Uploaded image for project: 'Cassandra'
  1. Cassandra
  2. CASSANDRA-1181

kinder gentler compaction

Agile BoardAttach filesAttach ScreenshotBulk Copy AttachmentsBulk Move AttachmentsVotersWatch issueWatchersCreate sub-taskConvert to sub-taskMoveLinkCloneLabelsUpdate Comment AuthorReplace String in CommentUpdate Comment VisibilityDelete Comments
    XMLWordPrintableJSON

Details

    • Task
    • Status: Resolved
    • Normal
    • Resolution: Fixed
    • 0.6.3
    • None
    • None

    Description

      I suggested this in a ML thread but it seems that nobody actually tried it. I think it's worth following up on:

      You could try setting the compaction thread to a lower priority. You could add a thread priority to NamedThreadPool, and pass that up from CompactionExecutor constructor. According to http://www.javamex.com/tutorials/threads/priority_what.shtml you have to run as root and add a JVM option to get this to work.

      In particular, Brandon saw stress.py read latencies spike to 100ms during [anti]compaction on a 2 core machine. I'd like to see if this can mitigate that.

      Attachments

        1. CompactionManager.java
          24 kB
          Edward Capriolo
        2. 1181.txt
          3 kB
          Jonathan Ellis
        3. stats.txt
          4 kB
          Edward Capriolo
        4. 1181-trunk.txt
          4 kB
          Brandon Williams

        Activity

          This comment will be Viewable by All Users Viewable by All Users
          Cancel

          People

            brandon.williams Brandon Williams Assign to me
            jbellis Jonathan Ellis
            Brandon Williams
            Jonathan Ellis
            Votes:
            0 Vote for this issue
            Watchers:
            4 Start watching this issue

            Dates

              Created:
              Updated:
              Resolved:

              Slack

                Issue deployment