IOException: Bad connect ack with firstBadLink

[#1]

Encountering these messages below while running a mapreduce job. Any ideas what's casuing or how to fix ?

Thanks.

Exception in createBlockOutputStream <java.io.IOException: Bad connect ack with firstBadLink as 172.18.9.141:50010>java.io.IOException: Bad connect ack with firstBadLink as 172.18.9.141:50010

    at org.apache.hadoop.hdfs.DFSOutputStream$DataStreamer.createBlockOutputStream(DFSOutputStream.java:1116)

    at org.apache.hadoop.hdfs.DFSOutputStream$DataStreamer.nextBlockOutputStream(DFSOutputStream.java:1039)

    at org.apache.hadoop.hdfs.DFSOutputStream$DataStreamer.run(DFSOutputStream.java:487)

[#2]

What are you seeing when you run this command on the primary network interface on the system you are seeing errors on and the CM server?

ifconfig (interface)

For example

ifconfig eth0

[#3]

I'm getting the same error. We just went through a power outage over the weekend and when we returned, there was a lot of data queued

up to be stored on hdfs. We let it catch up, but while doing so, oozie jobs we had scheduled were being kicked off and many times failing or

being suspended. 

When I run ifconfig, I see what I would expect to see on the interface. Is there anything we did or didn't do on power up that is having us thrashing 

and seemingly in a resource battle as these things run?

I'm two days into this research effort so any insight would be greatly appreciated.