# Hadoop cluster based on the CDH 4 packages. This is the playbook that I used to install and configure the Hadoop cluster @CNR, based on the deb packages found in the Cloudera repositories. No cloudera manager was used nor installed. ## The cluster. The cluster structure is the following: - jobtracker.t.hadoop.research-infrastructures.eu (3.5GB RAM, 4 CPUs): - mapreduce HA jobtracker - zookeeper quorum - HA HDFS journal - quorum4.t.hadoop.research-infrastructures.eu (3.5GB RAM, 4 CPUs): - mapreduce HA jobtracker - zookeeper quorum - HA HDFS journal - nn1.t.hadoop.research-infrastructures.eu (3.5GB RAM, 4 CPUs): - hdfs HA namenode - zookeeper quorum - HA HDFS journal - nn2.t.hadoop.research-infrastructures.eu (3.5GB RAM, 4 CPUs): - hdfs HA namenode - zookeeper quorum - HA HDFS journal - hbase-master.t.hadoop.research-infrastructures.eu (3.5GB RAM, 4 CPUs): - hbase primary master - hbase thrift - zookeeper quorum - HA HDFS journal - hbase-master2.t.hadoop.research-infrastructures.eu (2GB RAM, 2 CPUs): - HBASE secondary master - hbase thrift - node{2..13}.t.hadoop.research-infrastructures.eu (9GB RAM, 8 CPUs, 1000GB external storage for HDFS each): - mapreduce tasktracker - hdfs datanode - hbase regionserver - solr (sharded) - hive.t.hadoop.research-infrastructures.eu: - hue - hive - oozie - sqoop - db.t.hadoop.research-infrastructures.eu: - postgresql instance for hue and hive Su jobtracker.t.hadoop.research-infrastructures.eu sono installati gli script che gestiscono tutti i servizi. È possibile fermare/attivare i singoli servizi oppure tutto il cluster, rispettando l'ordine corretto. Hanno tutti prefisso "service-" e il nome dello script dà un'idea delle operazioni che verranno eseguite: service-global-hadoop-cluster service-global-hbase service-global-hdfs service-global-mapred service-global-zookeeper service-hbase-master service-hbase-regionserver service-hbase-rest service-hdfs-datanode service-hdfs-httpfs service-hdfs-journalnode service-hdfs-namenode service-hdfs-secondarynamenode service-mapreduce-jobtracker service-mapreduce-tasktracker service-zookeeper-server Prendono come parametro "start,stop,status,restart" - jobtracker URL: http://jobtracker.t.hadoop.research-infrastructures.eu:50030/jobtracker.jsp - HDFS URL: http://namenode.t.hadoop.research-infrastructures.eu:50070/dfshealth.jsp - HBASE master URL: http://hbase-master.t.hadoop.research-infrastructures.eu:60010/master-status - HUE Web Interface: http://quorum2.t.hadoop.research-infrastructures.eu:8888 - URL ganglia, per le metriche del cluster: http://monitoring.research-infrastructures.eu/ganglia/?r=hour&cs=&ce=&s=by+name&c=Openaire%252B%2520Hadoop%2520TEST&tab=m&vn= - URL Nagios, per lo stato dei servizi (da attivare): http://monitoring.research-infrastructures.eu/nagios3/ ------------------------------------------------------------------------------------------------ dom0/nodes/san map data dlib18x: *node8* e90.6 (dlibsan9) dlib19x: *node9* e90.7 (dlibsan9) dlib20x: *node10* e90.8 (dlibsan9) dlib22x: *node11* e90.5 (dlibsan9) *node7* e63.4 (dlibsan6) dlib23x: *node12* e80.3 (dlibsan8) *node13* e80.4 (dlibsan8) dlib24x: *node2* e25.1 (dlibsan2) *node3* e74.1 (dlibsan7) dlib25x: *node4* e83.4 (dlibsan8) dlib26x: *node5* e72.1 (dlibsan7) *node6* e63.3 (dlibsan6) ------------------------------------------------------------------------------------------------ Submitting a job (supporting multiple users) To support multiple users you create UNIX user accounts only in the master node. Sul namenode: #groupadd supergroup (da eseguire una sola volta) #adduser claudio ... # su - hdfs $ hadoop dfs -mkdir /home/claudio $ hadoop dfs -chown -R claudio:supergroup /home/claudio (aggiungere claudio al gruppo supergroup) Important: If you do not create /tmp