[HBASE-3081] Log Splitting & Replay: Distinguish between Network IOE and Parsing IOE

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 0.89.20100924, 0.90.0
• Fix Version/s: 0.89.20100924, 0.90.0
• Component/s: io, master, regionserver, Replication
• Hadoop Flags: Reviewed
• Assignee: Nicolas Spiegelberg
• Reporter: Nicolas Spiegelberg
• Created: 05/Oct/10 18:54
• Updated: 12/Oct/12 06:17
• Resolved: 16/Oct/10 05:36

Description

Originally, if HBase got an IOE from HDFS while splitting or opening a region, it would abort the operation. The assumption being that this is a network failure that will likely disappear at a later time or different partition of the network. However, if HBase gets parsing exceptions, we want to log the problem and continue opening/splitting the region anyways, because parsing is an idempotent problem and retries won't fix this issue.

Issue Links

• relates to: HBASE-2933 Skip EOF Errors during Log Recovery
• requires: HADOOP-6986 SequenceFile.Reader should distinguish between Network IOE and Parsing IOE