Spark ALS.train()迭代次数过多导致java.lang.StackOverflowError

Xiangrui Meng added a comment – 07/Oct/14 01:46

This is a known issue. We can fix it by checkpointing intermediate RDDs. For now, please use a smaller number of iterations (<30). Usually, ALS converges within that number of iterations.

https://issues.apache.org/jira/browse/SPARK-3370