Maybe you just have to add 'mapred' user to the group that owns the 'hdfs' root directory, it seems the group name is called: 'hdfs'.
On Sat, Feb 15, 2014 at 9:56 AM, Panshul Whisper <ouchwhis...@gmail.com>wrote: > Hello, > > I had a Cloudera Hadoop cluster running on AWS EC2 instances. I used the > Cloudera Manager to setup and configure the cluster. It is a CDH 4.4.0 > version package. > > Everything was running fine till I got a mail from Amazon saying that I > need to re-instantiate one of my EC2 instances and create a new one as the > old one was to be terminated. Un-luckily it was the the Master Node. > So I did the following: > > 1. Made an image of the running Master Node instance. > 2. Stopped the cluster services from the cloudera manager. > 3. Terminated the existing the node. > 4. Setup a new node with the AMI image of old Master node. > 5. Configured the same private IP Address as the old master node, so > it also got the same DNS address. > 6. Since the new machine was a clone of the old master node. I did not > change anything. > 7. I restarted all the services from Cloudera Manager. > 8. NameNode failed to start saying that DFS was not formatted. > 9. I formatted the DFS from the Cloudera Manager form the NameNode > service menu. > 10. The restarted the HDFS service with the Namenode service, > Successfully. > 11. Now All the Task Trackers are running. > 12. I can access the Master Node with the old SSH keys and the old IP > Adress. > 13. But the Task Tracker on the name node gives an error while > starting up. > 14. It gives the following error: > *org.apache.hadoop.ipc.RemoteException(org.apache.hadoop.security.AccessControlException): > Permission denied: user=mapred, access=WRITE, > inode="/":hdfs:supergroup:drwxr-xr-x* > > Please help me resolve this error. Earlier also I did not have a user > named mapred. Everything was running fine with the user hdfs. > > Thanking You, > > -- > Regards, > Ouch Whisper > 010101010101 >