From the command line as the hdfs user: # Create the directory $ hdfs dfs -mkdir /user/<username> # Set permissions and ownership $ hdfs dfs -chown <username> /user/<username> $ hdfs dfs -chmod 700 /user/<username> ##Optionally set $ hdfs dfsadmin -setSpaceQuota <bytes_allocated> /user/<username> ## Where bytes_allocated is bytes allowed for this directory (counting replication). This is allocating space for the directory, not by username. So if the user created files in other HDFS directories, this doesn't control that. ... View more

@Wing Lo - The solution may depend on what is actually wrong. It may be that the node is just out of memory. If that's the case, a restart may resolve the issue (but it could occur again). If the index data is actually corrupt, you can take the bad node offline and the errors will stop. However, if the index is actually corrupt, you will need to fix/replace the bad data. I have not used this technique, but you might look at this https://support.lucidworks.com/hc/en-us/articles/202091128-How-to-deal-with-Index-Corruption ... View more