If the active agent in a Secure Agent group goes down unexpectedly while an application ingestion and replication job or database ingestion and replication job is running, the job can automatically fail over to another available agent in the group.
Application Ingestion and Replication and Database Ingestion and Replication support automatic failovers for initial load jobs that have any supported source type. Database Ingestion and Replication also supports automatic failovers for incremental load and combined initial and incremental load jobs that have a Db2 for i, Db2 for z/OS, Oracle, or SQL Server source, subject to the following limitations:
•The job cannot have a Kafka target.
•The job cannot have persistent storage enabled.
•The job cannot use the Query-based CDC method to capture changes from an Oracle or SQL Server source.
Failover can be affected by how you set the supportedLoadTypes property of Database Ingestion agent service for the agents in the group. You can set the property separately for each agent in the group such that different agents run different load types. In this scenario, failover can occur only if another agent in the group is configured for the same load type as the failed agent or is configured for both load types (Initial and Incremental). For example, if you set the supportedLoadTypes property to Incremental for agent A, Initial for agent B, and Initial and Incremental for agent C, when agent A goes down, the job cannot fail over to agent B but it can fail over to agent C. Alternatively, if agent C goes down, the job might fail over to either agent A or Br, depending on internal load-type matching.
Also, failover is affected by agent capacity. If agent A fails, but agent C is already running with a full load, no failover occurs until capacity on agent C is available. For example, if a job that's running on agent C stops, enough capacity might become available for the failover from agent A to agent C to occur.
Another consideration is that all agents in a Secure Agent group are used for load balancing as well as failover. If multiple agents support the same load, jobs of that load type can be distributed across them. However, you can't specify which agent would be used for failover.