Note that there are some explanatory texts on larger screens.

plurals
  1. USraven_arkadon
    primarykey
    data
    text
    plurals
    1. This table or related slice is empty.
    1. COIf I understand correctly, you can only programmatically create servlets during initialization of the context - i.e. NOT as a side effect of a (HTTP) request: According to [ServletContext.addServlet(...)](http://docs.oracle.com/javaee/6/api/javax/servlet/ServletContext.html#addServlet%28java.lang.String,%20java.lang.Class%29): IllegalStateException - if this ServletContext has already been initialized
      singulars
    2. COSry for the late response; Is it possible, that my idea cannot work: The source file is split up and the splits are sent to the nodes. The nodes then read the records from their corresponding split. So duplicate records in the source file may be located in several splits and therefore become spread across several nodes. Therefore getting a grouping of the duplicates is only possible in the reducer. Is this correct?
      singulars
    3. COHi, I was a bit unclear: The point is, that I want the combiners to only hand the summary (the number of lines in 1, 2, and common) to the reducer - there is no need that all lines are sent back to the reducer. But for this to work, the combiners must see the records of both files together (my RecordReader already produces (line, fileId) pairs; the mapping from filename to fileId is passed around with the config object). However, when adding the files with two FileInputFormat.addInputPath(job, file) statements, the files get processed individually, so the combiners do not see their 'union'.
      singulars
 

Querying!

 
Guidance

SQuiL has stopped working due to an internal error.

If you are curious you may find further information in the browser console, which is accessible through the devtools (F12).

Reload