Hadoop Administration and Maintenance

“Admins are not only to stop people from doing stupid things, also to stop them from doing clever things!”

In the Hadoop world, a Systems Administrator is called a Hadoop Administrator. Hadoop Admin Roles and Responsibilities include setting up Hadoop clusters. Other duties involve backup, recovery and maintenance. Hadoop administration requires good knowledge of hardware systems and excellent understanding of Hadoop architecture.

It’s easy to get started with Hadoop administration because Linux system administration is a pretty well-known beast, and because systems administrators are used to administering all kinds of existing complex applications. However, there are many common missteps we’re seeing that make us believe there’s a need for some guidance in Hadoop administration. Most of these mistakes come from a lack of understanding about how Hadoop works. Here are just a few of the common issues we find.

With increased adoption of Hadoop in traditional enterprise IT solutions and increased number of Hadoop implementations in production environment, the need for Hadoop Operations and Administration experts to take care of the large Hadoop Clusters is becoming vital.

What does a Hadoop Admin do on day to day ?

• Installation and Configuration
• Cluster Maintenance
• Resource Management
• Security Management
• Troubleshooting
• Cluster Monitoring
• Backup And Recovery Task
• Aligning with the systems engineering team to propose and deploy new hardware and software environments required for Hadoop and to expand existing environments.
• Diligently teaming with the infrastructure, network, database, application and business intelligence teams to guarantee high data quality and availability.

Some of the potential problems, which Hadoop administrators face in day to day operations are caused due to:
Human: Humans cause most common errors in health of systems or machines. Even a simple mistake can create a disaster and lead to down time. To avoid these errors it is important to have proper and proactive diagnostic process in place.
Miss-configuration is another problem that Hadoop administrators come across. Even now, after so much of development in Hadoop, it is considered a young technology. Let us look at the solution to this problem. You need to perform the following steps to get rid of mis-configuration.
• Start with basic parameters related with storage and operation.
• Gain thorough understanding of what needs to be achieved and where we are?
• Do lot of trail runs before moving on with changes in production environment.
Next problem is Hardware. One would rarely hear about a problem related to memory, mother boards and disk controllers but hard drive is mostly the cause of most of the problems. Almost every manufacturer do specify different measures of drive such as: Mean time to failure, Mean time between failures. Hardware does not fail straight away, but they degrade over time and lead to different failures. HDFS is best in detecting corrupt data blocks and automatically replicate to new copies without human intervention.

Remaining factors causing problems for hadoop administrators are:
Resource exhaustion: It is also a major factor that causes problem. As an administrator, one should measure and track task failures so as to help the user identify and correct the processes. Repetitive task failures can occupy task slots and can also take away resources from other jobs. Therefore, it should be seen as a drain on overall capacity.Next is the Host identification and Naming. Incorrectly configured host configuration file, will lead to a situation where client will never be able to communicate with data nodes.
The other problem is network partition: it is a situation where network is unable to communicate with other hosts on segment of network. This means that host X on switch 1 cannot send messages to host Y present on switch 2.

If you are working on Hadoop, you’ll realize there are several shell commands available to manage your hadoop cluster.

1. Hadoop Namenode Commands

2. Hadoop fsck Commands

3. Hadoop Job Commands

4. Hadoop dfsadmin Commands

5. Hadoop Safe Mode (Maintenance Mode) Commands

The following dfsadmin commands helps the cluster to enter or leave safe mode, which is also called as maintenance mode. In this mode, Namenode does not accept any changes to the name space, it does not replicate or delete blocks.

6. Hadoop Configuration Files

7. Hadoop mradmin Commands

8. Hadoop Balancer Commands

9. Hadoop Filesystem Commands

10. Additional Hadoop Filesystem Commands