On a MapR 4.0.1 cluster, Revolution R mapreduce jobs may fail with the error:
To work around the problem, confirm the following post-installation configuration steps have been taken for Revolution R 7.3 on MapR 4.0.1 clusters:
On the "client" node, create the native filesystem shared directory for each user:
1. sudo mkdir -p /var/RevoShare/mapr
3. chmod 777 /var/RevoShare/mapr
4. add . /usr/lib64/Revo-7.3/scripts/RevoHadoopEnvVars.site to /home/mapr/.bash_profile
On EVERY worker node, make the following changes:
The installation of Revolution R automatically creates a link to scaleR-hadoop-0.1-SNAPSHOT.jar in $HADOOP_HOME/lib. However, this is the incorrect directory (a bug has been reported) and the file should be linked or copied in /opt/mapr/hadoop/hadoop-2.4.1/share/hadoop/common/lib/.
Linking /opt/mapr/hadoop/hadoop-0.20.2/c++/Linux-amd64-64/lib/libhdfs.so in /usr/lib64 will ensure that the proper libhdfs.so is found by Revolution R. The third step is necessary to move the problem libhdfs.so out of the way.
2015-02-13 08:00:37,657 FATAL [IPC Server handler 2 on 52243] org.apache.hadoop.mapred.TaskAttemptListenerImpl: Task: attempt_1423734868663_0289_m_000000_0 - exited : java.io.IOException: R VM exited with exit code: 1, output:
starting script
[1] TRUE
$fileName
[1] NA
$start
[1] 0
$length
[1] 0
/usr/lib64/Revo-7.3/R-3.1.1/lib64/R/library/RevoScaleR/rxLibs/x64/libExaCore.so.2(_Z21CriticalSignalHandleri+0x1e)[0x7fe02d3d423e]
/usr/lib/jvm/java-1.7.0-openjdk-1.7.0.71.x86_64/jre/lib/amd64/server/libjvm.so(+0x7e2628)[0x7fe0282f8628]
/usr/lib/jvm/java-1.7.0-openjdk-1.7.0.71.x86_64/jre/lib/amd64/server/libjvm.so(JVM_handle_linux_signal+0x95)[0x7fe0282fea55]
/lib64/libpthread.so.0[0x30cec0f710]
/usr/lib64/Revo-7.3/R-3.1.1/lib64/R/library/RevoScaleR/rxLibs/x64/libExaCore.so.2(_ZNK10CxAnalysis19HaveWriteDataSourceEv+0x0)[0x7fe02acbc280]
/usr/lib64/Revo-7.3/R-3.1.1/lib64/R/library/RevoScaleR/rxLibs/x64/libExaCore.so.2(_ZN10CxAnalysis18GetWriteDataSourceEv+0x9)[0x7fe02acbcb49]
/usr/lib64/Revo-7.3/R-3.1.1/lib64/R/library/RevoScaleR/rxLibs/x64/libRxLink.so.2(_Z19RxComputeHadoopTaskR11RListObjectR11RxAllocator+0x3b3)[0x7fe02e2a5d23]
/usr/lib64/Revo-7.3/R-3.1.1/lib64/R/library/RevoScaleR/rxLibs/x64/libRxLink.so.2(_Z10RxUtilBaseP7SEXPREC+0x6db)[0x7fe02e2b2fab]
/usr/lib64/Revo-7.3/R-3.1.1/lib64/R/library/RevoScaleR/rxLibs/x64/libRxLink.so.2(RxUtil+0x34)[0x7fe02e2b3f14]
/usr/lib64/Revo-7.3/R-3.1.1/lib64/R/lib/libR.so(+0x9f36b)[0x7fe04083536b]
/usr/lib64/Revo-7.3/R-3.1.1/lib64/R/lib/libR.so(Rf_eval+0x871)[0x7fe04086be91]
/usr/lib64/Revo-7.3/R-3.1.1/lib64/R/lib/libR.so(Rf_ReplIteration+0x212)[0x7fe0408946a2]
/usr/lib64/Revo-7.3/R-3.1.1/lib64/R/lib/libR.so(+0xfea59)[0x7fe040894a59]
/usr/lib64/Revo-7.3/R-3.1.1/lib64/R/lib/libR.so(run_Rmainloop+0x44)[0x7fe040894f64]
/usr/lib64/Revo-7.3/R-3.1.1/lib64/R/bin/exec/R(main+0x1b)[0x40084b]
/lib64/libc.so.6(__libc_start_main+0xfd)[0x30ce41ed5d]
/usr/lib64/Revo-7.3/R-3.1.1/lib64/R/bin/exec/R[0x400739]
Segmentation fault.