properly, with the right permissions as shown below, you may have problems with CDH components later. Specifically, if you don't create /tmp yourself, another process may create it automatically with restrictive permissions that will prevent your other applications from using it. Create the /tmp directory after HDFS is up and running, and set its permissions to 1777 (drwxrwxrwt), as follows: $ sudo -u hdfs hadoop fs -mkdir /tmp $ sudo -u hdfs hadoop fs -chmod -R 1777 /tmp Note: If Kerberos is enabled, do not use commands in the form sudo -u ; they will fail with a security error. Instead, use the following commands: $ kinit (if you are using a password) or $ kinit -kt (if you are using a keytab) and then, for each command executed by this user, $ Step 8: Create MapReduce /var directories sudo -u hdfs hadoop fs -mkdir -p /var/lib/hadoop-hdfs/cache/mapred/mapred/staging sudo -u hdfs hadoop fs -chmod 1777 /var/lib/hadoop-hdfs/cache/mapred/mapred/staging sudo -u hdfs hadoop fs -chown -R mapred /var/lib/hadoop-hdfs/cache/mapred Step 9: Verify the HDFS File Structure $ sudo -u hdfs hadoop fs -ls -R / You should see: drwxrwxrwt - hdfs supergroup 0 2012-04-19 15:14 /tmp drwxr-xr-x - hdfs supergroup 0 2012-04-19 15:16 /var drwxr-xr-x - hdfs supergroup 0 2012-04-19 15:16 /var/lib drwxr-xr-x - hdfs supergroup 0 2012-04-19 15:16 /var/lib/hadoop-hdfs drwxr-xr-x - hdfs supergroup 0 2012-04-19 15:16 /var/lib/hadoop-hdfs/cache drwxr-xr-x - mapred supergroup 0 2012-04-19 15:19 /var/lib/hadoop-hdfs/cache/mapred drwxr-xr-x - mapred supergroup 0 2012-04-19 15:29 /var/lib/hadoop-hdfs/cache/mapred/mapred drwxrwxrwt - mapred supergroup 0 2012-04-19 15:33 /var/lib/hadoop-hdfs/cache/mapred/mapred/staging Step 10: Create and Configure the mapred.system.dir Directory in HDFS After you start HDFS and create /tmp, but before you start the JobTracker (see the next step), you must also create the HDFS directory specified by the mapred.system.dir parameter (by default ${hadoop.tmp.dir}/mapred/system and configure it to be owned by the mapred user. To create the directory in its default location: $ sudo -u hdfs hadoop fs -mkdir /tmp/mapred/system $ sudo -u hdfs hadoop fs -chown mapred:hadoop /tmp/mapred/system Important: If you create the mapred.system.dir directory in a different location, specify that path in the conf/mapred-site.xml file. When starting up, MapReduce sets the permissions for the mapred.system.dir directory to drwx------, assuming the user mapred owns that directory. Step 11: Start MapReduce To start MapReduce, start the TaskTracker and JobTracker services On each TaskTracker system: $ sudo service hadoop-0.20-mapreduce-tasktracker start On the JobTracker system: $ sudo service hadoop-0.20-mapreduce-jobtracker start Step 12: Create a Home Directory for each MapReduce User Create a home directory for each MapReduce user. It is best to do this on the NameNode; for example: $ sudo -u hdfs hadoop fs -mkdir /user/ $ sudo -u hdfs hadoop fs -chown /user/ where is the Linux username of each user. Alternatively, you can log in as each Linux user (or write a script to do so) and create the home directory as follows: sudo -u hdfs hadoop fs -mkdir /user/$USER sudo -u hdfs hadoop fs -chown $USER /user/$USER ------------------------------------------------------------------------------------------------ We use the jobtracker as provisioning server Correct start order (reverse to obtain the stop order): • HDFS (NB: substitute secondarynamenode with journalnode when we will have HA) • MapReduce • Zookeeper • HBase • Hive Metastore • Hue • Oozie • Ganglia • Nagios I comandi di init si trovano nel file "init.sh" nel repository ansible. Errore da indagare: http://stackoverflow.com/questions/6153560/hbase-client-connectionloss-for-hbase-error # GC