Uploaded image for project: 'HBase'
  1. HBase
  2. HBASE-27888

Record readBlock message in log when it takes too long time

    XMLWordPrintableJSON

Details

    • Improvement
    • Status: Resolved
    • Minor
    • Resolution: Fixed
    • 2.5.3
    • 2.6.0, 2.5.6, 3.0.0-beta-1
    • HFile
    • None
    • Reviewed
    • Hide
      Add a configuration parameter,which control to record read block slow in logs.
      <property>
        <name>hbase.fs.reader.warn.time.ms</name>
        <value>-1</value>
      </property>
      If reading block cost time in milliseconds more than the threshold, a warning will be logged,the default value is -1, it means skipping record the read block slow warning log.
      Show
      Add a configuration parameter,which control to record read block slow in logs. <property>   <name>hbase.fs.reader.warn.time.ms</name>   <value>-1</value> </property> If reading block cost time in milliseconds more than the threshold, a warning will be logged,the default value is -1, it means skipping record the read block slow warning log.

    Description

      After HBASE-15160， we can record the readBlock message by TRACE log in org.apache.hadoop.hbase.io.hfile.HFileBlock.FSReaderImpl#readBlockDataInternal， But it record all read block message in TRACE log， some times， we only focus the block read cost too much time..

      Attachments

        Issue Links

          is related to

          Sub-task - The sub-task of the issue HBASE-15160 Put back HFile's HDFS op latency sampling code and add metrics for monitoring

          • Critical - Crashes, loss of data, severe memory leak.
          • Closed
          links to

          Web Link GitHub Pull Request #5255

          Activity

            People

              chaijunjie chaijunjie
              chaijunjie chaijunjie
              Votes:
              0 Vote for this issue
              Watchers:
              3 Start watching this issue

              Dates

                Created:
                Updated:
                Resolved: