The mailing list for listmasters using Sympa

[Miles Fidelman <address@concealed>]

> > >     We're looking to improve the availability of our Sympa system
> > >     here at Duke, to hopefully prevent the risk of having a single
> > >     point of failure as we do now, with one machine on one network in
> > >     one data center running the service.  It looks like running a
> > >     hot/cold failover system would be simple enough, with a load
> > >     balancer routing mail appropriately and each system being aware
> > >     of all the lists.  I don't know how well archives would work with
> > >     such a setup, though, unless we were doing some sort of shared
> > >     storage (e.g. NFS), which introduces its own set of problems.
What we do is run our mail system, including sympa, as a virtual image 
(xen); use DRBD to mirror the disk to a 2nd machine at the block level, 
and Pacemaker to do autofailover of the image.  Works like a charm.

The virtualization part isn't essential - for what you're doing, 
mirroring a disk partition with DRBD and using pacemaker for 
auto-failover would work just fine.

Miles Fidelman

--
In theory, there is no difference between theory and practice.
In<fnord>  practice, there is.   .... Yogi Berra