hints http://stackoverflow.com/questions/9792590/gc-tuning-preventing-a-full-gc?rq=1 HBASE troubleshooting - Se alcune region rimangono in "transition" indefinitamente, è possibile provare a risolvere il problema da shell: # su - hbase $ hbase hbck -fixAssignments Potrebbe essere utile anche $ hbase hbck -repairHoles ----------------------------------------------------- Quando si verifica: "ROOT stuck in assigning forever" bisogna: - verificare che non ci siano errori relativi a zookeeper. Se ci sono, far ripartire zookeeper e poi tutto il cluster hbase - Far ripartire il solo hbase master ----------------------------------------------------- Quando ci sono tabelle disabilitate, ma che risultano impossibili da abilitare o eliminare: # su - hbase $ hbase hbck -fixAssignments * Restart del master hbase ----------------------------------------------------- Vedi: http://comments.gmane.org/gmane.comp.java.hadoop.hbase.user/32838 Ed in generale, per capire il funzionamento: http://hortonworks.com/blog/apache-hbase-region-splitting-and-merging/ Tool per il monitoraggio di hbase quando è configurato per il manual splitting: https://github.com/sentric/hannibal --------------------------------------------------------------------------------- 2013-02-22 10:24:46,492 INFO org.apache.hadoop.mapred.TaskTracker: Using ResourceCalculatorPlugin : org.apache.hadoop.util.LinuxResourceCalculatorPlugin@41a7fead 2013-02-22 10:24:46,492 INFO org.apache.hadoop.mapred.TaskTracker: Starting thread: Map-events fetcher for all reduce tasks on tracker_node2.t.hadoop.research-infrastructures.eu:localhost/127.0.0.1:47798 2013-02-22 10:24:46,492 WARN org.apache.hadoop.mapred.TaskTracker: TaskTracker's totalMemoryAllottedForTasks is -1 and reserved physical memory is not configured. TaskMemoryManager is disabled. 2013-02-22 10:24:46,571 INFO org.apache.hadoop.mapred.IndexCache: IndexCache created with max memory = 10485760 --- Post interessante che tratta la configurazione ed i vari parametri: http://gbif.blogspot.it/2011/01/setting-up-hadoop-cluster-part-1-manual.html Lista di nomi di parametri deprecati e il loro nuovo nome: http://hadoop.apache.org/docs/current/hadoop-project-dist/hadoop-common/DeprecatedProperties.html --- How to decommission a worker node 1. If they are many, reduce the hdfs redundancy factor 2. Stop the regionserver on the node 3. Add the node to the hdfs and jobtracker exclude list ./run.sh mapred.yml -i inventory/hosts.production -l jt_masters --tags=hadoop_workers ./run.sh hadoop-hdfs.yml -i inventory/hosts.production -l hdfs_masters --tags=hadoop_workers 4. Refresh the hdfs and jobtracker configuration hdfs dfsadmin -refreshNodes mapred mradmin -refreshNodes 5. Remove the node from the list of allowed ones 5a. Edit the inventory 5b. Run ./run.sh hadoop-common.yml -i inventory/hosts.production --tags=hadoop_workers ./run.sh mapred.yml -i inventory/hosts.production -l jt_masters --tags=hadoop_workers ./run.sh hadoop-hdfs.yml -i inventory/hosts.production -l hdfs_masters --tags=hadoop_workers --------------------------------------------------------------------------------- Nagios monitoring - The handlers to restart the services are managed via nrpe. To get them work, we need to: - Add an entry in nrpe.cfg. The command name needs to start with "global_restart_" and the remaining part of the name must coincide with the name of the service. For example: command[global_restart_hadoop-0.20-mapreduce-tasktracker]=/usr/bin/sudo /usr/sbin/service hadoop-0.20-mapreduce-tasktracker restart - Add a handler to the nagios service. The command needs the service name as parameter Example: event_handler restart-service!hadoop-0.20-mapreduce-tasktracker ---------------------------------------------------------------------------------