Created 03-21-2017 02:49 AM
Greeings, all.
Can anybody tell me if a File URI is mandatory for each component of the value of the dfs.datanode.data.dir element in the hdfs-site.xml file?
For example, am I required to specify "file:///grid/hadoop/hdfs/dn, file:///grid1/hadoop/hdfs/dn"? Or, could I use "/grid/hadoop/hdfs/dn, /grid1/hadoop/hdfs/dn", instead, and have effectively the same metadata?
I'm using HDP 2.4.3.
Thanks very much!
~ Joseph
Created 03-21-2017 05:27 AM
@Joseph Hawkins - You can just have the value as "/grid/hadoop/hdfs/dn,/grid1/hadoop/hdfs/dn" if they are simple directories on the file system. This is the list of directories where the data will be stored. However, if you have Heterogeneous storage (like DISK, SSD, ARCHIVE or RAM_DISK) you would need to append the File URI.
Below is a very good article explaining HDFS Metadata Directories:
https://hortonworks.com/blog/hdfs-metadata-directories-explained/
Created 03-21-2017 03:14 AM
@Joseph Hawkins Can you please tell me the action/operation which you are trying ?
Created 03-21-2017 05:27 AM
@Joseph Hawkins - You can just have the value as "/grid/hadoop/hdfs/dn,/grid1/hadoop/hdfs/dn" if they are simple directories on the file system. This is the list of directories where the data will be stored. However, if you have Heterogeneous storage (like DISK, SSD, ARCHIVE or RAM_DISK) you would need to append the File URI.
Below is a very good article explaining HDFS Metadata Directories:
https://hortonworks.com/blog/hdfs-metadata-directories-explained/