Uploaded image for project: 'Hadoop HDFS'
  1. Hadoop HDFS
  2. HDFS-14379

WebHdfsFileSystem.toUrl double encodes characters

    XMLWordPrintableJSON

Details

    • Bug
    • Status: Resolved
    • Major
    • Resolution: Duplicate
    • 3.2.0
    • None
    • hdfs, hdfs-client
    • None

    Description

      When using DistCP over HTTPFS with data that contains Spark partitions, DistCP fails to access the partitioned parquet files since the "=" characters in file path gets double encoded:
      "/test/spark/partition/year=2019/month=1/day=1"
      to
      "/test/spark/partition/year%253D2019/month%253D1/day%253D1"

      This happens since fsPathItem containing the character '=' is encoded by URLEncoder.encode(fsPathItem, "UTF-8") to '%3D' and then encoded again by new Path(....) to '%253D'.

      Attachments

        1. HDFS-14379-patch1.patch
          2 kB
          Boris Vulikh
        2. HDFS-14379-patch3.patch
          2 kB
          Boris Vulikh

        Issue Links

          Activity

            People

              borisvu Boris Vulikh
              borisvu Boris Vulikh
              Votes:
              0 Vote for this issue
              Watchers:
              6 Start watching this issue

              Dates

                Created:
                Updated:
                Resolved: