How to avoid threadpool rejection

OpenSearch
1

Versions : Opensearch version-1.3.9

I am experiencing thread pool rejections in my OpenSearch production cluster, primarily in the search and write thread pool types. These rejections mostly occur during bulk indexing or heavy search activity across the cluster. To mitigate this issue, I found a suggestion online to increase the thread pool queue size above the default values. Is this a valid solution, or are there any other recommended approaches to prevent these rejections?

2

Hey @bshashan , I think you should first of all collect the operational stats of the cluster: increasing queues / resizing pools could be an option but would your cluster be able to handle even more work?

3

Hey @reta, thanks for the reply. Our cluster can handle the workload if we increase resources, and we are currently addressing this issue through vertical scaling. However, we’re looking for alternative solutions beyond scaling.