I have recently started working on hadoop and that too from 2.x onwards.
Also, I do understand the concept of High-Availability of NN in Hadoop 2.x .
But recently someone told me that even on Hadoop 1.x NN HA was possible, because Hadoop 1.x had ZK and QJN .
If so then why most of the articles on web always say NN was SPoF on Hadoop 1.x ?
Please do let me know the answer.
The branch 1.x do not have namenode HA with QJN based. It is production ready and available only from hadoop-2.x.
You can refer HDFS-HA umbrella jira, HDFS-3278.
View solution in original post
Thanks a lot for your response and the clarification.