Skip to content

[SPARK-43398][CORE] Executor timeout should be max of idle shuffle and rdd timeout #41082

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
warrenzhu25 wants to merge 2 commits into from
Closed

[SPARK-43398][CORE] Executor timeout should be max of idle shuffle and rdd timeout #41082

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
b9d2ad8
Executor timeout should be max of shuffle and rdd timeout
warrenzhu25 May 7, 2023
8bed581
Trigger Build
warrenzhu25 Jun 9, 2023
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
15 changes: 6 additions & 9 deletions core/src/main/scala/org/apache/spark/scheduler/dynalloc/ExecutorMonitor.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -563,17 +563,14 @@ private[spark] class ExecutorMonitor(
def updateTimeout(): Unit = {
val oldDeadline = timeoutAt
val newDeadline = if (idleStart >= 0) {
val timeout = if (cachedBlocks.nonEmpty || (shuffleIds != null && shuffleIds.nonEmpty)) {
val _cacheTimeout = if (cachedBlocks.nonEmpty) storageTimeoutNs else Long.MaxValue
val _shuffleTimeout = if (shuffleIds != null && shuffleIds.nonEmpty) {
shuffleTimeoutNs
} else {
Long.MaxValue
}
math.min(_cacheTimeout, _shuffleTimeout)
val _cacheTimeout = if (cachedBlocks.nonEmpty) storageTimeoutNs else 0
val _shuffleTimeout = if (shuffleIds != null && shuffleIds.nonEmpty) {
shuffleTimeoutNs
} else {
idleTimeoutNs
0
}
// timeout should be max of idleTimeout, storageTimeout and shuffleTimeout
val timeout = Seq(_cacheTimeout, _shuffleTimeout, idleTimeoutNs).max
val deadline = idleStart + timeout
if (deadline >= 0) deadline else Long.MaxValue
} else {
Expand Down
28 changes: 28 additions & 0 deletions core/src/test/scala/org/apache/spark/scheduler/dynalloc/ExecutorMonitorSuite.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -476,6 +476,34 @@ class ExecutorMonitorSuite extends SparkFunSuite {
assert(monitor.executorCount == 0 )
}

for (isShuffleTrackingEnabled <- Seq(true, false)) {
test(s"SPARK-43398: executor timeout should be max of shuffle and rdd timeout with" +
s" shuffleTrackingEnabled as $isShuffleTrackingEnabled") {
conf
.set(DYN_ALLOCATION_SHUFFLE_TRACKING_TIMEOUT.key, "240s")
.set(DYN_ALLOCATION_SHUFFLE_TRACKING_ENABLED, isShuffleTrackingEnabled)
.set(SHUFFLE_SERVICE_ENABLED, false)
monitor = new ExecutorMonitor(conf, client, null, clock, allocationManagerSource())

monitor.onExecutorAdded(SparkListenerExecutorAdded(clock.getTimeMillis(), "1", execInfo))
knownExecs += "1"
val stage1 = stageInfo(1, shuffleId = 0)
monitor.onJobStart(SparkListenerJobStart(1, clock.getTimeMillis(), Seq(stage1)))
monitor.onBlockUpdated(rddUpdate(1, 0, "1"))
val t1 = taskInfo("1", 1)
monitor.onTaskStart(SparkListenerTaskStart(1, 1, t1))
monitor.onTaskEnd(SparkListenerTaskEnd(1, 1, "foo", Success, t1, new ExecutorMetrics, null))
monitor.onJobEnd(SparkListenerJobEnd(1, clock.getTimeMillis(), JobSucceeded))

if (isShuffleTrackingEnabled) {
assert(monitor.timedOutExecutors(storageDeadline).isEmpty)
assert(monitor.timedOutExecutors(shuffleDeadline) == Seq("1"))
} else {
assert(monitor.timedOutExecutors(idleDeadline).isEmpty)
assert(monitor.timedOutExecutors(storageDeadline) == Seq("1"))
}
}
}

private def idleDeadline: Long = clock.nanoTime() + idleTimeoutNs + 1
private def storageDeadline: Long = clock.nanoTime() + storageTimeoutNs + 1
Expand Down