----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/25795/ -----------------------------------------------------------
(Updated Sept. 30, 2014, 1:14 a.m.) Review request for drill and Steven Phillips. Changes ------- Added configuration params to enable/disable the fragment memory planning. Default is OFF. Also added a parameter to set an overcommit factor that sets the fragment memory limit to be higher than if it was divided evenly. Repository: drill-git Description ------- Provides memory planning at the fragment level. Applies a dynamic limit per fragment. Allows certain operators to bypass the fragment limit. Diffs (updated) ----- exec/java-exec/src/main/java/io/netty/buffer/FakeAllocator.java 9ebf3e7 exec/java-exec/src/main/java/org/apache/drill/exec/ExecConstants.java c210541 exec/java-exec/src/main/java/org/apache/drill/exec/memory/Accountor.java 18c5072 exec/java-exec/src/main/java/org/apache/drill/exec/memory/AtomicRemainder.java 6a87ab4 exec/java-exec/src/main/java/org/apache/drill/exec/memory/TopLevelAllocator.java 583b388 exec/java-exec/src/test/java/org/apache/drill/exec/memory/TestAllocators.java c23cc10 Diff: https://reviews.apache.org/r/25795/diff/ Testing ------- Added new unit test. Ran all unit tests. TPC-H distributed tests are particularly good tests for this change. Tested external sort for regression. Thanks, Parth Chandra
