Skip to content

[SPARK-23637][YARN]Yarn might allocate more resource if a same executor is killed multiple times. #20781

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
jinxing64 wants to merge 3 commits into from
Closed

[SPARK-23637][YARN]Yarn might allocate more resource if a same executor is killed multiple times. #20781

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
bd6f8a1
[SPARK-23637][YARN]Yarn might allocate more resource if a same execut…
Mar 9, 2018
a177a63
it's possible that YarnAllocator process same completed container mul…
Mar 9, 2018
049ed49
Use runningExecutors as a set.
Mar 12, 2018
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
Next Next commit
[SPARK-23637][YARN]Yarn might allocate more resource if a same execut… 
…or is killed multiple times.
  • Loading branch information
jinxing committed Mar 9, 2018
commit bd6f8a1dd608d95f6faf2758cbd36e6328f48eb3
11 changes: 5 additions & 6 deletions resource-managers/yarn/src/main/scala/org/apache/spark/deploy/yarn/YarnAllocator.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -242,12 +242,11 @@ private[yarn] class YarnAllocator(
* Request that the ResourceManager release the container running the specified executor.
*/
def killExecutor(executorId: String): Unit = synchronized {
if (executorIdToContainer.contains(executorId)) {
val container = executorIdToContainer.get(executorId).get
internalReleaseContainer(container)
numExecutorsRunning.decrementAndGet()
} else {
logWarning(s"Attempted to kill unknown executor $executorId!")
executorIdToContainer.get(executorId) match {
case Some(container) if !releasedContainers.contains(container.getId) =>
internalReleaseContainer(container)
numExecutorsRunning.decrementAndGet()
case _ => logWarning(s"Attempted to kill unknown executor $executorId!")
}
}

Expand Down
24 changes: 24 additions & 0 deletions resource-managers/yarn/src/test/scala/org/apache/spark/deploy/yarn/YarnAllocatorSuite.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -256,6 +256,30 @@ class YarnAllocatorSuite extends SparkFunSuite with Matchers with BeforeAndAfter
handler.getPendingAllocate.size should be (1)
}

test("kill same executor multiple times") {
val handler = createAllocator(2)
handler.updateResourceRequests()
handler.getNumExecutorsRunning should be (0)
handler.getPendingAllocate.size should be (2)

val container1 = createContainer("host1")
val container2 = createContainer("host2")
handler.handleAllocatedContainers(Array(container1, container2))
handler.getNumExecutorsRunning should be (2)
handler.getPendingAllocate.size should be (0)

val executorToKill = handler.executorIdToContainer.keys.head
handler.killExecutor(executorToKill)
handler.getNumExecutorsRunning should be (1)
handler.killExecutor(executorToKill)
handler.killExecutor(executorToKill)
handler.killExecutor(executorToKill)
handler.getNumExecutorsRunning should be (1)
handler.requestTotalExecutorsWithPreferredLocalities(2, 0, Map.empty, Set.empty)
handler.updateResourceRequests()
handler.getPendingAllocate.size should be (1)
}

test("lost executor removed from backend") {
val handler = createAllocator(4)
handler.updateResourceRequests()
Expand Down