 |
Arun C Murthy |
at Dec 29, 2007 at 10:18 am
|
⇧ | 
| |
On Fri, Dec 28, 2007 at 11:53:42AM -0800, Jason Venner wrote:
Is it safe to use this to generate a single SequenceFile out of a set of
sequence files produced by reduce?
Nope.
FileUtil.copyMerge just copies bytes of src files into one large heap of a destination file. This will break if src files are SequenceFiles since we now have multiple headers mixed with data.
I've opened
http://issues.apache.org/jira/browse/HADOOP-2501 to cover _merge_ and other useful utilities for SequenceFiles.
this seems to be the source of my damaged sequence files.
Arun
