it caused by your batch size too small, which lead to a lot of ROS Containers created and reach the limitation(1024 default). you should do defragment using TupleMover task(mergeout) before the error raised.
To do troubleshooting:
select * from STORAGE_CONTAINERS where projection_name like '%DATASET_TABLE%';
SELECT *
FROM CONFIGURATION_PARAMETERS
WHERE parameter_name = 'ContainersPerProjectionLimit' ;
select count(*) from STORAGE_CONTAINERS where projection_name like '%DATASET_TABLE%';
-- change ContainersPerProjectionLimit settings
SELECT SET_CONFIG_PARAMETER('ContainersPerProjectionLimit', 2048);
or
-- change mergeout frequency
SELECT SET_CONFIG_PARAMETER('MergeOutInterval', 30);
or
-- Do TupleMover Task manually
select do_tm_task('mergeout','projection_name')