Uploaded image for project: 'Hadoop HDFS'
  1. Hadoop HDFS
  2. HDFS-8707 Implement an async pure c++ HDFS client
  3. HDFS-10231

libhdfs++: Fix race conditions in RPC layer

    XMLWordPrintableJSON

Details

    • Sub-task
    • Status: Resolved
    • Major
    • Resolution: Fixed
    • None
    • None
    • hdfs-client
    • None

    Description

      Namenode calls seem to hang and the retry logic never properly kicks in. It looks like there's a race condition that leads to a failed rpc call never properly passing the request object to the new RpcConnectionImpl so things hang forever. RpcConnectionImpl objects are also kept alive longer than they should because of a shared_ptr cycle between them and the timeout tracking objects.

      Attachments

        1. HDFS-10231.HDFS-8707.003.patch
          20 kB
          Bob Hansen
        2. HDFS-10231.HDFS-8707.002.patch
          20 kB
          James Clampffer
        3. HDFS-10231.HDFS-8707.001.patch
          20 kB
          James Clampffer
        4. HDFS-10231.HDFS-8707.000.patch
          55 kB
          James Clampffer

        Activity

          People

            bobthansen Bob Hansen
            James Clampffer James Clampffer
            Votes:
            0 Vote for this issue
            Watchers:
            5 Start watching this issue

            Dates

              Created:
              Updated:
              Resolved: