The mailing list for listmasters using Sympa

[Serge Aumont <address@concealed>]

 Le 8/26/10 12:11 AM, Malcolm Waltz a écrit :
> Hi Warren,
>
> Here at University of the Pacific, we are running Sympa on Solaris 10
> on-top of VMWare ESX 4.0 with vMotion enabled.  Disk images are stored
> on a NetApp filer using FibreChannel, though we are migrating to NFS and
> iSCSI over 10GB Ethernet.  We have an off-site ESX host that can be used
> with vMotion, once we have another NetApp set up for replication.  As
> Miles Fidelman put it, this is a "brute force approach."  I wouldn't
> call it simple or cheap though.  It also doesn't protect you from an OS
> level fault or give you the ability to do maintenance without downtime.
>
> I would much prefer to do clustering at the application level.  I
> believe I remember seeing some discussion of this on the "sympa-beta"
> list during the development of Sympa 6.0.  I believe some of the
> features added in Sympa 6.0 were intended to bring it closer to being
> clusterable, though still a ways from achieving that goal.
You are true. We known that the infrastructure with high disponibility 
of storage, database and host is a solutioon for high disponibility of a 
service such as Sympa. How ever, we are working on a new internal 
architecture that allow clustering with load balancing at the 
application level. Current step is to move all spools to database (going 
to the last step of this document 
http://www.sympa.org/manual_6.0/server_architecture_for_big_services ).

The job will continue for list config etc.

Serge