Github user passionke commented on the issue: https://github.com/apache/spark/pull/20712 @mgaido91 @hvanhovell @kiszk Thank's your attention, My application is a long term spark sql engine server, hundreds of spark sql run on it in spark local model. And these sql will not change until biz logic update. The sql depends on some tables which will be update before compute. So, In this special case, if I can make a larger CodeGenerator.cache size, I can reuse GeneratedClass instance, and In my view it will reduce meta space increase and decrease the full gc times
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org