Hadoop 2.3.0 - UnknownHostException: Invalid host name: local host is (unknown)

[Larry Compton <la...@gmail.com>]

I've recently upgraded my Hadoop instance from 0.20.2 to 2.3.0. I've
verified that M/R jobs work correctly on the instance and am in the process
of verifying that the instance is capable of executing Giraph jobs
successfully. I'm currently trying to run the
SimpleShortestPathsComputation from the examples and have encountered an
error. Here's an excerpt from the log...

ERROR [org.apache.hadoop.yarn.client.api.async.impl.NMClientAsyncImpl #0]
yarn.GiraphApplicationMaster
(GiraphApplicationMaster.java:onStartContainerError(688)) - Failed to start
Container container_...
java.net.UnknownHostException: Invalid host name: local host is:
(unknown)...

The command I'm using to execute the job is...

$HADOOP_HOME/bin/hadoop jar $GIRAPH_COMPLETE_JAR \
  org.apache.giraph.GiraphRunner \
  org.apache.giraph.examples.SimpleShortestPathsComputation \
  -vif
org.apache.giraph.io.formats.JsonLongDoubleFloatDoubleVertexInputFormat \
  -vip tiny.txt \
  -vof org.apache.giraph.io.formats.IdWithValueTextOutputFormat \
  -op shortest-paths \
  -w 1 \
  -yj $GIRAPH_COMPLETE_JAR

I'm guessing that something isn't configured correctly in my Hadoop
instance or I'm missing a parameter in the execution of the job. Note that
I have also tried tacking on the following additional parameters and I
still see the error...

-ca giraph.zkList=localhost:2181
-ca giraph.SplitMasterWorker=true
-ca mapred.job.tracker=54311
-ca mapreduce.job.tracker=54311

Any idea what might be wrong?

Larry

Re: Hadoop 2.3.0 - UnknownHostException: Invalid host name: local host is (unknown)

[Larry Compton <la...@gmail.com>]

I managed to get around this problem. The Hadoop instance is a single node
development cluster and all of the hostnames references in the Hadoop
configuration files are set to "localhost".
"GiraphConfiguration.getLocalHostname()" was retrieving the hostname set
via the "hostname" command, which was mapped to the actual IP address of
the machine rather than 127.0.0.1. I commented out the IP address line and
added the hostname to the 127.0.0.1 line in "/etc/hosts". This solved up
the problem. In a non-development cluster this wouldn't be a problem.

Larry


On Wed, Jul 2, 2014 at 4:12 PM, Larry Compton <la...@gmail.com>
wrote:

> I've recently upgraded my Hadoop instance from 0.20.2 to 2.3.0. I've
> verified that M/R jobs work correctly on the instance and am in the process
> of verifying that the instance is capable of executing Giraph jobs
> successfully. I'm currently trying to run the
> SimpleShortestPathsComputation from the examples and have encountered an
> error. Here's an excerpt from the log...
>
> ERROR [org.apache.hadoop.yarn.client.api.async.impl.NMClientAsyncImpl #0]
> yarn.GiraphApplicationMaster
> (GiraphApplicationMaster.java:onStartContainerError(688)) - Failed to start
> Container container_...
> java.net.UnknownHostException: Invalid host name: local host is:
> (unknown)...
>
> The command I'm using to execute the job is...
>
> $HADOOP_HOME/bin/hadoop jar $GIRAPH_COMPLETE_JAR \
>   org.apache.giraph.GiraphRunner \
>   org.apache.giraph.examples.SimpleShortestPathsComputation \
>   -vif
> org.apache.giraph.io.formats.JsonLongDoubleFloatDoubleVertexInputFormat \
>   -vip tiny.txt \
>   -vof org.apache.giraph.io.formats.IdWithValueTextOutputFormat \
>   -op shortest-paths \
>   -w 1 \
>   -yj $GIRAPH_COMPLETE_JAR
>
> I'm guessing that something isn't configured correctly in my Hadoop
> instance or I'm missing a parameter in the execution of the job. Note that
> I have also tried tacking on the following additional parameters and I
> still see the error...
>
> -ca giraph.zkList=localhost:2181
> -ca giraph.SplitMasterWorker=true
> -ca mapred.job.tracker=54311
> -ca mapreduce.job.tracker=54311
>
> Any idea what might be wrong?
>
> Larry
>