Details

    • Reviewed

    Description

      We introduce the notion of an RMProxy, running on each node (or once per rack). Upon start the AM is forced (via tokens and configuration) to direct all its requests to a new services running on the NM that provide a proxy to the central RM.

      This give us a place to:
      1) perform distributed scheduling decisions
      2) throttling mis-behaving AMs
      3) mask the access to a federation of RMs

      Attachments

        1. YARN-2884-V1.patch
          154 kB
          Kishore Chaliparambil
        2. YARN-2884-V10.patch
          138 kB
          Kishore Chaliparambil
        3. YARN-2884-V11.patch
          137 kB
          Kishore Chaliparambil
        4. YARN-2884-V12.patch
          144 kB
          Kishore Chaliparambil
        5. YARN-2884-V13.patch
          144 kB
          Kishore Chaliparambil
        6. YARN-2884-V2.patch
          134 kB
          Kishore Chaliparambil
        7. YARN-2884-V3.patch
          134 kB
          Kishore Chaliparambil
        8. YARN-2884-V4.patch
          134 kB
          Kishore Chaliparambil
        9. YARN-2884-V5.patch
          134 kB
          Kishore Chaliparambil
        10. YARN-2884-V6.patch
          134 kB
          Kishore Chaliparambil
        11. YARN-2884-V7.patch
          135 kB
          Kishore Chaliparambil
        12. YARN-2884-V8.patch
          136 kB
          Kishore Chaliparambil
        13. YARN-2884-V9.patch
          139 kB
          Kishore Chaliparambil

        Issue Links

          Activity

            People

              kishorch Kishore Chaliparambil
              curino Carlo Curino
              Votes:
              0 Vote for this issue
              Watchers:
              24 Start watching this issue

              Dates

                Created:
                Updated:
                Resolved: