I am trying to test various circuitbreakers to gain better understanding for tuning various breaker
limits
.. for example, for tuning indices.breaker.request.limit
on a cluster with 50% indexing and 50% querying load characteristics, what type of indexing operations can be simulated to trigger request circuitbreaker.
Often, during recovery of heavy shards or heavy scroll-based queries, I have always noticed parent circuitbreaker tripping. But have never observed individual breakers tripping.
I am using Elasticsearch 7.0.1 and heap allocation is around 14Gb per node and I use a 3 node cluster
Thanks in advance
- Dinesh