diff options
author | Andrew Or <andrew@databricks.com> | 2015-08-03 14:22:07 -0700 |
---|---|---|
committer | Josh Rosen <joshrosen@databricks.com> | 2015-08-03 14:22:07 -0700 |
commit | 702aa9d7fb16c98a50e046edfd76b8a7861d0391 (patch) | |
tree | 42f3a10ebd1086fda3d7b185b73cb452623964ca /sql/catalyst | |
parent | e4765a46833baff1dd7465c4cf50e947de7e8f21 (diff) | |
download | spark-702aa9d7fb16c98a50e046edfd76b8a7861d0391.tar.gz spark-702aa9d7fb16c98a50e046edfd76b8a7861d0391.tar.bz2 spark-702aa9d7fb16c98a50e046edfd76b8a7861d0391.zip |
[SPARK-8735] [SQL] Expose memory usage for shuffles, joins and aggregations
This patch exposes the memory used by internal data structures on the SparkUI. This tracks memory used by all spilling operations and SQL operators backed by Tungsten, e.g. `BroadcastHashJoin`, `ExternalSort`, `GeneratedAggregate` etc. The metric exposed is "peak execution memory", which broadly refers to the peak in-memory sizes of each of these data structure.
A separate patch will extend this by linking the new information to the SQL operators themselves.
<img width="950" alt="screen shot 2015-07-29 at 7 43 17 pm" src="https://cloud.githubusercontent.com/assets/2133137/8974776/b90fc980-362a-11e5-9e2b-842da75b1641.png">
<img width="802" alt="screen shot 2015-07-29 at 7 43 05 pm" src="https://cloud.githubusercontent.com/assets/2133137/8974777/baa76492-362a-11e5-9b77-e364a6a6b64e.png">
<!-- Reviewable:start -->
[<img src="https://reviewable.io/review_button.png" height=40 alt="Review on Reviewable"/>](https://reviewable.io/reviews/apache/spark/7770)
<!-- Reviewable:end -->
Author: Andrew Or <andrew@databricks.com>
Closes #7770 from andrewor14/expose-memory-metrics and squashes the following commits:
9abecb9 [Andrew Or] Merge branch 'master' of github.com:apache/spark into expose-memory-metrics
f5b0d68 [Andrew Or] Merge branch 'master' of github.com:apache/spark into expose-memory-metrics
d7df332 [Andrew Or] Merge branch 'master' of github.com:apache/spark into expose-memory-metrics
8eefbc5 [Andrew Or] Fix non-failing tests
9de2a12 [Andrew Or] Fix tests due to another logical merge conflict
876bfa4 [Andrew Or] Fix failing test after logical merge conflict
361a359 [Andrew Or] Merge branch 'master' of github.com:apache/spark into expose-memory-metrics
40b4802 [Andrew Or] Fix style?
d0fef87 [Andrew Or] Fix tests?
b3b92f6 [Andrew Or] Address comments
0625d73 [Andrew Or] Merge branch 'master' of github.com:apache/spark into expose-memory-metrics
c00a197 [Andrew Or] Fix potential NPEs
10da1cd [Andrew Or] Fix compile
17f4c2d [Andrew Or] Fix compile?
a87b4d0 [Andrew Or] Fix compile?
d70874d [Andrew Or] Fix test compile + address comments
2840b7d [Andrew Or] Merge branch 'master' of github.com:apache/spark into expose-memory-metrics
6aa2f7a [Andrew Or] Merge branch 'master' of github.com:apache/spark into expose-memory-metrics
b889a68 [Andrew Or] Minor changes: comments, spacing, style
663a303 [Andrew Or] UnsafeShuffleWriter: update peak memory before close
d090a94 [Andrew Or] Fix style
2480d84 [Andrew Or] Expand test coverage
5f1235b [Andrew Or] Merge branch 'master' of github.com:apache/spark into expose-memory-metrics
1ecf678 [Andrew Or] Minor changes: comments, style, unused imports
0b6926c [Andrew Or] Oops
111a05e [Andrew Or] Merge branch 'master' of github.com:apache/spark into expose-memory-metrics
a7a39a5 [Andrew Or] Strengthen presence check for accumulator
a919eb7 [Andrew Or] Add tests for unsafe shuffle writer
23c845d [Andrew Or] Add tests for SQL operators
a757550 [Andrew Or] Address comments
b5c51c1 [Andrew Or] Re-enable test in JavaAPISuite
5107691 [Andrew Or] Add tests for internal accumulators
59231e4 [Andrew Or] Fix tests
9528d09 [Andrew Or] Merge branch 'master' of github.com:apache/spark into expose-memory-metrics
5b5e6f3 [Andrew Or] Add peak execution memory to summary table + tooltip
92b4b6b [Andrew Or] Display peak execution memory on the UI
eee5437 [Andrew Or] Merge branch 'master' of github.com:apache/spark into expose-memory-metrics
d9b9015 [Andrew Or] Track execution memory in unsafe shuffles
770ee54 [Andrew Or] Track execution memory in broadcast joins
9c605a4 [Andrew Or] Track execution memory in GeneratedAggregate
9e824f2 [Andrew Or] Add back execution memory tracking for *ExternalSort
4ef4cb1 [Andrew Or] Merge branch 'master' of github.com:apache/spark into expose-memory-metrics
e6c3e2f [Andrew Or] Move internal accumulators creation to Stage
a417592 [Andrew Or] Expose memory metrics in UnsafeExternalSorter
3c4f042 [Andrew Or] Track memory usage in ExternalAppendOnlyMap / ExternalSorter
bd7ab3f [Andrew Or] Add internal accumulators to TaskContext
Diffstat (limited to 'sql/catalyst')
-rw-r--r-- | sql/catalyst/src/main/java/org/apache/spark/sql/execution/UnsafeExternalRowSorter.java | 7 |
1 files changed, 7 insertions, 0 deletions
diff --git a/sql/catalyst/src/main/java/org/apache/spark/sql/execution/UnsafeExternalRowSorter.java b/sql/catalyst/src/main/java/org/apache/spark/sql/execution/UnsafeExternalRowSorter.java index 5e4c6232c9..193906d247 100644 --- a/sql/catalyst/src/main/java/org/apache/spark/sql/execution/UnsafeExternalRowSorter.java +++ b/sql/catalyst/src/main/java/org/apache/spark/sql/execution/UnsafeExternalRowSorter.java @@ -106,6 +106,13 @@ final class UnsafeExternalRowSorter { sorter.spill(); } + /** + * Return the peak memory used so far, in bytes. + */ + public long getPeakMemoryUsage() { + return sorter.getPeakMemoryUsedBytes(); + } + private void cleanupResources() { sorter.freeMemory(); } |