SPARK-1252. On YARN, use container-log4j.properties for executors

container-log4j.properties is a file that YARN provides so that containers can have log4j.properties distinct from that of the NodeManagers.

Logs now go to syslog, and stderr and stdout just have the process's standard err and standard out.

I tested this on pseudo-distributed clusters for both yarn (Hadoop 2.2) and yarn-alpha (Hadoop 0.23.7)/

Author: Sandy Ryza <sandy@cloudera.com>

Closes #148 from sryza/sandy-spark-1252 and squashes the following commits:

c0043b8 [Sandy Ryza] Put log4j.properties file under common
55823da [Sandy Ryza] Add license headers to new files
10934b8 [Sandy Ryza] Add log4j-spark-container.properties and support SPARK_LOG4J_CONF
e74450b [Sandy Ryza] SPARK-1252. On YARN, use container-log4j.properties for executors
This commit is contained in:
Sandy Ryza 2014-04-07 13:28:14 -05:00 committed by Thomas Graves
parent 83f2a2f14e
commit 9dd8b91662
7 changed files with 53 additions and 5 deletions

View file

@ -81,7 +81,8 @@ class ExecutorRunnable(
credentials.writeTokenStorageToStream(dob)
ctx.setContainerTokens(ByteBuffer.wrap(dob.getData()))
val commands = prepareCommand(masterAddress, slaveId, hostname, executorMemory, executorCores)
val commands = prepareCommand(masterAddress, slaveId, hostname, executorMemory, executorCores,
localResources.contains(ClientBase.LOG4J_PROP))
logInfo("Setting up executor with commands: " + commands)
ctx.setCommands(commands)

View file

@ -0,0 +1,24 @@
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License. See accompanying LICENSE file.
# Set everything to be logged to the console
log4j.rootCategory=INFO, console
log4j.appender.console=org.apache.log4j.ConsoleAppender
log4j.appender.console.target=System.err
log4j.appender.console.layout=org.apache.log4j.PatternLayout
log4j.appender.console.layout.ConversionPattern=%d{yy/MM/dd HH:mm:ss} %p %c{1}: %m%n
# Settings to quiet third party logs that are too verbose
log4j.logger.org.eclipse.jetty=WARN
log4j.logger.org.apache.spark.repl.SparkIMain$exprTyper=INFO
log4j.logger.org.apache.spark.repl.SparkILoop$SparkILoopInterpreter=INFO

View file

@ -266,11 +266,11 @@ trait ClientBase extends Logging {
localResources: HashMap[String, LocalResource],
stagingDir: String): HashMap[String, String] = {
logInfo("Setting up the launch environment")
val log4jConfLocalRes = localResources.getOrElse(ClientBase.LOG4J_PROP, null)
val env = new HashMap[String, String]()
ClientBase.populateClasspath(yarnConf, sparkConf, log4jConfLocalRes != null, env)
ClientBase.populateClasspath(yarnConf, sparkConf, localResources.contains(ClientBase.LOG4J_PROP),
env)
env("SPARK_YARN_MODE") = "true"
env("SPARK_YARN_STAGING_DIR") = stagingDir
env("SPARK_USER") = UserGroupInformation.getCurrentUser().getShortUserName()
@ -344,6 +344,10 @@ trait ClientBase extends Logging {
JAVA_OPTS += " " + env("SPARK_JAVA_OPTS")
}
if (!localResources.contains(ClientBase.LOG4J_PROP)) {
JAVA_OPTS += " " + YarnSparkHadoopUtil.getLoggingArgsForContainerCommandLine()
}
// Command for the ApplicationMaster
val commands = List[String](
Environment.JAVA_HOME.$() + "/bin/java" +

View file

@ -50,7 +50,8 @@ trait ExecutorRunnableUtil extends Logging {
slaveId: String,
hostname: String,
executorMemory: Int,
executorCores: Int) = {
executorCores: Int,
userSpecifiedLogFile: Boolean) = {
// Extra options for the JVM
var JAVA_OPTS = ""
// Set the JVM memory
@ -63,6 +64,10 @@ trait ExecutorRunnableUtil extends Logging {
JAVA_OPTS += " -Djava.io.tmpdir=" +
new Path(Environment.PWD.$(), YarnConfiguration.DEFAULT_CONTAINER_TEMP_DIR) + " "
if (!userSpecifiedLogFile) {
JAVA_OPTS += " " + YarnSparkHadoopUtil.getLoggingArgsForContainerCommandLine()
}
// Commenting it out for now - so that people can refer to the properties if required. Remove
// it once cpuset version is pushed out.
// The context is, default gc for server class machines end up using all cores to do gc - hence

View file

@ -22,6 +22,7 @@ import org.apache.hadoop.mapred.JobConf
import org.apache.hadoop.security.Credentials
import org.apache.hadoop.security.UserGroupInformation
import org.apache.hadoop.yarn.conf.YarnConfiguration
import org.apache.hadoop.yarn.api.ApplicationConstants
import org.apache.hadoop.conf.Configuration
import org.apache.spark.deploy.SparkHadoopUtil
@ -67,3 +68,9 @@ class YarnSparkHadoopUtil extends SparkHadoopUtil {
}
}
object YarnSparkHadoopUtil {
def getLoggingArgsForContainerCommandLine(): String = {
"-Dlog4j.configuration=log4j-spark-container.properties"
}
}

View file

@ -167,6 +167,12 @@
<outputDirectory>target/scala-${scala.binary.version}/classes</outputDirectory>
<testOutputDirectory>target/scala-${scala.binary.version}/test-classes</testOutputDirectory>
<resources>
<resource>
<directory>../common/src/main/resources</directory>
</resource>
</resources>
</build>
</project>

View file

@ -78,7 +78,8 @@ class ExecutorRunnable(
credentials.writeTokenStorageToStream(dob)
ctx.setTokens(ByteBuffer.wrap(dob.getData()))
val commands = prepareCommand(masterAddress, slaveId, hostname, executorMemory, executorCores)
val commands = prepareCommand(masterAddress, slaveId, hostname, executorMemory, executorCores,
localResources.contains(ClientBase.LOG4J_PROP))
logInfo("Setting up executor with commands: " + commands)
ctx.setCommands(commands)