ArcFiles:

From Lawa
Revision as of 04:07, 19 May 2011 by Kereno (Talk | contribs)

Jump to: navigation, search

The ArcTool package is parsing and processing arc files. I uploaded the Jar with sources on /cs/phd/ouaknine/process_arc.jar
Job run example: bin/hadoop jar ~/process_arc.jar ArcProcessing -libjars=/cs/phd/ouaknine/PhD/lawa/arcTools/heritrix-1.14.4.jar,/cs/phd/ouaknine/PhD/lawa/arcTools/fastutil-6.1.0.jar keren1GB/ /user/ouaknine/output/sum4

The reader of the arc file is based on Heritrix parser.

Done:

  • Implemented new InputFormat (ArcInputFormat, ArcRecordReader, ArcRecord)
  • Implemented processing class
  • Implemented layer between HDFS and heritrix's parser
  • Processed 1GB of URLs count on the test cluster. More to come.
  • Memory issues - solved.

To be checked:

  • Corrupted files exist in our data sets. Even though rare (less than 2%), these files prevents to complete the job, thus removed. I am checking issue with Internet Memory.