I am trying to start the datanode on the slave node but when I check the
dfs I only have one node.
When I check the logs on the slave node I find the following output.
2013-03-13 10:22:14,608 INFO
org.apache.hadoop.hdfs.server.datanode.DataNode: STARTUP_MSG:
was this namenode part of any other hadoop cluster?
did you format your namenode and forgot to clean up the datanode?
you can refer Michael's blog for more details and solutions
Hello Cyril,
This is because your datanode has a different namespaceID from the
one which master(namenode) actually has. Have you formatted the HDFS
recently? Were you able to format it properly? Everytime you format HDFS,
NameNode generates new namespaceID, which should be same in both
Thank you both.
So what both of you were saying which will be true is that is order to
start and synchronize the cluster, I will have to format both nodes at the
same time ok.
I was working on the master node without the second node and did not format
before trying to start the second one.
I
No, you don't have to format the NN everytime you add a DN. Looking at your
case, it seems the second DN was part of some other cluster and has the ID
of that NN.
Warm Regards,
Tariq
https://mtariq.jux.com/
cloudfront.blogspot.com
On Wed, Mar 13, 2013 at 9:13 PM, Cyril Bogus