[jira] [Created] (TRAFODION-1740) Use CQDs to prevent time-outs and speed sampling in UPDATE STATS

["David Wayne Birdsall (JIRA)" <ji...@apache.org>]

David Wayne Birdsall created TRAFODION-1740:
-----------------------------------------------

             Summary: Use CQDs to prevent time-outs and speed sampling in UPDATE STATS
                 Key: TRAFODION-1740
                 URL: https://issues.apache.org/jira/browse/TRAFODION-1740
             Project: Apache Trafodion
          Issue Type: Bug
          Components: sql-cmp
    Affects Versions: 2.0-incubating, 1.3-incubating
         Environment: All; however this is particularly noticeable in clusters
            Reporter: David Wayne Birdsall
            Assignee: David Wayne Birdsall


When doing UPDATE STATISTICS on large tables (one example had 17.7 billion rows), we sometimes time-out with an error like the following:

SQL>update statistics for table table1 on every column sample;

*** ERROR[9200] UPDATE STATISTICS for table TRAFODION.SEABASE.TABLE1 encounter
ed an error (8448) from statement Process_Query. [2015-12-15 12:18:45]
*** ERROR[8448] Unable to access Hbase interface. Call to ExpHbaseInterface::nex
tRow returned error HBASE_ACCESS_ERROR(-706). Cause:
java.util.concurrent.ExecutionException: org.apache.hadoop.hbase.client.RetriesE
xhaustedException: Failed after attempts=36, exceptions:
Tue Dec 15 17:18:19 GMT 2015, null, java.net.SocketTimeoutException: callTimeout
=60000, callDuration=60303: row '?. [2015-12-15 12:18:45] 



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)