Uploaded image for project: 'Hadoop Common'
  1. Hadoop Common
  2. HADOOP-13226

Support async call retry and failover

    XMLWordPrintableJSON

Details

    • New Feature
    • Status: Resolved
    • Major
    • Resolution: Fixed
    • None
    • 2.8.0, 3.0.0-alpha1
    • io, ipc
    • None
    • Reviewed

    Description

      In current Async DFS implementation, file system calls are invoked and returns Future<?> immediately to clients. Clients call Future#get to retrieve final results. Future#get internally invokes a chain of callbacks residing in ClientNamenodeProtocolTranslatorPB, ProtobufRpcEngine and ipc.Client. The callback path bypasses the original retry layer/logic designed for synchronous DFS. This proposes refactoring to make retry also works for Async DFS.

      Attachments

        1. h10433_20160524.patch
          31 kB
          Tsz-wo Sze
        2. h10433_20160525.patch
          37 kB
          Tsz-wo Sze
        3. h10433_20160525b.patch
          37 kB
          Tsz-wo Sze
        4. h10433_20160527.patch
          49 kB
          Tsz-wo Sze
        5. h10433_20160528.patch
          45 kB
          Tsz-wo Sze
        6. h10433_20160528c.patch
          47 kB
          Tsz-wo Sze

        Issue Links

          relates to

          Improvement - An improvement or enhancement to an existing feature or task. HADOOP-13227 AsyncCallHandler should use an event driven architecture to handle async calls

          • Major - Major loss of function.
          • Resolved

          Activity

            People

              szetszwo Tsz-wo Sze
              xiaobingo Xiaobing Zhou
              Votes:
              0 Vote for this issue
              Watchers:
              5 Start watching this issue

              Dates

                Created:
                Updated:
                Resolved: