HELP: architecture with replication

Russ Allbery rra at
Mon Apr 10 08:08:14 UTC 2000

Long-delayed response....

Elena Samsonova <E.Samsonova at> writes:

> I'm trying to describe distributed architecture with replication even
> though I have no personal experience with it.  So please HELP!

> First, you replicate the whole thing, that is the active file, history and
> spool, right?

> In this case, how do you expire on the slaves?  Do you have them run
> their own expire or do you somehow replicate expired spool?  And if they
> do run their own expire, how do you make sure that their data is
> synchronized?

I've seen this done in one of two ways.  One way is to use Xref slaving,
in which case each reader has its own spool and runs its own expire and
the like.  Another way is to just NFS-export the active and history files
and the news spool and then just run nnrpd in daemon mode on all of the
reader machines.  (A problem with this approach, of course, is that the
history and active files are very frequently changing and NFS is
notoriously bad with this.)

Russ Allbery (rra at             <>

More information about the inn-workers mailing list