Active Failover Limitations

The Active Failover setup in ArangoDB has a few limitations.

  • In contrast to the ArangoDB Cluster:
    • Active Failover has only asynchronous replication, and hence no guarantee on how many database operations may have been lost during a failover.
    • Active Failover has no global state, and hence a failover to a bad follower (see the example above) overrides all other followers with that state (including the previous leader, which might have more up-to-date data). In a Cluster setup, a global state is provided by the agency and hence ArangoDB is aware of the latest state.
  • Should you add more than one follower, be aware that during a failover situation the failover attempts to pick the most up-to-date follower as the new leader on a best-effort basis.
  • Should you be using the ArangoDB Starter or the Kubernetes Operator to manage your Active-Failover deployment, be aware that upgrading might trigger an unintentional failover between machines.