Hi Jeff,
It is not causing it to re-register to the NN.
The PacketResponder gets interrupted when the write pipeline fails in
certain situations - the interruption should cause the pipeline to tear
down, and then the client will retry with a new pipeline (minus the
offending node).
-Todd
On Sun
Hi all,
In my data node logs, it says PacketResponder is interrupt,and it cuase my
data node to re-register to namenode.
The following is the log: (Anyone has any clues ? Thanks )
868_1195229 java.io.EOFException: while trying to read 65557 bytes
2010-07-05 00:00:36,719 INFO
org.apache.hadoop.hd
