Github user rxin commented on a diff in the pull request:
https://github.com/apache/spark/pull/1587#discussion_r15390285
--- Diff:
sql/core/src/main/scala/org/apache/spark/sql/columnar/ColumnBuilder.scala ---
@@ -118,7 +118,7 @@ private[sql] class BinaryColumnBuilder extends
ComplexColumnBuilder(BINARY)
private[sql] class GenericColumnBuilder extends
ComplexColumnBuilder(GENERIC)
private[sql] object ColumnBuilder {
- val DEFAULT_INITIAL_BUFFER_SIZE = 10 * 1024 * 104
+ val DEFAULT_INITIAL_BUFFER_SIZE = 10 * 1024 * 1024
--- End diff --
While this is definitely a bug fix, it is also pretty dangerous. This means
a 100 column table would use 1G of buffer per task, and with 32 cores, that's
32G of ram just for the buffer. @marmbrus @liancheng
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
