Please support me on the below question for Opensearch CCR feature. If I have setup one leader cluster and one follower cluster and start replication from leader to follower index and it is working. but it’s possible to automatic stop replication when my leader cluster goes down or my leader index not found in leader cluster.
If leader index goes down(or not present), replication automatically moves to the paused state.
Thanks @ankikala for your support and quick responses !
Here what we are trying to achieve is the Term Automatic Failover In Open Search Cluster whenever CCR is running.
Let me introduce my issue for better understanding :- we have setup development cluster for open search which is deployed on two instance cluster A (Leader Cluster ) and Cluster B (Follower Cluster) , started CCR in between them ,At the current situation cluster B is only read only just copying the leader indexes data into it , but the situation at which leader cluster goes down or not available , we want our cluster B follower cluster to write indexes data automatically , changes to READWRITE but it is not happening and we need to do manual process for stopping the CCR and make it to read write.
I hope it is clear to you
Yep. Currently we only support active-passive replication where follower is only in read-only mode.
If you want to create a bi-directional replication then you can maintain 2 separate indices(let’s say index1 and index2) on leader and follower. Leader can ingest data to index1 which is replicated to the follower. Similarly, follower can ingest to index2 which gets replicated on leader.
We don’t have any documentation for this. But I’ve seen similar setup mentioned by elastic here: Bi-directional replication with Elasticsearch cross-cluster replication (CCR) | Elastic Blog