Uploaded image for project: 'IMPALA'
  1. IMPALA
  2. IMPALA-1070

Incorrect Parameter Name in Documentation (dfs.client.file-block-storage-locations.timeout.millis)

    XMLWordPrintableJSON

Details

    • Bug
    • Status: Resolved
    • Minor
    • Resolution: Won't Fix
    • Impala 1.3.1
    • None
    • Docs
    • None
    • CentOS 6.4 x86_64 (kernel 2.6.32-358.23.2.el6.x86_64) running on VMWare VM, with CDH 4.7.0 and Impala 1.3.1.

    Description

      The documentation (http://www.cloudera.com/content/cloudera-content/cloudera-docs/Impala/latest/Installing-and-Using-Impala/ciiu_config_performance.html) says to configure a parameter 'dfs.client.file-block-storage-locations.timeout.millis' in the hdfs-site.xml file, providing it with a value of '10000' (this appears under both the 'To configure DataNodes for short-circuit reads with CDH 4.2 or later' heading and the 'To configure DataNodes for short-circuit reads with CDH 4.1' heading.

      I did so, but the impala-server daemon failed to start. Upon investigating the problem, I saw the following in the /var/log/impala/impalad.ERROR file:

      Log file created at: 2014/07/04 13:22:11
      Running on machine: localhost.localdomain
      Log line format: [IWEF]mmdd hh:mm:ss.uuuuuu threadid file:line] msg
      E0704 13:22:11.658120 1302 impala-server.cc:208] ERROR: block location tracking is not properly enabled because

      • dfs.client.file-block-storage-locations.timeout is too low. It should be at least 3000.
        E0704 13:22:11.658340 1302 impala-server.cc:210] Aborting Impala Server startup due to improper configuration

      I doublechecked my /etc/hadoop/conf/hdfs-site.xml and /etc/impala/conf/hdfs-site.xml files and verified that I had configured the timeout parameter as described.

      I then noticed a subtle but apparently important distinction. The error message is complaining about a parameter named 'dfs.client.file-block-storage-locations.timeout' while the documentation instructed me to set one named 'dfs.client.file-block-storage-locations.timeout.millis' (i.e. the parameter in the log is missing the '.millis' on the end).

      After updating the /etc/hadoop/conf/hdfs-site.xml and /etc/impala/conf/hdfs-site.xml files to set a parameter 'dfs.client.file-block-storage-locations.timeout' with a value of 10000, I restarted the impala-server daemon. It then worked as expected.

      Attachments

        Activity

          People

            jrussell John Russell
            twheeler@cloudera.com_impala_982d Tom Wheeler
            Votes:
            0 Vote for this issue
            Watchers:
            4 Start watching this issue

            Dates

              Created:
              Updated:
              Resolved: