[ https://issues.apache.org/jira/browse/SPARK-8890?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14628533#comment-14628533 ]
Ilya Ganelin commented on SPARK-8890: ------------------------------------- Once data is sorted, is the number of partitions guaranteed to be under that limit? When we're talking about sorting, are we talking about which columns are in which partition? I want to make sure I understand what is happening. When we ingest a data frame, we consume a set of data organized by columns (the schema). When this data is partitioned, does all data under a certain column go to the same partition? If not, what happens in this stage? We create a new ```outputWriter``` for each row based on the columns within that row (from the projected columns). New ```outputWriters``` become necessary when the columns within a row are different. However, given that the schema is fixed, where does this variability come from and what does it mean to "sort" in this context? > Reduce memory consumption for dynamic partition insert > ------------------------------------------------------ > > Key: SPARK-8890 > URL: https://issues.apache.org/jira/browse/SPARK-8890 > Project: Spark > Issue Type: Sub-task > Components: SQL > Reporter: Reynold Xin > Priority: Critical > > Currently, InsertIntoHadoopFsRelation can run out of memory if the number of > table partitions is large. The problem is that we open one output writer for > each partition, and when data are randomized and when the number of > partitions is large, we open a large number of output writers, leading to OOM. > The solution here is to inject a sorting operation once the number of active > partitions is beyond a certain point (e.g. 50?) -- This message was sent by Atlassian JIRA (v6.3.4#6332) --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org For additional commands, e-mail: issues-h...@spark.apache.org