nutch-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Markus Jelsma <markus.jel...@openindex.io>
Subject RE: Independent Map Reduce to parse Nutch content (Cont.)
Date Mon, 06 Jan 2014 16:13:56 GMT
 Hi - Check the logs first.

-----Original message-----
From: Bin Wang<binwang.cu@gmail.com>
Sent: Saturday 4th January 2014 21:47
To: dev@nutch.apache.org
Subject: Re: Independent Map Reduce to parse Nutch content (Cont.)

Hi Tejas,

I started an AWS instance and run hadoop in single node mode.

When I do..

hadoop -jar example.jar hdfsinput/ hdfsoutput/

Everything works perfect as I expected: a bunch of staff got printed to the screen and both
mappers and reducers got finished without question. In the end, the expected output sits in
the hdfs output directory.

However, when I tried to run the jar file without hadoop:

java -jar example.jar localinput/ localoutput/

It will finish all the mappers without problem but still.. errored out after all the mappers....

Exception in thread "main" java.io.IOException: Job failed!

	at org.apache.hadoop.mapred.JobClient.runJob(JobClient.java:784)

	at arrow.ParseMapred.run(ParseMapred.java:70)

	at org.apache.hadoop.util.ToolRunner.run(ToolRunner.java:70)

	at arrow.ParseMapred.main(ParseMapred.java:18)

I am so confused now why my code doesnt work locally...

Based on my understanding, I can see Nutch constantly uses Hadoop API without hadoop pre-installed..
why cant my code work..

Well, any hint or directional guidance will be appreciated, many thanks!

/usr/bin

On Sat, Jan 4, 2014 at 12:38 AM, Tejas Patil <tejas.patil.cs@gmail.com <mailto:tejas.patil.cs@gmail.com>>
wrote:

Hi Bin Wang,

I would suggest you to NOT use eclipse and run your code over command line. Use logger statements
and see the logs for full stack traces of the failure. In my personal experience, logs are
the best way to debug hadoop code compared to Eclipse debugger.

Thanks,

Tejas

On Fri, Jan 3, 2014 at 8:56 PM, Bin Wang <binwang.cu@gmail.com <mailto:binwang.cu@gmail.com>>
wrote:

Hi,

I tried to modify the code here to parse the nutch content data...

http://svn.apache.org/viewvc/nutch/trunk/src/java/org/apache/nutch/parse/ParseSegment.java?view=markup
<http://svn.apache.org/viewvc/nutch/trunk/src/java/org/apache/nutch/parse/ParseSegment.java?view=markup>

And in the end of this email is a prototype that I have written to run map reduce to calculate
the HTML content length of each URL that I have scraped.

The mapper part runs perfectly fine as expected, however, the whole program stops after all
the mappers finished and the reducer did not get a chance to run: (I am sure there are certain
number of pages got scraped and in the Eclipse console, there are same number of Mapper..
so I assume all the mapper finished.)

Can anyone, who is pretty into writing java map reduce job take a look at my code and see
what the error might be... I am not a Java developer at all so any debug trick or common sense
will be appreciated!

(I heard that it is fairly hard to debug code written using hadoop API... is that true?)

Many thanks!

/usr/bin

_____________________________________________________

Eclipse Console Info

Starting Mapper ...

Key: http://url1 <http://url1>

Result: 134943

Starting Mapper ...

Key: http://url2 <http://url2>

Result: 258588

Exception in thread "main" java.io.IOException: Job failed!

	at org.apache.hadoop.mapred.JobClient.runJob(JobClient.java:784)

	at arrow.ParseMapred.run(ParseMapred.java:68)

	at org.apache.hadoop.util.ToolRunner.run(ToolRunner.java:70)

	at arrow.ParseMapred.main(ParseMapred.java:18)

_____________________________________________________

// my code

package example;

import ...;

public class ParseMapred extends Configured implements Tool,

		Mapper<WritableComparable<?>, Content, Text, IntWritable>,

		Reducer<Text, IntWritable, Text, IntWritable> {

	public static void main(String[] args) throws Exception {

		int res = ToolRunner.run(NutchConfiguration.create(),

				new ParseMapred(), args);

		System.exit(res);

	}

	public void configure(JobConf job) {

		setConf(job);

	}

	public void close() throws IOException {}

	public void reduce(Text key, Iterator<IntWritable> values,

			OutputCollector<Text, IntWritable> output, Reporter reporter)

			throws IOException {

		System.out.println("Starting Reducer ...");

		System.out.println("Reducer: " + "key" + key);

	    output.collect(key, values.next()); // collect first value

	}

	public void map(WritableComparable<?> key, Content content,

			OutputCollector<Text, IntWritable> output, Reporter reporter)

			throws IOException {

		Text url = new Text();

		IntWritable result = new IntWritable();

		url.set("fail");

		result = new IntWritable(1);

		try {

			System.out.println("Starting Mapper ...");

			url.set(key.toString());

			result = new IntWritable(content.getContent().length);

			System.out.println("Key: " + url);

			System.out.println("Result: " + result);

			output.collect(url, result);

		} catch (Exception e) {

			// TODO Auto-generated catch block

			output.collect(url, result);

		}

	}

	public int run(String[] args) throws Exception {

	    JobConf job = new NutchJob(getConf());

	    job.setJobName("ParseData");

	    FileInputFormat.addInputPath(job, new Path("/Users/.../data/"));

	    FileOutputFormat.setOutputPath(job, new Path("/Users/.../result"));

	    job.setInputFormat(SequenceFileInputFormat.class);

	    job.setOutputFormat(TextOutputFormat.class);

	    job.setOutputKeyClass(Text.class);

	    job.setOutputValueClass(IntWritable.class);

	    job.setMapperClass(ParseMapred.class);

	    job.setReducerClass(ParseMapred.class);

	    JobClient.runJob(job);

		return 0;

	}

}



Mime
View raw message