Troubleshoot the Mart

Owned by Becky Weinberg (Unlicensed)
Last updated: Jul 26, 2023 by Pamela Totas (Deactivated)
2 min read

The use of log files generated as part of a Data Mart build. These files, combined with ERX files, are your best troubleshooting tools. Here are some suggestions for how to use those files for particular purposes.

When Data Mart Returns No or Unexpected Results

You may find that a Data Mart build sometimes returns no data, or data that you were not expecting to see.  The log files produced at the lowest level of drill-down in Data Mart Execution Logs (known as the Report, OLAP or Concentration engine level) can help if you run the problem scenario at log level 10, the most detailed.

In the log file that produces unexpected results, look for the words “Begin:Concentration data acquisition”.  What follows will be the query that fetches the basic data required to build the fields provided by that OLAP engine.  You can copy that and paste it into a SQL tool to verify what is returned.  If you don’t see expected results, you may experiment with commenting-out parts of the query such as source, to see if the anomaly is due to one of the assumptions (like source) built into your configuration of the mart.  This exercise may lead you to correct an aspect of your configuration and correct the problem.

When the Build Runs Long

A long-running mart build can result from a number of factors.  Some of these might be resolved with database tuning intervention from your database administrator or Eagle.  The following strategies may be helpful:

  • Determine if processing time is concentrated in one or a few steps of the build.  After running at log level 10, check the “time stamps” of the log file to see if there are sizeable gaps between one or more successive rows of the log.  If you find that one or a few steps account for most of the run time, that can indicate the nature of what is taking time and lead to a faster solution.

  • Sometimes the time-consuming step will be the execution of a particular query. Have your database administrator check the query’s execution plan to determine if indexing or a database hint can reduce run time.  Your particular use of warehouse tables may be atypical, and require intervention to accommodate it