During the past few hours the master node of our cluster is operating under high load (~20, even spikes to 30-40). However, there has been no significant change in the amount of indexing/fetching requests recently. Even though the load is high, CPU cycles and GC seem normal. On the other hand, indexing/searching response times have been affected.
The cluster consists of two nodes and each index has 1 primary and 1 replica shard, thus I was expecting that both nodes would operate under roughly the same load.
Cluster ID: "82fc82"