General Troubleshooting
Ambari Server: Check /var/log/ambari-server/ambari-server.[log|out] for errors.
Ambari Agent: Check /var/log/ambari-agent/ambari-agent.[log|out] for errors.
Note that if Ambari Agent has any output in /var/log/ambari-agent/ambari-agent.out, it is indicative of a significant problem.
Services fail to start up
- HDFS: Check log files under /var/log/hadoop/hdfs
- MapReduce: Check log files under /var/log/hadoop/mapred
- HBase: Check log files under /var/log/hbase
- Hive: Check log files under /var/log/hive
- Oozie: Check log files under /var/log/oozie
- ZooKeeper: Check log files under /var/log/zookeeper
- WebHCat: Check log files under /var/log/webhcat
- Nagios: Check log files under /var/log/nagios
Nagios alerts don't show up in Ambari Web
- Try running "service httpd restart" on the Nagios Server host.
Install Wizard fails during Install phase
- Click on the Retry button. In most cases, this will solve install failures due to package install problems due to intermittent software repository availability.
Install Wizard failed with warning during Start/Test phase
- Proceed by hitting "Next". Once you are in the Dashboard, go to individual services, reconfigure, and start them to resolve any startup issues.
Installing a new cluster on top of an existing cluster
When installing a Hadoop cluster via Ambari on hosts that already have Hadoop bits installed (including an existing cluster deployed via Ambari), perform the following:
- Stop all the services on all the nodes(including ganglia and nagios)
- It is also a good practice to delete the rpms from all nodes.
search for the rpms:
rpm -qa | grep ganglia
rpm -qa | grep oozie
rpm -qa | grep sqoop
rpm -qa | grep pig
rpm -qa | grep nagios
rpm -qa | grep hadoop
and remove them :
rpm -e <package name>