StarRocks, a Linux Foundation project, is a next-generation sub-second MPP OLAP database for full analytics scenarios, including multi-dimensional analytics, real-time analytics, and ad-hoc queries. StarRocks received InfoWorld’s 2023 BOSSIE Award for best open source software.

StarRocks

During load testing I am seeing
```ScanOperator failed to offer io task due to thread pool overload: BE:60836```
what thread pool specifically can I increase to help this issue? My nodes are not using the full amount of cpu on them so I'd like to increase the number of threads available

pipeline_scan_thread_pool_queue_size
pipeline_scan_thread_pool_thread_num
pipeline_connector_scan_thread_num_per_cpu

a pstack can be obtained to know what are those thread busy with.

Screenshot 2026-03-04 at 11.50.33 PM.png

fantastic, ty. I'm running some very expensive iceberg queries so there will def be a limit here. but the defaults appear to be resulting in 40% cpu usage on my CN nodes and stacking tasks. I'll try those out

what is the difference between pipeline_scan_thread_pool_thread_num and pipeline_connector_scan_thread_num_per_cpu? Is one for internal tables and one for external tables like iceberg tables?

if you are running iceberg scanning, the connector one should be the thread pool.

non-connector one is for native internal table, connector one is for external catalog scanning and internal cloud native table scanning.