Suresh Subbiah updated TRAFODION-1235:
    Fix Version/s:     (was: 2.2.0)

> LP Bug: 1453969 - User CQDs should not affect internal query compilation and 
> execution
> --------------------------------------------------------------------------------------
>                 Key: TRAFODION-1235
>                 URL: https://issues.apache.org/jira/browse/TRAFODION-1235
>             Project: Apache Trafodion
>          Issue Type: Bug
>          Components: sql-exe
>            Reporter: Weishiun Tsai
>            Assignee: Sandhya Sundaresan
>            Priority: Critical
> Currently, user CQD settings would impact internal query compilation and 
> execution.  This is causing strange errors in unexpected places.  The 
> following scenario is one of such examples.  Setting the CQD 
> UPDATE_CLUSTERING_OR_UNIQUE_INDEX_KEY ‘OFF’ caused the next alter rename 
> statement to fail with a perplexing 4033 error.  The statement worked fine 
> after the CQD was reset.
> This case is filed not only about this particular CQD.  In general, we need 
> to avoid sending user CQDs to internal query compilation and execution.  
> Trafodion has hundreds of CQDs.  Some of them may be recommended to customers 
> in the future to improve performance or to work around certain problems.  If 
> we don’t stop sending CQDs to internal query compilation, this problem is 
> bound to show up again and again in the future.
> This problem can be seen on the v0505 build.
> -------------------------------------------------------------------------------
> Here is the entire script to reproduce it:
> create schema testsch;
> set schema testsch;
> create table t1 (a int);
> alter table t1 rename to t2;
> control query default UPDATE_CLUSTERING_OR_UNIQUE_INDEX_KEY reset;
> alter table t1 rename to t2;
> drop schema testsch cascade;
> -------------------------------------------------------------------------------
> Here is the execution output:
> >>create schema testsch;
> --- SQL operation complete.
> >>set schema testsch;
> --- SQL operation complete.
> >>create table t1 (a int);
> --- SQL operation complete.
> >>control query default UPDATE_CLUSTERING_OR_UNIQUE_INDEX_KEY 'OFF';
> --- SQL operation complete.
> >>alter table t1 rename to t2;
> *** ERROR[4033] Column CATALOG_NAME is a primary or clustering key column and 
> cannot be updated.
> *** ERROR[8822] The statement was not prepared.
> --- SQL operation failed with errors.
> >>control query default UPDATE_CLUSTERING_OR_UNIQUE_INDEX_KEY reset;
> --- SQL operation complete.
> >>alter table t1 rename to t2;
> --- SQL operation complete.
> >>drop schema testsch cascade;
> --- SQL operation complete.

This message was sent by Atlassian JIRA

Reply via email to