Application Startup Behavior When Couchbase Node Is Unreachable Before Auto-Failover

Couchbase Server
1

Hi Everyone,

I have a question regarding application startup behavior when a Couchbase node becomes unavailable but the auto-failover threshold has not yet been reached.**

Cluster Configuration**

  • Couchbase cluster with:

    • 3 Data Nodes (DN)

    • 2 Index + Query Nodes (IQ)

  • Auto-failover threshold: 30 seconds (default)

Assume one of the Data Nodes becomes unavailable for some reason.

At the same time, an application attempts to start and connect to the Couchbase cluster.

Since the failure occurs within the auto-failover threshold window (30 seconds), the cluster has not yet triggered failover, and the failed node still appears in the cluster topology.

Known Behavior

  1. The application bootstraps using a static connection string.

  2. During bootstrap, the SDK retrieves the cluster map from one of the reachable nodes.

  3. The cluster map contains all 5 nodes in the cluster.

Questions

  1. Can the application start successfully in this situation?

  2. How does the Couchbase SDK behave when one node is unreachable but still present in the cluster map?

Important Condition

All the above events occur within the auto-failover threshold window, meaning:

  • The failed node has not yet been failed over.

  • The cluster topology still includes the failed node.**
    **