wankunde commented on a change in pull request #34234:
URL: https://github.com/apache/spark/pull/34234#discussion_r739662025
##########
File path: core/src/test/scala/org/apache/spark/scheduler/MapStatusSuite.scala
##########
@@ -191,4 +191,61 @@ class MapStatusSuite extends SparkFunSuite {
assert(count === 3000)
}
}
+
+ def compressAndDecompressSize(size: Long): Long = {
+ MapStatus.decompressSize(MapStatus.compressSize(size))
+ }
+
+ test("SPARK-36967: HighlyCompressedMapStatus should record accurately the
size " +
+ "of skewed shuffle blocks") {
+ val conf = new
SparkConf().set(config.SHUFFLE_ACCURATE_BLOCK_SKEWED_FACTOR.key, "5")
+ val env = mock(classOf[SparkEnv])
+ doReturn(conf).when(env).conf
+ SparkEnv.set(env)
+
+ val sizes = Array.tabulate[Long](3000)(i => (if (i < 2990) i else i + 350
* 1024).toLong)
+ val avg = sizes.filter(_ < 3000).sum / sizes.count(i => i > 0 && i < 3000)
+ val loc = BlockManagerId("a", "b", 10)
+ val mapTaskAttemptId = 5
+ val status = MapStatus(loc, sizes, mapTaskAttemptId)
+ val status1 = compressAndDecompressMapStatus(status)
+ assert(status1.isInstanceOf[HighlyCompressedMapStatus])
+ assert(status1.location == loc)
+ assert(status1.mapId == mapTaskAttemptId)
+ assert(status1.getSizeForBlock(0) == 0)
+ for (i <- 1 until 3000) {
+ val estimate = status1.getSizeForBlock(i)
+ if (i < 2990) {
+ assert(estimate === avg)
+ } else {
+ assert(estimate === compressAndDecompressSize(sizes(i)))
+ }
Review comment:
Updated
##########
File path: core/src/test/scala/org/apache/spark/scheduler/MapStatusSuite.scala
##########
@@ -191,4 +191,61 @@ class MapStatusSuite extends SparkFunSuite {
assert(count === 3000)
}
}
+
+ def compressAndDecompressSize(size: Long): Long = {
+ MapStatus.decompressSize(MapStatus.compressSize(size))
+ }
+
+ test("SPARK-36967: HighlyCompressedMapStatus should record accurately the
size " +
+ "of skewed shuffle blocks") {
+ val conf = new
SparkConf().set(config.SHUFFLE_ACCURATE_BLOCK_SKEWED_FACTOR.key, "5")
+ val env = mock(classOf[SparkEnv])
+ doReturn(conf).when(env).conf
+ SparkEnv.set(env)
+
+ val sizes = Array.tabulate[Long](3000)(i => (if (i < 2990) i else i + 350
* 1024).toLong)
+ val avg = sizes.filter(_ < 3000).sum / sizes.count(i => i > 0 && i < 3000)
+ val loc = BlockManagerId("a", "b", 10)
+ val mapTaskAttemptId = 5
+ val status = MapStatus(loc, sizes, mapTaskAttemptId)
+ val status1 = compressAndDecompressMapStatus(status)
+ assert(status1.isInstanceOf[HighlyCompressedMapStatus])
+ assert(status1.location == loc)
+ assert(status1.mapId == mapTaskAttemptId)
+ assert(status1.getSizeForBlock(0) == 0)
+ for (i <- 1 until 3000) {
+ val estimate = status1.getSizeForBlock(i)
+ if (i < 2990) {
+ assert(estimate === avg)
+ } else {
+ assert(estimate === compressAndDecompressSize(sizes(i)))
+ }
+ }
+ }
+
+ test("SPARK-36967: Limit accurated skewed block number if too many blocks
are skewed") {
+ val conf = new
SparkConf().set(config.SHUFFLE_ACCURATE_BLOCK_SKEWED_FACTOR.key, "5")
+ val env = mock(classOf[SparkEnv])
+ doReturn(conf).when(env).conf
+ SparkEnv.set(env)
+
+ val sizes = Array.tabulate[Long](3000)(i => (if (i < 2500) i else i + 3500
* 1024).toLong)
Review comment:
Updated
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
