache layer on top of a persistent database, where the 
> > > cache is maintained by the clients of the persistent database.
> > 
> > Completed in snapshot 20111213:
> > 
> > Support for a persistent backup database in the memcache client.
> > The memcache client updates the memcache whenever it looks up or modifies 
> > information in the persistent database.
> > 
> > The persistent database can be shared with the proxymap service, provided 
> > that it is listed as "proxy:maptype:mapname" in the proxy_read_maps or 
> > proxy_write_maps parameter value (depending on whether the access is 
> > read-only or read-write).
> > 
> > Support for proxymap-over-tcp (proxy:maptype:mapname@host:port) is under 
> > development. This would support postscreen/verify cache sharing, with or 
> > without memcache (memcache would help to avoid a performance bottleneck).
> > 
> > Elimination of dependencies on the libmemcache library. Postfix memcache 
> > support is now compiled in by default.
> > 
> >     Wietse
> > 
> 

Reply via email to