Can elasticsearch incorrectly show search threadpool rejected count

I have been load testing my elastic cluster by sending a large number of search requests. I see a lot of "rejected" when I do GET /_cat/thread_pool/search in kibana. However, I see that all the queries got successful responses. I didn't see any errors in the elastic logs either. Typically I see errors that mention search queue full and requests rejected etc in the Elasticsearch log. I see CPU spike to 100% on data nodes, but drop also rapidly. Is it possible elastic is incorrectly showing search requests as rejected?
I run Elasticsearch on EC2 instances launched in AWS. Use Elasticsearch version 7.11.2

Or could it be elastic is internally retrying it after marking it as "rejected" and hence the reason it is successful?

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.