Cross Cluster Replication - Disaster Recovery Failback


I'm currently working on a CCR architecture, to use a follower cluster as a Disaster Recovery.
I have a question with the failback:

  1. When production cluster is down, then I use the DR (follower) one
  2. I promote the follower as my main cluster, as users need to continue indexing data
  3. Then I need to resync the production one with the up-to-date DR cluster, how can I manage this?
  • Making my production cluster as follower, and once every data is sync'ed, promote to my main cluster?
  • Restore my last known backup, but how can I get documents which where indexed in the DR cluster it there is no Message Queue system in front of the cluster?
  • Any other idea?

Thanks for your feedback,


You will probably want to setup bi-directional CCR in this case - Bi-directional replication with Elasticsearch cross-cluster replication (CCR) | Elastic Blog

If you used specific index names in each cluster (eg indexA and indexB) and an alias pointing to the underlying index (eg index), you can use that to switch between which ever one you want to be the one the app writes to in each cluster.

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.