Interviews with Hadoop examine candidates from different perspectives from the big data perspective. Be prepared to answer questions about Hadoop’s ecosystem components when you are interviewed for the job. HDFS is no exception. Interview questions about HDFS are an important part of any Hadoop interview, as HDFS is a key component of Hadoop.
Are you ready to become a certified Hadoop professional. Get started with preparation and take our online courses for Cloudera Certification and Hortonworks Certification.
This blog will focus on the most important and relevant HDFS interview questions and answers. These Hadoop Interview Questions for HDFS will also highlight the key areas of HDFS that you should focus on.
Most Common HDFS Interview Questions & Answers
1. What is HDFS?
Answer: HDFS is Hadoop Distributed File System. It stores large data sets in Hadoop. It is extremely fault-tolerant and runs on commodity hardware. HDFS uses Master/Slave architecture, which allows for multiple machines to run in a single cluster. The cluster consists of a Nameode and multiple slave nodes, known as DataNodes.

The Namenode stores meta data, i.e. the number of Data Blocks, replicas, locations, as well as other details. Data Node, on the other hand stores the actual data and responds to client requests.
2. What are the components of HDFS?
Answer: HDFS consists of three components:
Namenode
DataNode
Secondary Nameode
3. What is the default block size for DataBlock in HDFS DataNode
Answer: In Hadoop 1.x, the default block size for DataBlock is 64MB. In Hadoop 2.x, it is 128MB.
4. Explain NameNode in Hadoop.
Answer: NameNode is the Master node in HDFS. It contains two important information:
Concerning Hadoop metadata and the file system tree
The in-memory mapping of data blocks and nodes
NameNode contains metadata information such as file permission, file replication factor and block size. It also contains information about the owner of the file. It also includes the mapping between blocks and data nodes.
5. What are fsimages and editlogs in HDFS?
Answer: The metadata for Hadoop files is stored within a file in HDFS Namenode memory, also known as fsimage.
Any changes to the Hadoop filesystem, such as adding or removing files, are made. It is not immediately written to Fsimage, but it is kept in an editlog file on disk. The editlog is synced to the old fsimage file when a name node is created. A new copy is then created.
6. Unix and Linux default block sizes are 4KB. Why is HDFS set to 64MB or 128MB for HDFS?
Answer: A block is the smallest unit that can be stored in a file system. If we take into account the default block size of Linux/Unix to store data in Hadoop, then for a large set of data (petabytes), it will take a large amount of blocks. NameNode will experience performance issues as a result of the increased metadata. In Hadoop 1.x, the default block size was 64MB. In Hadoop 2.x, it is 128MB.
Are you a newbie looking to start a career in Hadoop as a Hadoop developer? You can read our previous blog to learn Hadoop for beginners.
7. What happens when the NameNode begins?
Answer: The NameNode performs the following operations when it starts:
It loads the file system namespace from the last saved FsImage file and the editlog file into its main memory.
This creates the new fsimage by merging the editslog and previous fsimage files to create a new file system namespace.
Receives information from all DataNodes about block locations
8. What is Safe mode in Hadoop
Answer: Safe mode is the maintenance state for the NameNode. The HDFS cluster is read-only during safe mode. The filesystem cannot be modified. This mode does not allow you to delete or duplicate any data blocks.
9. What happens to existing data if you change the HDFS block size?
Answer: The HDFS block size can be changed but it will not affect existing data.
10. What is HDFS replication? What is the default replication factor for HDFS replication?
Answer: HDFS is fault-tolerant to prevent data loss. HDFS stores three copies of each DataBlock on different racks and at different data nodes, which is known to be replication.
The default replication factor for this is 3.
11. What is Secondary NameNode?
Answer: Hadoop metadata are stored in NameNode main storage and disk. This purpose is served by two files –
Editlogs
Fsimage
All HDFS updates are recorded in the editlogs. The file size for the Fsimage file grows as the number of entries increases, but it remains the same. The contents of the editlogs files are written to the fsimage file when the server is restarted. This file is then loaded into main memory, which can be time-consuming. The larger the editlogs files are, the longer it will take to load into Fsimage. This can lead to a prolonged downtime.