So, 2 of our indexers sometime have very high I/O due to a known issue, but this is causing index queueing on all our 31 other indexers in same cluster. When we turn off the 2 indexers that are going to have high I/O, we dont see any issue.
We are assuming that replication to these problem indexers are blocking other indexers and causing ripple effect across cluster. Is it expected that the cluster behaves this way?
Are there any configuration to optimize or create dedicated threads only for replication with out blocking indexing and searching?