Scribd
Upload
124 views14 pages

Recover From Namenode Failure

This document discusses recovering from Namenode failure in Hadoop. It explains that the Namenode is a single point of failure and stores all HDFS metadata. If it fails, the entire cluster becomes unavailable. However, checkpoints of the metadata are periodically taken by the Secondary Namenode and stored on its local disk. These checkpoints can be used to recover the Namenode by copying the latest fsimage file from the Secondary Namenode to the primary Namenode before restarting it and the HDFS services.

Uploaded by

vishnu
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
124 views14 pages

Recover From Namenode Failure

This document discusses recovering from Namenode failure in Hadoop. It explains that the Namenode is a single point of failure and stores all HDFS metadata. If it fails, the entire cluster becomes unavailable. However, checkpoints of the metadata are periodically taken by the Secondary Namenode and stored on its local disk. These checkpoints can be used to recover the Namenode by copying the latest fsimage file from the Secondary Namenode to the primary Namenode before restarting it and the HDFS services.

Uploaded by

vishnu
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 14

Recover from Namenode failure

For online Hadoop training, send mail to [email protected]


Agenda
What is Namenode
Responsibility of Namenode
Single point of failure
Causes of Namenode failures.
Namenode recovery
Role of Secondary Namenode
FsImage & Edits files.
Checkpoints in Hadoop
Creating checkpoints
Recovery with the help of checkpoint
What is Namenode

Namenode is a process which runs on master


machine of Hadoop cluster.

We need to contact Namenode for any read/write


operation in HDFS.

Namenode keeps metadata of the data which is


stored in HDFS.

Namenode cordinates with datanodes to


read/write data in HDFS.
Responsibility of a Namenode

Namenode keeps a block map of all the files in


HDFS.

It contacts each datanode & ask for block report.

It creates bigger block report from all datanode


block reports.

It keeps list of live nodes & dead nodes.

It balances the storage of Hadoop cluster.


Single point of failure

Namenode is single point of failure in Hadoop


cluster.

Hadoop cluster is not accessible if Namenode is


down.

We can't do any read/write operation, even


datanodes have all the data.

Hot backup is not yet supported in Hadoop.


Causes of Namenode failure

Master machine can stop working due to


hardware problem.

Namenode metadata can get corrupt.

Without metadata, Namenode is not capable of


finding the data in HDFS.

We can't contact Datanode directly for data.

Checkpoint can be used to recover metadata.


Namenode recovery

Namenode must be recovered in order to


access HDFS.

Hadoop cluster will remain offline, untill we


recover Namenode.

Secondary Namenode can help Namenode


to recover.
We can only recover till the last checkpoint saved.
Stale data is far better than no data.
Role of Secondary Namenode

Secondary Namenode must be on separate machine in


Hadoop production cluster.

Add the following entry in hdfs-site.xml to run Secondary


Namnode on another machine.
<property>
<name>dfs.secondary.http.address</name>
<value>192.168.1.2:50090</value>
<description>The secondary Namenode address and port </description>
</property>

Checkpoints, which are stored on Secondary Namenode,


helps in Namenode recovery.
FsImage & Edits files

FSImage contains snapshot of HDFS metadata.

Namenode loads FSImage at it's startup.

After every read/write operation, FSImage is not


updated.

Instead, all the changes are recorded in edits file.

Later, a new FSImage can be created by merging


old FSImage & edits file.
FsImage & Edits files
Creating checkpoints

Checkpoints are taken after every 1 hour ( by default )

Checkpoint are useful for recovering from failure.

We can create checkpoints, by running the below command


Checkpoints in Hadoop
Checkpoint are stored in below shown directory

Fsimage is used to recover the namenode


Recovery with the help of checkpoint
Follow below mentioned steps to recover namenode
1. Stop Hadoop by runnning ./stop-all.sh command

2. Copy fsimage file from checkpoint directory to current directory.

3. Start hadoop by runnning ./start-all.sh command


…Thanks…

For online Hadoop training, send mail to [email protected]

You might also like