I was trying to setup namenode HA using ambari 2.2.0, on the initialize metadata stage I got error while restarting additional namenode and whole operation just got failed.
I am using CentOS 7 for cluster.
What I did is first I created fresh 2 node cluster and then add 3rd node in this cluster and then I tried to setup namenode HA.
Here is my log snippet of hdfs.
How are you enabling NN HA? I would recommend you to use Ambari to do the same. Have at least three nodes setup in the cluster before you try that.
one thing I noticed is I can ping that machine on which standby namenode is going to be installed but in main namenode startup log I found 3-4 times that connection refused message to that stand by namenode machine
I can show you that log messages if you want @vpoornalingam, See if it helps.
I'm assuming that this attempt was using the Ambari UI wizard to enable HDFS NameNodeHA.
Here are a few things to check to debug this further:
1. Is your Zookeeper quorum up and running? HDFS HA requires Zookeeper, so both NameNodes will need to connect to the Zookeeper quorum in order to start up properly.
2. During the wizard execution, did you run all of the required command-line prompts? For the most part, the UI will guide you to run the correct commands, but if you accidentally forgot to run one, it may be causing other NameNode startup problems in your example.