2 Replies Latest reply on Mar 22, 2013 4:09 PM by somasek

    ftp to hdfs route requires a localworkdirectory

    somasek

      Hi,

      I use Fuse 7.1

               I have two routes that transfers files to hdfs.

      The first route is defined to transfer the file via file endpoint to hdfs. (file to hdfs)

      The second route is defined to tranfer the file via ftp endpoint to hdfs. (ftp to hdfs)

       

      Ftp to Hdfs route throws an OutofMemory when i attempt to transfer large files while the first route does not. Currently i use the localWorkDirectory option on ftp endpoint url to overcome this issue.

       

      But this could cause a problem if there is not enough space on the system running the route when the files get really large. And i can also see that the time taken for streaming is increased since the file is transferred to hdfs in two steps (ftp to localworkdirectory and then to hdfs).

       

      I would like to avoid streaming to localworkdirectory. Is it possible?

      I was expecting ftp to hdfs route to behave similar to file to hdfs route.

       

      Thanks,

      Soma