I have a cluster with coordinating nodes that are relatively much smaller than data nodes - 4Gi vs 256Gi memory, respectively. I am starting to see indexing pressure rejections from the coordinating stage, only on the coordinating nodes. All of these coordinating nodes are using <50% heap. I would like to increase indexing_pressure.memory.limit
but only on these coordinating nodes as they have plenty of heap available. This does not seem possible as the setting is cluster-wide. Is there a workaround? I'd rather not scale up the coordinating nodes while there is still room in the heap.
Hi @andy-w00d :
indexing_pressure.memory.limit
is a node setting (docs). You should be able to specify this setting in your elasticsearch configuration.
Hope that helps!