Skip to content

[SPARK-11583][Scheduler][Core] Make MapStatus use less memory uage #9559

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
yaooqinn wants to merge 8 commits into from
Closed

[SPARK-11583][Scheduler][Core] Make MapStatus use less memory uage #9559

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
79353eb
map status smaller
Nov 7, 2015
492adeb
Let MapStatus be smaller in sparse/tense cases
yaooqinn Nov 9, 2015
cb4bce5
Let MapStatus be smaller in sparse/tense cases 1
yaooqinn Nov 9, 2015
d17e989
repair scalastyle check failures
yaooqinn Nov 9, 2015
e1d1106
simliar classes merged
yaooqinn Nov 9, 2015
8582522
use OpenHashSet instead of scala's HashSet
yaooqinn Nov 10, 2015
e6fb687
fix style
yaooqinn Nov 10, 2015
4fc89ad
fix ser/der bug
yaooqinn Nov 12, 2015
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Prev Previous commit
Next Next commit
repair scalastyle check failures
  • Loading branch information
@yaooqinn
yaooqinn committed Nov 9, 2015
commit d17e989d2d3f73e93984d9d28fe2d71fea6b5bb1
14 changes: 10 additions & 4 deletions core/src/main/scala/org/apache/spark/scheduler/MapStatus.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -172,8 +172,11 @@ private[spark] class MapStatusTrackingNoEmptyBlocks private (
}

private[spark] object MapStatusTrackingNoEmptyBlocks {
def apply(loc: BlockManagerId, numNonEmptyBlocks: Int , nonEmptyBlocks: mutable.HashSet[Int], avgSize: Long): MapStatusTrackingNoEmptyBlocks =
{
def apply(
loc: BlockManagerId,
numNonEmptyBlocks: Int,
nonEmptyBlocks: mutable.HashSet[Int],
avgSize: Long): MapStatusTrackingNoEmptyBlocks = {
new MapStatusTrackingNoEmptyBlocks(loc, numNonEmptyBlocks, nonEmptyBlocks, avgSize )
}
}
Expand Down Expand Up @@ -227,8 +230,11 @@ private[spark] class MapStatusTrackingEmptyBlocks private (
}

private[spark] object MapStatusTrackingEmptyBlocks {
def apply(loc: BlockManagerId, numNonEmptyBlocks: Int , emptyBlocksHashSet: mutable.HashSet[Int], avgSize: Long): MapStatusTrackingEmptyBlocks =
{
def apply(
loc: BlockManagerId,
numNonEmptyBlocks: Int ,
emptyBlocksHashSet: mutable.HashSet[Int],
avgSize: Long): MapStatusTrackingEmptyBlocks = {
new MapStatusTrackingEmptyBlocks(loc, numNonEmptyBlocks, emptyBlocksHashSet, avgSize )
}
}
Expand Down
11 changes: 7 additions & 4 deletions core/src/test/scala/org/apache/spark/scheduler/MapStatusSuite.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -66,7 +66,8 @@ class MapStatusSuite extends SparkFunSuite {
}
}

test("large tasks with dense non-empty blocks should use " + classOf[MapStatusTrackingEmptyBlocks].getName) {
test("large tasks with dense non-empty blocks should use" +
classOf[MapStatusTrackingEmptyBlocks].getName) {
val sizes = Array.fill[Long](2001)(150L)
val status = MapStatus(null, sizes)
assert(status.isInstanceOf[MapStatusTrackingEmptyBlocks])
Expand All @@ -76,7 +77,8 @@ class MapStatusSuite extends SparkFunSuite {
assert(status.getSizeForBlock(2000) === 150L)
}

test("large tasks with sparse non-empty blocks should use " + classOf[MapStatusTrackingNoEmptyBlocks].getName) {
test("large tasks with sparse non-empty blocks should use " +
classOf[MapStatusTrackingNoEmptyBlocks].getName) {
val sizes = Array.fill[Long](2001)(0L)
sizes(0) = 1L
val status = MapStatus(null, sizes)
Expand All @@ -87,10 +89,11 @@ class MapStatusSuite extends SparkFunSuite {
assert(status.getSizeForBlock(2000) === 0L)
}

test("large tasks with not dense or sparse non-empty blocks should use " + classOf[HighlyCompressedMapStatus].getName) {
test("large tasks with not dense or sparse non-empty blocks should use " +
classOf[HighlyCompressedMapStatus].getName) {
val sizes = Array.fill[Long](2001)(0L)
for(i <- 0 to sizes.length - 1){
if(i % 2 == 1){
if (i % 2 == 1) {
sizes(i) = 1L
}
}
Expand Down