at com.revolutionanalytics.scaleR.hadoop.Rscript.waitFor(Rscript.java:103)
at com.revolutionanalytics.scaleR.hadoop.Rscript.run(Rscript.java:59)
at com.revolutionanalytics.scaleR.hadoop.RDelegatingMapper.run(RDelegatingMapper.java:73)
at org.apache.hadoop.mapred.MapTask.runNewMapper(MapTask.java:773)
at org.apache.hadoop.mapred.MapTask.run(MapTask.java:345)
at org.apache.hadoop.mapred.YarnChild$2.run(YarnChild.java:167)
at java.security.AccessController.doPrivileged(Native Method)
at javax.security.auth.Subject.doAs(Subject.java:415)
at org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1469)
at org.apache.hadoop.mapred.YarnChild.main(YarnChild.java:162)
On MapR 4.0.1 platforms, this is due to the Revolution R Enterprise environment configuration using the wrong libhdfs.so library.starting script
[1] TRUE
$fileName
[1] NA
$start
[1] 0
$length
[1] 0
/usr/lib64/Revo-7.3/R-3.1.1/lib64/R/library/RevoScaleR/rxLibs/x64/libExaCore.so.2(_Z21CriticalSignalHandleri+0x1e)[0x7fe02d3d423e]
/usr/lib/jvm/java-1.7.0-openjdk-1.7.0.71.x86_64/jre/lib/amd64/server/libjvm.so(+0x7e2628)[0x7fe0282f8628]
/usr/lib/jvm/java-1.7.0-openjdk-1.7.0.71.x86_64/jre/lib/amd64/server/libjvm.so(JVM_handle_linux_signal+0x95)[0x7fe0282fea55]
/lib64/libpthread.so.0[0x30cec0f710]
/usr/lib64/Revo-7.3/R-3.1.1/lib64/R/library/RevoScaleR/rxLibs/x64/libExaCore.so.2(_ZNK10CxAnalysis19HaveWriteDataSourceEv+0x0)[0x7fe02acbc280]
/usr/lib64/Revo-7.3/R-3.1.1/lib64/R/library/RevoScaleR/rxLibs/x64/libExaCore.so.2(_ZN10CxAnalysis18GetWriteDataSourceEv+0x9)[0x7fe02acbcb49]
/usr/lib64/Revo-7.3/R-3.1.1/lib64/R/library/RevoScaleR/rxLibs/x64/libRxLink.so.2(_Z19RxComputeHadoopTaskR11RListObjectR11RxAllocator+0x3b3)[0x7fe02e2a5d23]
/usr/lib64/Revo-7.3/R-3.1.1/lib64/R/library/RevoScaleR/rxLibs/x64/libRxLink.so.2(_Z10RxUtilBaseP7SEXPREC+0x6db)[0x7fe02e2b2fab]
/usr/lib64/Revo-7.3/R-3.1.1/lib64/R/library/RevoScaleR/rxLibs/x64/libRxLink.so.2(RxUtil+0x34)[0x7fe02e2b3f14]
/usr/lib64/Revo-7.3/R-3.1.1/lib64/R/lib/libR.so(+0x9f36b)[0x7fe04083536b]
/usr/lib64/Revo-7.3/R-3.1.1/lib64/R/lib/libR.so(Rf_eval+0x871)[0x7fe04086be91]
/usr/lib64/Revo-7.3/R-3.1.1/lib64/R/lib/libR.so(Rf_ReplIteration+0x212)[0x7fe0408946a2]
/usr/lib64/Revo-7.3/R-3.1.1/lib64/R/lib/libR.so(+0xfea59)[0x7fe040894a59]
/usr/lib64/Revo-7.3/R-3.1.1/lib64/R/lib/libR.so(run_Rmainloop+0x44)[0x7fe040894f64]
/usr/lib64/Revo-7.3/R-3.1.1/lib64/R/bin/exec/R(main+0x1b)[0x40084b]
/lib64/libc.so.6(__libc_start_main+0xfd)[0x30ce41ed5d]
/usr/lib64/Revo-7.3/R-3.1.1/lib64/R/bin/exec/R[0x400739]
Segmentation fault.
at com.revolutionanalytics.scaleR.hadoop.Rscript.waitFor(Rscript.java:103)
at com.revolutionanalytics.scaleR.hadoop.Rscript.run(Rscript.java:59)
at com.revolutionanalytics.scaleR.hadoop.RDelegatingMapper.run(RDelegatingMapper.java:73)
at org.apache.hadoop.mapred.MapTask.runNewMapper(MapTask.java:773)
at org.apache.hadoop.mapred.MapTask.run(MapTask.java:345)
at org.apache.hadoop.mapred.YarnChild$2.run(YarnChild.java:167)
at java.security.AccessController.doPrivileged(Native Method)
at javax.security.auth.Subject.doAs(Subject.java:415)
at org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1469)
at org.apache.hadoop.mapred.YarnChild.main(YarnChild.java:162)
To work around the problem, confirm the following post-installation configuration steps have been taken for Revolution R 7.3 on MapR 4.0.1 clusters:
On the "client" node, create the native filesystem shared directory for each user:
1. sudo mkdir -p /var/RevoShare/mapr
3. chmod 777 /var/RevoShare/mapr
4. add . /usr/lib64/Revo-7.3/scripts/RevoHadoopEnvVars.site to /home/mapr/.bash_profile
On EVERY worker node, make the following changes:
# cp scaleR-hadoop-0.1-SNAPSHOT.jar /opt/mapr/hadoop/hadoop-2.4.1/share/hadoop/common/lib/
# ln -s /opt/mapr/hadoop/hadoop-0.20.2/c++/Linux-amd64-64/lib/libhdfs.so /usr/lib64
# mv /opt/mapr/hadoop/hadoop-2.4.1/lib/native/libhdfs.so /opt/mapr/hadoop/hadoop-2.4.1/lib/native/libhdfs.so.saveThe installation of Revolution R automatically creates a link to scaleR-hadoop-0.1-SNAPSHOT.jar in $HADOOP_HOME/lib. However, this is the incorrect directory (a bug has been reported) and the file should be linked or copied in /opt/mapr/hadoop/hadoop-2.4.1/share/hadoop/common/lib/.
Linking /opt/mapr/hadoop/hadoop-0.20.2/c++/Linux-amd64-64/lib/libhdfs.so in /usr/lib64 will ensure that the proper libhdfs.so is found by Revolution R. The third step is necessary to move the problem libhdfs.so out of the way.