Re: Failover and circular replication

[Lee <lee_hoffman@brown.edu>]

We use multimaster with hardware loadbalancers that do load balancing 
on a read ip and failover on write ip (to preserve atomicity of writes).

Sincerely,
Lee

On Tuesday, August 19, 2003, at 12:00 AM, Beast wrote:

> Monday, August 18, 2003, 8:16:49 PM, Frank wrote:
>
> > Today at 1:12pm, Markus Schaber wrote:
>
> > > Is this possible by setting up multiple replication rules making 
> > > every
> > > server replicating to every other? Or will we create update loops 
> > > this
> > > way? (As I understand, the replication log contains timestamps, so it
> > > should be possible by slurpd/slapd to achieve this goal, but I don't
> > > know whether it is implemented in such a way.)
>
> > Markus,
> >   What you are looking for is called "multi-master" in OpenLDAP.  
> > There
> > has been quite a bit of discussion on the mailinglists about this.
> > Basically, it is possible, but it is full of danger.
>
> Hi,
> I'm looking for similar setup.
> So what is the best possible solution or workaround for these
> problems?
>
> --beast