понедельник, июля 30, 2012

Install cloudera cluster cheats, add VM node.

  1. disable selinux at all hosts /etc/selinux/config
  2. disable iptables at master or fix rules
  3. set all node hosts of cluster in all /etc/hosts
  4. change the host name in all /etc/cloudera-scm-agent/config.ini

  1. enter into runned instance and change mac adress in /etc/sysconfig/network-scripts/ifcfg-eth0
  2. shutdown
  3. change mac address into Virtual Box VM preferences
  4. start VM
  5. ifconfig
  6. add host to /etc/hosts
    • 10.0.0.10   node10.cloudera node10
  7. change host at  /etc/sysconfig/network
    • HOSTNAME=node10.clouder

вторник, июля 17, 2012

MapReduce doens't run when i tried install hadoop (CDH4) on Centos 6 (minimal) using Cloudera Manager (Free Edition)

I tried install full package include: HBase, HDFS, Hue, MapReduce, Oozie, ZooKeeper

But on start MapReduce I got error that JobTracker doesn't want to run.

After a few hours I found in log /var/log/hadoop-0.20-mapreduce/hadoop-cmf-mapreduce1-TASKTRACKER-localhost.localdomain.log.out this row:
 INFO org.apache.hadoop.mapred.TaskTracker: Tasktracker disallowed by JobTracker
solution found in google :
I added the following to the "MapReduce Service Configuration Safety
Valve" in the configuration of the MapReduce service page.

 
    mapred.hosts
    /root/mapred_approved_

hosts
    true
 

Then created /root/mapred_approved_hosts containing:
localhost
localhost.localdomain
In  Cloudera Manager click Services->MapReduce in category search type: "MapReduce Service Configuration Safety Valve" edit value and save, after it restart all services

p.s. so MapReduse run but i have problem to start Beeswax Server :)

Beeswax Server problem fixed, port 8002 used by other program

[me@localhost]# nmap -sT -O localhost
8002/tcp  open  teradataordbms

so I changed  port of Beeswax to 8004 and restarted service