dzcxzl created HDFS-17995:
-----------------------------

             Summary: DirectoryScanner should schedule scans with a fixed delay 
and log scan duration
                 Key: HDFS-17995
                 URL: https://issues.apache.org/jira/browse/HDFS-17995
             Project: Hadoop HDFS
          Issue Type: Improvement
          Components: hdfs
            Reporter: dzcxzl


  DirectoryScanner schedules itself with 
ScheduledThreadPoolExecutor#scheduleAtFixedRate.
  With fixed-rate scheduling the next execution is computed from the previous 
*scheduled*
  time, not from the time the previous run finished. If a single scan takes 
longer than
  dfs.datanode.directoryscan.interval, the executor runs the missed executions 
back-to-back
  until it has caught up with the schedule. On DataNodes with a large number of 
blocks the
  scanner therefore ends up running almost continuously.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to