Skip to content

[SPARK-12101][Core]Fix thread pools that cannot cache tasks in Worker and AppClient #10108

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
wants to merge 2 commits into from
Closed

[SPARK-12101][Core]Fix thread pools that cannot cache tasks in Worker and AppClient #10108

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
40b3d3f
Fix thread pools that cannot cache tasks in Worker and AppClient
zsxwing Dec 2, 2015
2b9f4da
Remove unused imports
zsxwing Dec 2, 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
10 changes: 4 additions & 6 deletions core/src/main/scala/org/apache/spark/deploy/client/AppClient.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -68,12 +68,10 @@ private[spark] class AppClient(
// A thread pool for registering with masters. Because registering with a master is a blocking
// action, this thread pool must be able to create "masterRpcAddresses.size" threads at the same
// time so that we can register with all masters.
private val registerMasterThreadPool = new ThreadPoolExecutor(
0,
masterRpcAddresses.length, // Make sure we can register with all masters at the same time
60L, TimeUnit.SECONDS,
new SynchronousQueue[Runnable](),
ThreadUtils.namedThreadFactory("appclient-register-master-threadpool"))
private val registerMasterThreadPool = ThreadUtils.newDaemonCachedThreadPool(
"appclient-register-master-threadpool",
masterRpcAddresses.length // Make sure we can register with all masters at the same time
)

// A scheduled executor for scheduling the registration actions
private val registrationRetryThread =
Expand Down
10 changes: 4 additions & 6 deletions core/src/main/scala/org/apache/spark/deploy/worker/Worker.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -146,12 +146,10 @@ private[deploy] class Worker(
// A thread pool for registering with masters. Because registering with a master is a blocking
// action, this thread pool must be able to create "masterRpcAddresses.size" threads at the same
// time so that we can register with all masters.
private val registerMasterThreadPool = new ThreadPoolExecutor(
0,
masterRpcAddresses.size, // Make sure we can register with all masters at the same time
60L, TimeUnit.SECONDS,
new SynchronousQueue[Runnable](),
ThreadUtils.namedThreadFactory("worker-register-master-threadpool"))
private val registerMasterThreadPool = ThreadUtils.newDaemonCachedThreadPool(
"worker-register-master-threadpool",
masterRpcAddresses.size // Make sure we can register with all masters at the same time
)

var coresUsed = 0
var memoryUsed = 0
Expand Down
14 changes: 4 additions & 10 deletions 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 @@ -25,8 +25,6 @@ import scala.collection.mutable
import scala.collection.mutable.{ArrayBuffer, HashMap, HashSet}
import scala.collection.JavaConverters._

import com.google.common.util.concurrent.ThreadFactoryBuilder

import org.apache.hadoop.conf.Configuration
import org.apache.hadoop.yarn.api.records._
import org.apache.hadoop.yarn.client.api.AMRMClient
Expand All @@ -40,7 +38,7 @@ import org.apache.spark.deploy.yarn.YarnSparkHadoopUtil._
import org.apache.spark.rpc.{RpcCallContext, RpcEndpointRef}
import org.apache.spark.scheduler.{ExecutorExited, ExecutorLossReason}
import org.apache.spark.scheduler.cluster.CoarseGrainedClusterMessages.RemoveExecutor
import org.apache.spark.util.Utils
import org.apache.spark.util.ThreadUtils

/**
* YarnAllocator is charged with requesting containers from the YARN ResourceManager and deciding
Expand Down Expand Up @@ -117,13 +115,9 @@ private[yarn] class YarnAllocator(
// Resource capability requested for each executors
private[yarn] val resource = Resource.newInstance(executorMemory + memoryOverhead, executorCores)

private val launcherPool = new ThreadPoolExecutor(
// max pool size of Integer.MAX_VALUE is ignored because we use an unbounded queue
sparkConf.getInt("spark.yarn.containerLauncherMaxThreads", 25), Integer.MAX_VALUE,
1, TimeUnit.MINUTES,
new LinkedBlockingQueue[Runnable](),
new ThreadFactoryBuilder().setNameFormat("ContainerLauncher #%d").setDaemon(true).build())
launcherPool.allowCoreThreadTimeOut(true)
private val launcherPool = ThreadUtils.newDaemonCachedThreadPool(
Copy link
Member Author

@zsxwing zsxwing Dec 2, 2015

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The previous codes are right. Just use newDaemonCachedThreadPool to make codes simple.

"ContainerLauncher",
sparkConf.getInt("spark.yarn.containerLauncherMaxThreads", 25))

// For testing
private val launchContainers = sparkConf.getBoolean("spark.yarn.launchContainers", true)
Expand Down