I'm pretty excited about this one so I want to pre-annouce it. I'm currently working on rolling release/upgrade support in Exhibitor and Curator.

For Exhibitor:
    * Ensemble list, config values, etc. are centralized (via Amazon S3, a shared file, etc.)
    * If the ensemble list changes, Exhibitor will stage the change so that each instance in the ensemble
       applies the change one at a time, thus accomplishing a rolling release
    * New servers can be added to the ensemble this way without taking down the cluster

For Curator:
    * Curator can now be configured to poll the Exhibitor instances to get the list of ZK servers in the ensemble
    * Curator will use the updated ensemble list when creating connections.
    * If a Sys-Disconnected is received and there is a new ensemble list, Curator will close the current ZK
       handle and create a new one using the updated list.

Coming soon!

-Jordan

Woohoo! This is great news Jordan!

On 12.03.2012, at 18:32, "Jordan Zimmerman" <[EMAIL PROTECTED]> wrote:

> I'm pretty excited about this one so I want to pre-annouce it. I'm currently working on rolling release/upgrade support in Exhibitor and Curator.
>
> For Exhibitor:
>    * Ensemble list, config values, etc. are centralized (via Amazon S3, a shared file, etc.)
>    * If the ensemble list changes, Exhibitor will stage the change so that each instance in the ensemble
>       applies the change one at a time, thus accomplishing a rolling release
>    * New servers can be added to the ensemble this way without taking down the cluster
>
> For Curator:
>    * Curator can now be configured to poll the Exhibitor instances to get the list of ZK servers in the ensemble
>    * Curator will use the updated ensemble list when creating connections.
>    * If a Sys-Disconnected is received and there is a new ensemble list, Curator will close the current ZK
>       handle and create a new one using the updated list.
>
> Coming soon!
>
> -Jordan