Question About Hadoop WordCount Exercise on a Cluster

[KJV]

Hello,

I've been trying to run wordcount exercise using the HDFS cluster. I'm
able to complete the "configure hdfs in a cluster" step and "configure
mapreduce daemon" steps successfully. I can format the file system but
when I run the start-dfs.sh script, I see the following error:

192.168.11.10: starting secondarynamenode, logging to /N/u/train010/
hadoop-0.20.2/bin/../logs/hadoop-train010-secondarynamenode-s10.out
192.168.11.10: Exception in thread "main" java.net.BindException:
Address already in use
192.168.11.10: at sun.nio.ch.Net.bind(Native Method)

I looked at my log file and I do see the Duplicate Protocol Version
error:

2010-07-27 19:11:41,116 INFO org.apache.hadoop.ipc.Server: Error
register getProtocolVersion
java.lang.IllegalArgumentException: Duplicate
metricsName:getProtocolVersion
at org.apache.hadoop.metrics.util.MetricsR

I've followed the instructions and removed all my /tmp files and
started from scratch but I'm still seeing this same error. Has anyone
else seen this? If so, could you let me know how you resolved it.

Thanks,

KJV

[Stephen TAK-LON WU]

Hi,

 

It seems the port of HDFS are already used by some other users ( from your message "Address already in use" ).

 

first, try to stop all the thing with "./stop-all.sh".

 

Then, Please try "pgrep java" too see whether there is process running on your machine.

 

if yes, unfortunately, the previous user of this machine didn't kill the HDFS process before he logout.

 

Please send us a e-mail the node number (e.g. sX or iX, where X is the node number), and we will forward it to IT support team.

 

Best,
Stephen

--
Stephen Wu
Pervasive Technology Institute
Indiana University, Bloomington