[HADOOP-7123] Hadoop Disk Fail Inplace

• Type: Improvement
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 0.20.203.0
• Assignee: Bharath Mundlapudi
• Reporter: Bharath Mundlapudi
• Created: 28/Jan/11 20:57
• Updated: 22/Mar/13 09:52

Description

This is an umbrella Jira for tracking all the issues and work related for handling disk failures across the Hadoop stack. Today, Hadoop handles disk failures to some extent but not all the issues. This involves understanding both startup and runtime issues related to disk failures in the Hadoop components.

Scope: The initial scope of work will only be for DataNode and TaskTracker related disk failure issues.

Methodology: Injecting disk failures in SCSI layer at various runtime points of Hadoop and observing Hadoop's behavior for improvement.

Issue Links

• incorporates: MAPREDUCE-2657 TaskTracker should handle disk failures
• incorporates: HDFS-2137 Datanode Disk Fail Inplace