[HDFS-4048] Use ERROR instead of INFO for volume failure logs

• Type: Improvement
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.0.0-alpha
• Fix Version/s: 2.0.3-alpha
• Hadoop Flags: Reviewed
• Assignee: Stephen Chu
• Reporter: Stephen Chu
• Created: 14/Oct/12 20:42
• Updated: 15/Feb/13 13:12
• Resolved: 09/Nov/12 03:14

Description

I misconfigured the permissions of the DataNode data directories (they were owned by root, instead of hdfs).

I wasn't aware of this misconfiguration until a few days later. I usually search through the logs for WARN and ERROR but didn't find messages at these levels that indicated volume failure.

After more carefully reading the logs, I found:

2012-10-01 13:07:10,440 INFO org.apache.hadoop.hdfs.server.common.Storage: Cannot access storage directory /data/4/dfs/dn
2012-10-01 13:07:10,440 INFO org.apache.hadoop.hdfs.server.common.Storage: Storage directory /data/4/dfs/dn does not exist.

I think we should bump the log level to ERROR. This will make the problem more visible to users.