Today the list of processors of ingest pipelines is compared during cluster state updates in order to detect if the pipeline changed. In a cluster with a large number of ingest pipelines (1500+) with some of them having hundreds of processors (+300) parsing and comparing the list of processor can be inefficient and can slow down cluster state update applying.
Today the list of processors of ingest pipelines is compared during cluster state updates in order to detect if the pipeline changed. In a cluster with a large number of ingest pipelines (1500+) with some of them having hundreds of processors (+300) parsing and comparing the list of processor can be inefficient and can slow down cluster state update applying.
This shows up in Node Hot Threads like this: