Using GiraphRunner

[neilac333]

I am trying to use SHDP to run a Giraph job by using GiraphRunner and its associated switches like so:

hadoop jar giraphjob.jar 
                    org.apache.giraph.GiraphRunner 
                    com.myapp.MyVertex 
                    -w  3
                    -if  com.myapp.VertexInputFormat 
                    -of  com.myapp.VertexOutputFormat 
                    -ip  inputPath
                    -op  outputPath

It is not obvious to me how to do this with <hdp:tool-runner>. Any help is appreciated.

Thanks.

[shekhina]

(I'm not familiar with english)

I don't know Giraph.
Hmm... GiraphRunner is implemented org.apache.hadoop.util.Tool.
I will explain my setting/codes. (Perhaps, it is not good ways but run correctly)
In xml

<hdp:tool-runner id="giraphRunnerBean"
tool-class="org.apache.giraph.GiraphRunner"
jar="file://${giraphjob.jar.location}"
run-at-startup="false">
</hdp:tool-runner>

'giraphRunnerBean' is general spring bean and extends org.springframework.data.hadoop.mapreduce.ToolRunn er.

You have to write some another bean that is injected giraphRunnerBean.
Inside the another bean (my case spring-batch Tasklet),

giraphRunnerBean.setArguments(
"com.myapp.MyVertex",
"-w", "3",
"-if", "com.myapp.VertexInputFormat",
"-of", "com.myapp.VertexOutputFormat ", ...)
giraphRunnerBean.call();

I hope it helps.

[neilac333]

It seems a little cumbersome to write two declarations for this one task, but I will try this approach. Thanks for your help.

[shekhina]

neilac333/ You are right.
There is a simpler way.

<hdp:tool-runner id="giraphRunnerBean"
tool-class="org.apache.giraph.GiraphRunner"
jar="file://${giraphjob.jar.location}"
run-at-startup="false">

<hdp:arg value="-a"/>
<hdp:arg value="someVal"/>
<hdp:arg value="-b"/>
<hdp:arg value="otherVal"/>
</hdp:tool-runner>

[Costin Leau]

Indeed - check out the documentation which has an entry exactly for this: http://static.springsource.org/sprin...p:tool-scripts