giraph-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Eli Reisman <apache.mail...@gmail.com>
Subject Re: Running SimpleShortestPathsVertex with ToolRunner
Date Fri, 05 Apr 2013 05:39:26 GMT
Also if you write your own Giraph apps, you can package them in the
examples repo for an easy build, or build your own jar and simply include
it in the class path and refer to it where we called
o.a.g.SimpleShortestPathsVertex above.



On Fri, Apr 5, 2013 at 1:37 AM, Eli Reisman <apache.mailbox@gmail.com>wrote:

> Hey guys sorry for the confusion, we're busting our humps to get a release
> done and the docs on the wiki are next up after that (and a book in the
> works as well)
>
> So...here's the problem. Benchmarks are old code, and work great for us
> but run using the API you described.
>
> EXAMPLES however, use a newer execution command. You want to run
> org.apache.giraph.GiraphRunner with the hadoop jar command. the first
> argument to this will be the name of the example class you wan tot run
> (SimpleShortestPathsVertex in your case)
>
> try something like (I'm simplifying here):
>
> hadoop jar --config etc/hadoop path/to/giraph/examples.jar
> org.apache.giraph.GiraphRunner
> org.apache.giraph.examples.SimpleShortestPathsVertex ...
>
> where "..." is the in/out dirs on HDFS, other command line switches like
> -w and even optional hadoop/giraph config params with -ca
> key1=val1,key2=val2,... and others -- use -h option with GiraphRunner (I
> think) to get a dump of the command line opts that are not specific to just
> one example, you will have to refer to the source code to see what custom
> options an application might have.
>
> Hope this helps, sorry for the frustration. Spread the word, docs are
> coming!
>
>
>
>
> On Thu, Apr 4, 2013 at 2:44 PM, David Boyd <dboyd@data-tactics-corp.com>wrote:
>
>>  Jonathan:
>>      The ShortestPathBenchmark code you ran should be a good starting
>> point for your
>> task.  You will certainly have to write some code depending on the format
>> of your input
>> and any specific weights you wanted to apply to edges.   If you can put
>> your data into
>> the same format as that benchmark/example then you should be able to
>> start from
>> there.
>>
>>
>> On 4/4/2013 1:03 PM, Jonathan Ellithorpe wrote:
>>
>> David,
>>
>>  Thank you very much for your prompt response, and also the tip about
>> the number of workers being no more than one less than the available map
>> slots.
>>
>>  I tried the ShortestPathsBenchmark and that does indeed seem to work,
>> so that's good.
>>
>>  My higher level goal is to use giraph to compute the shortest paths in
>> an input graph. Is giraph 0.2 capable of doing this out of the box, or do I
>> need to write my own code? Would switching to giraph 0.1 give me this
>> ability?
>>
>>  Best,
>> Jonathan
>>
>>
>> On Thu, Apr 4, 2013 at 9:35 AM, David Boyd <dboyd@data-tactics-corp.com>wrote:
>>
>>>  All of the documentation for Giraph on both the Githiub wiki and the
>>> the Apache site is out of date as it related to the 0.2 version.
>>> Currently I have tested the following benchmark examples from the core
>>> library:
>>>
>>> PageRankBenchmark
>>> ShortestPathsBenchmark
>>> AggregatorsBenchmark
>>> RandomMessageBenchmark
>>> Help for these examples is provided via the -h option as shown below:
>>>
>>> hadoop jar giraph-jar-with-dependencies.jar
>>> org.apache.giraph.benchmark.PageRankBenchmark -h
>>> usage: org.apache.giraph.benchmark.PageRankBenchmark [-e <arg>] [-h] [-s
>>> <arg>] [-v] [-V <arg>] [-w <arg>]
>>>  -e,--edgesPerVertex <arg>      Edges per vertex
>>>  -h,--help                      Help
>>>  -s,--supersteps <arg>          Supersteps to execute before finishing
>>>  -v,--verbose                   Verbose
>>>  -V,--aggregateVertices <arg>   Aggregate vertices
>>>  -w,--workers <arg>             Number of workers
>>> Just substitute the class name of the example you want help for in the
>>> command above.
>>>
>>> It is critical that the number of workers be NO MORE than one less than
>>> the available number of map slots on your cluster.
>>>
>>> The following was used to execute the PageRankBenchmark example on my
>>> cluster:
>>>
>>>  hadoop jar giraph-jar-with-dependencies.jar
>>> org.apache.giraph.benchmark.PageRankBenchmark
>>> -Dmapred.child.java-opts="-Xmx64g -Xms64g XX:+UseConcMarkSweepGC
>>> -XX:-UseGCOverheadLimit" -Dgiraph.zkList=10.1.94.104:2181 -e 1 -s 3 -v
>>> -V 50000 -w 83
>>>
>>>
>>>
>>>
>>>  On 4/4/2013 12:23 PM, Jonathan Ellithorpe wrote:
>>>
>>> Hi,
>>>
>>>  I am using version 0.2 of giraph and version 0.20.203 of hadoop. I
>>> would like to run the SimpleShortestPathsVertex example, and so I tried the
>>> command in the documentation (
>>> https://cwiki.apache.org/confluence/display/GIRAPH/Shortest+Paths+Example
>>> ):
>>>
>>>  hadoop jar
>>> ~/giraph/giraph-examples/target/giraph-examples-0.2-SNAPSHOT-for-hadoop-0.20.203.0-jar-with-dependencies.jar
>>> org.apache.giraph.examples.SimpleShortestPathsVertex graph output 0 3
>>>
>>>  but I get:
>>>
>>>  Exception in thread "main" java.lang.NoSuchMethodException:
>>> org.apache.giraph.examples.SimpleShortestPathsVertex.main([Ljava.lang.String;)
>>>
>>>  Indicating that the class has no main method. The documentation (
>>> https://cwiki.apache.org/confluence/display/GIRAPH/Shortest+Paths+Example)
>>> in fact states that this is the expected behavior in giraph 0.2 because it
>>> must be run with ToolRunner, but I don't see any instructions for executing
>>> SimpleShortestPathsVertex using ToolRunner.
>>>
>>>  This is the first time that I am running hadoop and giraph, so I am
>>> probably missing something here, and any help would be greatly appreciated.
>>>
>>>  Best,
>>> Jonathan
>>>
>>>
>>>
>>
>>
>

Mime
View raw message