[HDFS-3764] Hadoop fsck reports files with blocks w/o live replicas as corrupt

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 2.0.0-alpha
• Target Version/s: 2.0.2-alpha
• Assignee: Andy Isaacson
• Reporter: Eli Collins
• Created: 05/Aug/12 02:45
• Updated: 07/Aug/12 21:33

Description

fsck calls FSN#listCorruptFileBlocks, this returns info for files where there's no live replicas, which doesn't necessarily indicate the file is "corrupt". Let's make fsck distinguish between:

• Blocks w/o any lives replicas (mostly the case when repl=1 and a DN goes down)
• Blocks where are live replicas are/were corrupt

Perhaps this could leverage the stats from HDFS-2554.

Issue Links

• relates to: HDFS-2554 Add separate metrics for missing blocks with desired replication level 1