Torsten Curdt wrote:
On 29.08.2007, at 23:26, Doug Cutting wrote:
Raghu Angadi wrote:
What context are you seeing this?
This recently happened during unit testing:
http://lucene.zones.apache.org:8080/hudson/job/Hadoop-Patch/637/testReport/org.apache.hadoop.fs/TestCopyFiles/testCopyFromLocalToDfs/
I'm not sure if that's what caused this test to fail or not. In any
case, the unit test logs might help diagnosis.
Hah! Great! Well, sort of ;)
I just checked our mapred.submit.replication and it is higher than the
nodes in the cluster - maybe that's the problem?
This pretty much assures at least a few of these exceptions.
You can argue that Namenode should not schedule a block to a node
twice.. and I agree.
Raghu.
