Failover is automatic for applications with recovery enabled so long as they are deployed to a deployment group with two or more servers. Continued operation after failover is subject to the limitations discussed in Recovering applications.
When an application is deployed ON ALL (see DEPLOY APPLICATION), a failover copy of each event is stored on a different server from the one processing it. If one server goes down, Striim will automatically reallocate the application among the remaining servers. When the server comes back up, the application will automatically be reallocated as per its original deployment. If two servers in the deployment group go down, the application will crash.
When an application is deployed ON ONE (see DEPLOY APPLICATION), a failover copy of each event is stored on a different server. If the server running the application goes down, Striim will automatically deploy the application on the other server. The application will continue running on the other server even after the original server comes back up.
If a Forwarding Agent's output is partitioned (see Using the Striim Forwarding Agent) and one of several servers receiving its output goes down, the output it was receiving will be redistributed to the remaining servers while it is offline.
The web UI runs on all servers, so if the server where users normally access the UI goes down, they can manually change the URL to use another server. Alternatively, you may implement standard Linux load-balancing across multiple servers so that in the event one fails its web UI users will automatically be switched to another.
If the server hosting the Derby metadata repository goes down, the cluster will crash. See Recovering applications for further discussion.