An example Apache Hadoop Yarn upgrade - PowerPoint PPT Presentation

About This Presentation
Title:

An example Apache Hadoop Yarn upgrade

Description:

This is a simple example of how Hadoop on Ubuntu Linux can be upgraded from V1 to Yarn. It shows the steps, the configuration, a mapreduce check and the errors encountered. – PowerPoint PPT presentation

Number of Views:308
Slides: 20
Provided by: semtechs
Category:

less

Transcript and Presenter's Notes

Title: An example Apache Hadoop Yarn upgrade


1
Apache Yarn Upgrade
  • Example upgrade
  • From V1 -gt Yarn
  • Environment
  • Approach
  • Install steps
  • Install check

www.semtech-solutions.co.nz info_at_semtech-solutions
.co.nz
2
Yarn Upgrade Environment
  • Java OpenJDK 1.6.0_27
  • Ubuntu 12.04
  • Maven 3.0.4
  • Hadoop 1.2.0
  • Mahout 0.9
  • Hadoop to install
  • 2.0.6-alpha
  • Full details are available from our web site site
  • under guides folder

www.semtech-solutions.co.nz info_at_semtech-solutions
.co.nz
3
Yarn Upgrade Approach
  • Install along side existing Hadoop on all nodes
  • Use existing hdfs
  • Change cfg files on all nodes
  • Set up as single nodes and test via mapreduce
  • Create cluster and test via mapreduce
  • Check web GUI access
  • Full details are available from our web site site
  • under guides folder

www.semtech-solutions.co.nz info_at_semtech-solutions
.co.nz
4
Yarn Upgrade Install
  • Build with Maven into a distribution directory
  • mvn clean package -Pdist -Dtar -DskipTests
    -Pnative
  • release created under ./hadoop-dist/target/hadoop
    -2.0.6-alpha
  • Only skip tests after first build to speed
    things up
  • Configure HOME/.bashrc
  • HADOOP_COMMON_HOME
  • HADOOP_HDFS_HOME
  • HADOOP_MAPRED_HOME
  • HADOOP_YARN_HOME
  • HADOOP_CONF_DIR
  • YARN_CONF_DIR
  • MAPRED_CONF_DIR
  • HADOOP_PREFIX
  • PATH
  • YARN_CLASSPATH

www.semtech-solutions.co.nz info_at_semtech-solutions
.co.nz
5
Yarn Upgrade Install
  • Set up core-site.xml
  • cd HADOOP_COMMON_HOME/etc/hadoop
  • Alter values for
  • fs.default.name
  • hadoop.tmp.dir
  • fs.checkpoint.dir

www.semtech-solutions.co.nz info_at_semtech-solutions
.co.nz
6
Yarn Upgrade Install
  • Set up hdfs-site.xml
  • cd HADOOP_HDFS_HOME/etc/hadoop
  • Alter values for
  • dfs.name.dir
  • dfs.data.dir
  • dfs.http.address
  • dfs.secondary.http.address
  • dfs.https.address

www.semtech-solutions.co.nz info_at_semtech-solutions
.co.nz
7
Yarn Upgrade Install
  • Set up yarn-site.xml
  • cd YARN_CONF_DIR
  • Alter values for
  • yarn.resourcemanager.resource-tracker.address
  • yarn.resourcemanager.scheduler.address
  • yarn.resourcemanager.scheduler.class
  • yarn.resourcemanager.address
  • yarn.nodemanager.local-dirs
  • yarn.nodemanager.address
  • yarn.nodemanager.resource.memory-mb
  • yarn.nodemanager.remote-app-log-dir
  • yarn.nodemanager.log-dirs
  • yarn.nodemanager.aux-services
  • yarn.web-proxy.address

www.semtech-solutions.co.nz info_at_semtech-solutions
.co.nz
8
Yarn Upgrade Install
  • Set up mapred-site.xml
  • cd MAPRED_CONF_DIR
  • Alter values for
  • mapreduce.cluster.temp.dir
  • mapreduce.cluster.local.dir
  • mapreduce.jobhistory.address
  • mapreduce.jobhistory.webapp.address

www.semtech-solutions.co.nz info_at_semtech-solutions
.co.nz
9
Yarn Upgrade Install
  • Set up capcity-scheduler.xml
  • cd HADOOP_YARN_HOME/etc/hadoop
  • Alter values for
  • yarn.scheduler.capacity.maximum-applications
  • yarn.scheduler.capacity.maximum-am-resource-percen
    t
  • yarn.scheduler.capacity.resource-calculator
  • yarn.scheduler.capacity.root.queues
  • yarn.scheduler.capacity.child.queues
  • yarn.scheduler.capacity.child.unfunded.capacity
  • yarn.scheduler.capacity.child.default.capacity
  • yarn.scheduler.capacity.root.capacity
  • yarn.scheduler.capacity.root.unfunded.capacity
  • yarn.scheduler.capacity.root.default.capacity
  • yarn.scheduler.capacity.root.default.user-limit-fa
    ctor
  • yarn.scheduler.capacity.root.default.maximum-capac
    ity
  • yarn.scheduler.capacity.root.default.state
  • yarn.scheduler.capacity.root.default.acl_submit_ap
    plications
  • yarn.scheduler.capacity.root.default.acl_administe
    r_queue
  • yarn.scheduler.capacity.node-locality-delay

