MAPREDUCE-5451. Merging change r1539413 from trunk to branch-2.

git-svn-id: https://svn.apache.org/repos/asf/hadoop/common/branches/branch-2@1539416 13f79535-47bb-0310-9956-ffa450edef68
This commit is contained in:
Chris Nauroth 2013-11-06 18:17:32 +00:00
parent 825103c331
commit c02e97c531
4 changed files with 102 additions and 7 deletions

View File

@ -106,6 +106,9 @@ Release 2.2.1 - UNRELEASED
MAPREDUCE-5604. TestMRAMWithNonNormalizedCapabilities fails on Windows due to
exceeding max path length. (cnauroth)
MAPREDUCE-5451. MR uses LD_LIBRARY_PATH which doesn't mean anything in
Windows. (Yingda Chen via cnauroth)
Release 2.2.0 - 2013-10-13
INCOMPATIBLE CHANGES

View File

@ -570,8 +570,10 @@ public interface MRJobConfig {
public static final String MAPRED_ADMIN_USER_ENV =
"mapreduce.admin.user.env";
public static final String DEFAULT_MAPRED_ADMIN_USER_ENV =
"LD_LIBRARY_PATH=$HADOOP_COMMON_HOME/lib/native";
public final String DEFAULT_MAPRED_ADMIN_USER_ENV =
Shell.WINDOWS ?
"PATH=%PATH%;%HADOOP_COMMON_HOME%\\bin":
"LD_LIBRARY_PATH=$HADOOP_COMMON_HOME/lib/native";
public static final String WORKDIR = "work";

View File

@ -439,11 +439,16 @@
<property>
<name>mapreduce.admin.user.env</name>
<value>LD_LIBRARY_PATH=$HADOOP_COMMON_HOME/lib/native</value>
<description>Expert: Additional execution environment entries for
<value></value>
<description>
Expert: Additional execution environment entries for
map and reduce task processes. This is not an additive property.
You must preserve the original value if you want your map and
reduce tasks to have access to native libraries (compression, etc).
When this value is empty, the command to set execution
envrionment will be OS dependent:
For linux, use LD_LIBRARY_PATH=$HADOOP_COMMON_HOME/lib/native.
For windows, use PATH = %PATH%;%HADOOP_COMMON_HOME%\\bin.
</description>
</property>

View File

@ -23,7 +23,9 @@
import static org.junit.Assert.fail;
import java.io.*;
import java.util.HashMap;
import java.util.Iterator;
import java.util.Map;
import org.junit.AfterClass;
import org.junit.BeforeClass;
@ -46,6 +48,7 @@
import org.apache.hadoop.hdfs.MiniDFSCluster;
import org.apache.hadoop.mapreduce.v2.MiniMRYarnCluster;
import org.apache.hadoop.util.Shell;
import org.apache.hadoop.yarn.util.Apps;
/**
* Class to test mapred task's
@ -107,6 +110,29 @@ public void configure(JobConf job) {
}
}
}
/**
* Map class which checks if hadoop lib location
* is in the execution path
*/
public static class ExecutionEnvCheckMapClass extends MapReduceBase
implements Mapper<LongWritable, Text, Text, IntWritable> {
public void map (LongWritable key, Text value,
OutputCollector<Text, IntWritable> output,
Reporter reporter) throws IOException {
}
public void configure(JobConf job) {
String executionEnvPathVariable = System.getenv(Shell.WINDOWS ? "PATH"
: "LD_LIBRARY_PATH");
String hadoopHome = System.getenv("HADOOP_COMMON_HOME");
if (hadoopHome == null) {
hadoopHome = "";
}
String hadoopLibLocation = hadoopHome
+ (Shell.WINDOWS ? "\\bin" : "/lib/native");
assertTrue(executionEnvPathVariable.contains(hadoopLibLocation));
}
}
// configure a job
private void configure(JobConf conf, Path inDir, Path outDir, String input,
@ -153,8 +179,6 @@ public void launchTest(JobConf conf,
Path outDir,
String input)
throws IOException, InterruptedException, ClassNotFoundException {
configure(conf, inDir, outDir, input,
MapClass.class, IdentityReducer.class);
FileSystem outFs = outDir.getFileSystem(conf);
@ -359,7 +383,8 @@ public void testTaskTempDir(){
Path inDir = new Path("testing/wc/input");
Path outDir = new Path("testing/wc/output");
String input = "The input";
configure(conf, inDir, outDir, input,
MapClass.class, IdentityReducer.class);
launchTest(conf, inDir, outDir, input);
} catch(Exception e) {
@ -369,6 +394,66 @@ public void testTaskTempDir(){
}
}
/**
* To test OS dependent setting of default execution path for a MapRed task.
* Mainly that we can use MRJobConfig.DEFAULT_MAPRED_ADMIN_USER_ENV to set -
* for WINDOWS: %HADOOP_COMMON_HOME%\bin is expected to be included in PATH - for
* Linux: $HADOOP_COMMON_HOME/lib/native is expected to be included in
* LD_LIBRARY_PATH
*/
@Test
public void testMapRedExecutionEnv() {
// test if the env variable can be set
try {
// Application environment
Map<String, String> environment = new HashMap<String, String>();
String setupHadoopHomeCommand = Shell.WINDOWS ?
"HADOOP_COMMON_HOME=C:\\fake\\PATH\\to\\hadoop\\common\\home" :
"HADOOP_COMMON_HOME=/fake/path/to/hadoop/common/home";
Apps.setEnvFromInputString(environment, setupHadoopHomeCommand);
// Add the env variables passed by the admin
Apps.setEnvFromInputString(environment, conf.get(
MRJobConfig.MAPRED_ADMIN_USER_ENV,
MRJobConfig.DEFAULT_MAPRED_ADMIN_USER_ENV));
String executionPaths = environment.get(
Shell.WINDOWS ? "PATH" : "LD_LIBRARY_PATH");
String toFind = Shell.WINDOWS ?
"C:\\fake\\PATH\\to\\hadoop\\common\\home\\bin" :
"/fake/path/to/hadoop/common/home/lib/native";
// Ensure execution PATH/LD_LIBRARY_PATH set up pointing to hadoop lib
assertTrue("execution path does not include the hadoop lib location "
+ toFind, executionPaths.contains(toFind));
} catch (Exception e) {
e.printStackTrace();
fail("Exception in testing execution environment for MapReduce task");
tearDown();
}
// now launch a mapreduce job to ensure that the child
// also gets the configured setting for hadoop lib
try {
JobConf conf = new JobConf(mr.getConfig());
// initialize input, output directories
Path inDir = new Path("input");
Path outDir = new Path("output");
String input = "The input";
// set config to use the ExecutionEnvCheckMapClass map class
configure(conf, inDir, outDir, input,
ExecutionEnvCheckMapClass.class, IdentityReducer.class);
launchTest(conf, inDir, outDir, input);
} catch(Exception e) {
e.printStackTrace();
fail("Exception in testing propagation of env setting to child task");
tearDown();
}
}
/**
* Test to test if the user set env variables reflect in the child
* processes. Mainly