Uploaded image for project: 'Hadoop HDFS'
  1. Hadoop HDFS
  2. HDFS-11384

Add option for balancer to disperse getBlocks calls to avoid NameNode's rpc.CallQueueLength spike

    XMLWordPrintableJSON

Details

    • Reviewed

    Description

      When running balancer on hadoop cluster which have more than 3000 Datanodes will cause NameNode's rpc.CallQueueLength spike. We observed this situation could cause Hbase cluster failure due to RegionServer's WAL timeout.

      Attachments

        1. balancer.day.png
          58 kB
          yunjiong zhao
        2. balancer.week.png
          50 kB
          yunjiong zhao
        3. HDFS-11384.001.patch
          4 kB
          yunjiong zhao
        4. HDFS-11384.002.patch
          5 kB
          yunjiong zhao
        5. HDFS-11384.003.patch
          5 kB
          Konstantin Shvachko
        6. HDFS-11384.004.patch
          5 kB
          Konstantin Shvachko
        7. HDFS-11384.005.patch
          5 kB
          Konstantin Shvachko
        8. HDFS-11384.006.patch
          13 kB
          Konstantin Shvachko
        9. HDFS-11384.008.patch
          15 kB
          Konstantin Shvachko
        10. HDFS-11384.009.patch
          14 kB
          Konstantin Shvachko
        11. HDFS-11384.010.patch
          16 kB
          Konstantin Shvachko
        12. HDFS-11384.011.patch
          16 kB
          Konstantin Shvachko
        13. HDFS-11384-007.patch
          13 kB
          Konstantin Shvachko
        14. HDFS-11384-branch-2.7.011.patch
          13 kB
          Konstantin Shvachko
        15. HDFS-11384-branch-2.8.011.patch
          13 kB
          Konstantin Shvachko

        Issue Links

          Activity

            People

              shv Konstantin Shvachko
              zhaoyunjiong yunjiong zhao
              Votes:
              0 Vote for this issue
              Watchers:
              13 Start watching this issue

              Dates

                Created:
                Updated:
                Resolved: