Uploaded image for project: 'Cassandra'
  1. Cassandra
  2. CASSANDRA-8164

OOM due to slow memory meter

Agile BoardAttach filesAttach ScreenshotVotersWatch issueWatchersCreate sub-taskConvert to sub-taskMoveLinkCloneLabelsUpdate Comment AuthorReplace String in CommentUpdate Comment VisibilityDelete Comments
    XMLWordPrintableJSON

    Details

    • Type: Bug
    • Status: Resolved
    • Priority: Normal
    • Resolution: Fixed
    • Fix Version/s: 2.0.12
    • Component/s: None
    • Labels:
      None
    • Severity:
      Normal
    • Since Version:

      Description

      Memory meter holds strong reference to memtable while it iterates over memtable cells. Because meter is not fast, it prevents memtable from being GCed after it has been flushed for some time.

      If write rate is fast enough, this makes node OOM.

      Fixed this by aborting metering if table becomes not active in datatracker, i.e. flushing or flushed.

        Attachments

        Issue Links

          Activity

          $i18n.getText('security.level.explanation', $currentSelection) Viewable by All Users
          Cancel

            People

            • Assignee:
              m0nstermind Oleg Anastasyev Assign to me
              Reporter:
              m0nstermind Oleg Anastasyev
              Authors:
              Oleg Anastasyev
              Reviewers:
              Jonathan Ellis

              Dates

              • Created:
                Updated:
                Resolved:

                Issue deployment