Uploaded image for project: 'Cassandra'
  1. Cassandra
  2. CASSANDRA-1099

make CFS.flushWriter_ size configurable

    Details

    • Type: New Feature
    • Status: Resolved
    • Priority: Low
    • Resolution: Fixed
    • Fix Version/s: 0.7 beta 1
    • Component/s: None
    • Labels:
      None

      Description

      For small heaps (including our out-of-the-box 1G) we want to allow reducing this from the automatic (1 + 2 * DatabaseDescriptor.getAllDataFileLocations().length); having 3 memtables in the queue, plus another being written, is over 200MB which is a significant amount of data for the GC to chew through (see CASSANDRA-1014).

      For large heaps, we might also want to allow increasing this value, to accommodate load spikes better.

      Unfortunately making this configurable at runtime does not appear to be an option due to the lack of a setCapacity method on any of the BlockingQueue classes.

        Attachments

        1. 1099.txt
          4 kB
          Brandon Williams

          Activity

            People

            • Assignee:
              brandon.williams Brandon Williams
              Reporter:
              jbellis Jonathan Ellis
              Authors:
              Brandon Williams
            • Votes:
              0 Vote for this issue
              Watchers:
              0 Start watching this issue

              Dates

              • Created:
                Updated:
                Resolved: