I think this is an interesting extension to the architecture.
We have heard requirements for a standby cluster, but frankly,
a cluster itself is a composite logical entity in that if you
have designed the cluster appropriately (w.r.t. throughput and
availability requirements) then you should not have a cluster
that has a problem as a whole.
You'll have one or more dysfunctional servers in a cluster and
the support crew needs to fix them. The availability will be
impacted, but not completely impaired. The load balancer will
take the affected servers out of group of responsive servers and
will direct the requests to the ones that are responsive.
Do you see a strong need of a standby "cluster"? This surely
makes the actual failover across clusters more complicated.
Regarding your other question -- sorry, 2 different DAS's can't
administer the same cluster. A cluster is contained in a given
administrative domain in GlassFish architecture.
Regards,
Kedar
Legolas wood wrote:
> Hi
> thank you for reading my post
> Is there any way to setup a Active passive schema for glassfish
> application server?
> imagine that we have a two cluster each has 3 instance of glassfish
> which each instance is running on a single machie.
> What do we need (hardware, software, module,...) to setup a Active/
> passive scenario with this two cluster?
>
> I mean
> when cluster A is running cluster B just stay up to date with latest
> session, transaction,.... Data.
> When Cluster A become problematic then Cluster B take the control of
> everything until support crew fix the cluster A.
>
>
> Some other questions :
>
> Can we have 2 DAS to administrate this two clusters and in the same time
> keep them synchronized?
>
> Thanks
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe_at_glassfish.dev.java.net
> For additional commands, e-mail: users-help_at_glassfish.dev.java.net
>