Informatica Standards: Difference between revisions

From ChipWiki
Jump to navigation Jump to search
Line 7: Line 7:
<LI>Try to avoid overriding reuseable Tasks and Sessions... if you get a "Revert" button while making a change, make sure the change should only apply to THIS instance, rather than every instance, and consider making the change in another place.</LI>
<LI>Try to avoid overriding reuseable Tasks and Sessions... if you get a "Revert" button while making a change, make sure the change should only apply to THIS instance, rather than every instance, and consider making the change in another place.</LI>
<LI>Scripts should always write to <stdout>; only the highest level call should redirect to a log file with "> <log> 2>&1" style</LI>
<LI>Scripts should always write to <stdout>; only the highest level call should redirect to a log file with "> <log> 2>&1" style</LI>
<LI>Use Admin Tables:  ADMIN_FEED_TO_TABLE, etc. (needed for checking MLoad errors, etc.)</LI>
<LI>Set Error Threshold to 1 in MLoad configuration.  This does NOT mean you can avoid a separate check for MLoad errors; UV errors are not properly reported back to Informatica, but ET errors will at least fail the load</LI>
</UL>
</UL>



Revision as of 22:45, 24 August 2007

Informatica Standards and Thoughts

Good Standards

  • Always set the error threshold to 1
  • Always check "fail parent if this task fails"
  • Always check "Run if previous completed" on command tasks
  • Try to avoid overriding reuseable Tasks and Sessions... if you get a "Revert" button while making a change, make sure the change should only apply to THIS instance, rather than every instance, and consider making the change in another place.
  • Scripts should always write to <stdout>; only the highest level call should redirect to a log file with "> <log> 2>&1" style
  • Use Admin Tables: ADMIN_FEED_TO_TABLE, etc. (needed for checking MLoad errors, etc.)
  • Set Error Threshold to 1 in MLoad configuration. This does NOT mean you can avoid a separate check for MLoad errors; UV errors are not properly reported back to Informatica, but ET errors will at least fail the load

Discussion Items

  • Where should shell script redirects to log files be? In the Informatica command task that calls them? Should the script handle its own log file? If a BTEQ script is called from a shell script, should the shell script handle the BTEQ log but something else handle the shell script log? Or should the BTEQ just pass through stdout and stderr to the script, which is then redirected by Informatica?