Details
-
Improvement
-
Resolution: Fixed
-
Critical
-
2.0, 2.0.1, 2.1.0
Description
When the whole cluster is restarted, we sometimes observe that the replication starts from the beginning, which would be very expensive. To avoid this, we should be able to identify the list of items that weren't replicated upon restarting the cluster, and replicate them only when the cluster is ready.