Hadoop Secondary NameNode & DataNode Web Interface
Following on from my previous which provided a tour of the NameNode web interface and servlets; in this post I'll be covering the Secondary NameNode.
Like the NN web UI, you can also view the SNN's log files. The metrics exposed via JMX can be viewed from the URL http://<secondary_namenode>:50090/jmx
That's pretty much it for the Secondary NameNode!
When you click the 'browse the filesystem' link from the NameNode web interface (refer to my post on the NameNode), this redirects you to one of the DataNode's web UI's to browse the directory structure and view files stored in HDFS.
It's also possible to view the contents of a file through the web interface (if it is a text file).
Clicking advanced options shows the blocks associated with the file and the DateNode on which it resides and allows you to download the file or 'tail' the file which works much like the tail command on Linux systems.
DataNodes periodically scan blocks to detect corruption and report corrupt blocks to the NameNode so that they can be re-replicated to maintain the minimum replication factor for the file that the block is associated with.
The DataNode has a web page http://<datanode_host>:50075/blockScannerReport displays information on block scanning statistics.
Secondary NameNode (SNN)
The key information provided on the Secondary NameNode web interface from an administrative standpoint is the last checkpoint time.Like the NN web UI, you can also view the SNN's log files. The metrics exposed via JMX can be viewed from the URL http://<secondary_namenode>:50090/jmx
That's pretty much it for the Secondary NameNode!
DataNode (DN)
Browsing the File System
When you click the 'browse the filesystem' link from the NameNode web interface (refer to my post on the NameNode), this redirects you to one of the DataNode's web UI's to browse the directory structure and view files stored in HDFS.
It's also possible to view the contents of a file through the web interface (if it is a text file).
Clicking advanced options shows the blocks associated with the file and the DateNode on which it resides and allows you to download the file or 'tail' the file which works much like the tail command on Linux systems.
Block Scanner
DataNodes periodically scan blocks to detect corruption and report corrupt blocks to the NameNode so that they can be re-replicated to maintain the minimum replication factor for the file that the block is associated with.
The DataNode has a web page http://<datanode_host>:50075/blockScannerReport displays information on block scanning statistics.
Comments
Post a Comment