Hi Everyone, I am trying to run this query on External Iceberg Table, there are around 1.1 Billion event for the below query.

select count(*) from audit_events_log where date_trunc('day', occurredAt) = '2025-09-23';

I have 3 BE Node with 6CPU and 24GB Mem. Following properties are enabled:

SET GLOBAL query_timeout='3600';
SET GLOBAL enable_spill='true';

## BE Config

datacache_mem_size = 10GB
datacache_enable = 'true'

## Updated the default_wg

ALTER RESOURCE GROUP default_wg WITH ('mem_limit' = '50%', 'spill_mem_limit_threshold' = '0.2');

My BE PODs memory reached 100% and I get the error. ERROR 1064 (HY000): Memory of default_wg exceed limit. get chunk from scanner Backend: starrockscluster-be-1.starrockscluster-be-search.starrocks.svc.cluster.local, Limit: 9393093476. Mem usage has exceed the limit of query pool: file = <some-s3-file-path>. Since, I have enabled the spill, why the Query is not spilling over to the disk?

Have you set up

spill_local_storage_dir

under be.conf (or cn.conf)? Also, can you share your backends memory usage when the query is running here. For your cluster, it should be accessible via

<http://starrockscluster-be-1.starrockscluster-be-search.starrocks.svc.cluster.local:8040>