diff options
author | Marcelo Vanzin <vanzin@cloudera.com> | 2015-10-08 09:47:58 -0700 |
---|---|---|
committer | Marcelo Vanzin <vanzin@cloudera.com> | 2015-10-08 09:47:58 -0700 |
commit | 56a9692fc06077e31b37c00957e8011235f4e4eb (patch) | |
tree | ee7c0972e968f1e2fae401649c1304d099fdb8bd | |
parent | 2df882ef14d376f9e49380ff15a8a5e6997024a7 (diff) | |
download | spark-56a9692fc06077e31b37c00957e8011235f4e4eb.tar.gz spark-56a9692fc06077e31b37c00957e8011235f4e4eb.tar.bz2 spark-56a9692fc06077e31b37c00957e8011235f4e4eb.zip |
[SPARK-10987] [YARN] Workaround for missing netty rpc disconnection event.
In YARN client mode, when the AM connects to the driver, it may be the case
that the driver never needs to send a message back to the AM (i.e., no
dynamic allocation or preemption). This triggers an issue in the netty rpc
backend where no disconnection event is sent to endpoints, and the AM never
exits after the driver shuts down.
The real fix is too complicated, so this is a quick hack to unblock YARN
client mode until we can work on the real fix. It forces the driver to
send a message to the AM when the AM registers, thus establishing that
connection and enabling the disconnection event when the driver goes
away.
Also, a minor side issue: when the executor is shutting down, it needs
to send an "ack" back to the driver when using the netty rpc backend; but
that "ack" wasn't being sent because the handler was shutting down the rpc
env before returning. So added a change to delay the shutdown a little bit,
allowing the ack to be sent back.
Author: Marcelo Vanzin <vanzin@cloudera.com>
Closes #9021 from vanzin/SPARK-10987.
4 files changed, 17 insertions, 0 deletions
diff --git a/core/src/main/scala/org/apache/spark/executor/CoarseGrainedExecutorBackend.scala b/core/src/main/scala/org/apache/spark/executor/CoarseGrainedExecutorBackend.scala index fcd76ec527..49059de50b 100644 --- a/core/src/main/scala/org/apache/spark/executor/CoarseGrainedExecutorBackend.scala +++ b/core/src/main/scala/org/apache/spark/executor/CoarseGrainedExecutorBackend.scala @@ -110,6 +110,11 @@ private[spark] class CoarseGrainedExecutorBackend( case StopExecutor => logInfo("Driver commanded a shutdown") + // Cannot shutdown here because an ack may need to be sent back to the caller. So send + // a message to self to actually do the shutdown. + self.send(Shutdown) + + case Shutdown => executor.stop() stop() rpcEnv.shutdown() diff --git a/core/src/main/scala/org/apache/spark/scheduler/cluster/CoarseGrainedClusterMessage.scala b/core/src/main/scala/org/apache/spark/scheduler/cluster/CoarseGrainedClusterMessage.scala index d947436777..e0d25dc50c 100644 --- a/core/src/main/scala/org/apache/spark/scheduler/cluster/CoarseGrainedClusterMessage.scala +++ b/core/src/main/scala/org/apache/spark/scheduler/cluster/CoarseGrainedClusterMessage.scala @@ -100,4 +100,11 @@ private[spark] object CoarseGrainedClusterMessages { case class KillExecutors(executorIds: Seq[String]) extends CoarseGrainedClusterMessage + // Used internally by executors to shut themselves down. + case object Shutdown extends CoarseGrainedClusterMessage + + // SPARK-10987: workaround for netty RPC issue; forces a connection from the driver back + // to the AM. + case object DriverHello extends CoarseGrainedClusterMessage + } diff --git a/core/src/main/scala/org/apache/spark/scheduler/cluster/YarnSchedulerBackend.scala b/core/src/main/scala/org/apache/spark/scheduler/cluster/YarnSchedulerBackend.scala index e0107f9d3d..38218b9c08 100644 --- a/core/src/main/scala/org/apache/spark/scheduler/cluster/YarnSchedulerBackend.scala +++ b/core/src/main/scala/org/apache/spark/scheduler/cluster/YarnSchedulerBackend.scala @@ -170,6 +170,8 @@ private[spark] abstract class YarnSchedulerBackend( case RegisterClusterManager(am) => logInfo(s"ApplicationMaster registered as $am") amEndpoint = Option(am) + // See SPARK-10987. + am.send(DriverHello) case AddWebUIFilter(filterName, filterParams, proxyBase) => addWebUIFilter(filterName, filterParams, proxyBase) diff --git a/yarn/src/main/scala/org/apache/spark/deploy/yarn/ApplicationMaster.scala b/yarn/src/main/scala/org/apache/spark/deploy/yarn/ApplicationMaster.scala index a2ccdc05d7..3791eea5bf 100644 --- a/yarn/src/main/scala/org/apache/spark/deploy/yarn/ApplicationMaster.scala +++ b/yarn/src/main/scala/org/apache/spark/deploy/yarn/ApplicationMaster.scala @@ -564,6 +564,9 @@ private[spark] class ApplicationMaster( case x: AddWebUIFilter => logInfo(s"Add WebUI Filter. $x") driver.send(x) + + case DriverHello => + // SPARK-10987: no action needed for this message. } override def receiveAndReply(context: RpcCallContext): PartialFunction[Any, Unit] = { |