We are trying to use BulkProcessor (Async) for indexing, the operation has millions of documents.
On the application client side we are using High Level REST Client and ThreadPool. We are able to generate a high amount of load, for eg lets assume 30threads in bulkActionSize=500.
On ES side, lets assume i have only one master node with 5shards and 0 replicas.
Machine is a Octa-Core machine
- While setting thread pool we can set the executor. We have set it to 'Names.BULK'. ES documentation says, by default this type of operation will use FixedExecutorBuilder. This operation will use fixed number of threads on ES side.
What will happen if we do the bulk indexing using Names.GENERIC which has ScalingExecutorBuilder, to achieve scaling ?? This can give us more threads to index.