www.semtech-solutions.co.nz info_at_semtech-solutions
.co.nz
10
Yarn Upgrade Install
  • Start Resource Manager
  • cd HADOOP_YARN_HOME
  • sbin/yarn-deamon.sh start resourcemanager
  • Start Node Manager
  • cd HADOOP_YARN_HOME
  • sbin/yarn-deamon.sh start ndemanager
  • Test via map reduce job
  • cd HADOOP_MAPRED_HOME/share/hadoop/mapreduce
  • HADOOP_COMMON_HOME/bin/hadoop jar \
  • hadoop-mapreduce-examples-2.0.6-alpha.jar
    randomwriter out

www.semtech-solutions.co.nz info_at_semtech-solutions
.co.nz
11
Yarn Upgrade Install
  • Mapreduce job should end with
  • BYTES_WRITTEN1073750341
  • RECORDS_WRITTEN102099
  • File Input Format Counters
  • Bytes Read0
  • File Output Format Counters
  • Bytes Written1085699265
  • Job ended Sun Aug 25 124535 NZST 2013
  • The job took 89 seconds.
  • Run this test on each node being upgraded

www.semtech-solutions.co.nz info_at_semtech-solutions
.co.nz
12
Yarn Upgrade Install
  • Stop the servers
  • cd HADOOP_YARN_HOME
  • sbin/yarn-daemon.sh stop resourcemanager
  • stopping resourcemanager
  • sbin/yarn-daemon.sh stop nodemanager
  • stopping nodemanager
  • Alter Hadoop env
  • cd HADOOP_CONF_DIR
  • vi hadoop-env.sh
  • add a JAVA_HOME definition at the end. i.e.
  • export JAVA_HOME/usr/lib/jvm/java-6-openjdk-i386

www.semtech-solutions.co.nz info_at_semtech-solutions
.co.nz
13
Yarn Upgrade Install
  • Alter HADOOP_CONF_DIR/slaves file
  • Add details ( one per line ) for slave nodes
  • Format the cluster
  • DONT have the cluster running else you will lose
    data
  • hdfs namenode -format
  • Now proceed to start the cluster

www.semtech-solutions.co.nz info_at_semtech-solutions
.co.nz
14
Yarn Upgrade Install
  • cd HADOOP_COMMON_HOME
  • sbin/hadoop-daemon.sh --config HADOOP_COMMON_HOME
    /etc/hadoop
  • --script hdfs start namenode
  • cd HADOOP_COMMON_HOME
  • sbin/hadoop-daemon.sh --config HADOOP_CONF_DIR
    --script hdfs start datanode
  • cd HADOOP_YARN_HOME
  • sbin/yarn-daemon.sh --config HADOOP_CONF_DIR
    start resourcemanager
  • cd HADOOP_YARN_HOME
  • sbin/yarn-daemon.sh --config HADOOP_CONF_DIR
    start nodemanager
  • cd HADOOP_YARN_HOME
  • bin/yarn start proxyserver --config
    HADOOP_CONF_DIR
  • cd HADOOP_MAPRED_HOME
  • sbin/mr-jobhistory-daemon.sh start historyserver
    --config HADOOP_CONF_DIR

www.semtech-solutions.co.nz info_at_semtech-solutions
.co.nz
15
Yarn Upgrade Install
  • Use jps to check servers running
  • jps
  • 5856 DataNode
  • 6434 Jps
  • 5776 NameNode
  • 6181 NodeManager
  • 6255 WebAppProxyServer
  • 5927 ResourceManager
  • 6352 JobHistoryServer
  • Then run the same mapreduce job on the cluster

www.semtech-solutions.co.nz info_at_semtech-solutions
.co.nz
16
Web Access
www.semtech-solutions.co.nz info_at_semtech-solutions
.co.nz
17
Web Access
www.semtech-solutions.co.nz info_at_semtech-solutions
.co.nz
18
Web Access
www.semtech-solutions.co.nz info_at_semtech-solutions
.co.nz
19
Contact Us
  • Feel free to contact us at
  • www.semtech-solutions.co.nz
  • info_at_semtech-solutions.co.nz
  • We offer IT project consultancy
  • We are happy to hear about your problems
  • You can just pay for those hours that you need
  • To solve your problems
Write a Comment
User Comments (0)
About PowerShow.com