Re: python udf out of memory

Hi tom Flink will create individual classloader for each job in task manager. When jobs are frequently started and stopped, the usage of memory in Metaspace will increase. I found out that your OOM was caused by metaspace. I think you can check the size of metaspace and try to increase the size by

python udf out of memory

Hi, I am running a standalone cluster setup and submit flinksql job with python udf following the examples here github.com/ververica/flink-sql-cookbook/blob/main/udfs/01_python_udfs/01_python_udfs.md