How to configure DC–DR with OpenSearch CCR and Failback Behavior

OpenSearch
1

Versions:

  • OpenSearch: >= 2.8.0

Describe the issue:

We are designing a DC–DR architecture using OpenSearch Cross-Cluster Replication (CCR) with a leader → follower (active–passive) model.

The initial setup works as expected:

  • DC is configured as the leader

  • DR is configured as the follower

  • CCR successfully replicates data from DC to DR

However, the issue arises after a DR failover:

  1. The leader (DC) becomes unavailable

  2. The DR follower is promoted by stopping replication (stop replication / unfollow) so it can accept writes

  3. Applications start writing to DR (now acting as primary)

  4. Later, the original DC comes back online

At this point, it is unclear how the original DC can rejoin the topology and follow the new primary (DR).

Based on our understanding:

  • Once the follower is promoted and starts accepting writes, CCR does not support syncing those changes back to the old leader

  • Re-establishing CCR typically requires deleting the existing indices on the old DC and replicating all data from scratch, which is expensive for large datasets

Main question:
Is there any supported or recommended way for the original DC to automatically follow the new primary (DR) after failover without rebuilding the entire dataset?