Im using 2017 Professional with Oracle OLEDB drivers with a very simple connection to an Oracle view on the source side. Im sending directly to a clean and flat SQL Server table with no indexes, constraints, nothing. There are no transformations or any other SSIS objects between the the source and destination, just pure pipeline with two OLEDB connections. I have done no tuning and there are no parameters being used. I want the entire view to populate the SQL Server table. I am getting 130,000 rows of the 1.1 million rows of the source Oracle view and it completes with no errors except a shared memory error that is there even when I limit the row count at the view to 500 rows.
Things tried:
We expanded table space on the destination database - resulting in same row count at point of completion
Things not tried:
Tuning buffers since the speed was not an issue and 130k rows made it fine
Monitor pipeline errors
My questions are this:
- What could cause this view and another like it (that is about 800 columns wide) to stop and not report any error that interrupted the data flow on the "Execution Results"?
- With ONLY the Source and Destination OLEDB connectons tied to each other, can I FORCE it to resume following a data error instead of stopping at the point of error (if that is what it is doing)?
- Is error monitoring a requirement in order to FORCE a RESUME of the package processing when an error occurs?
Thanks in advance for your help with this.
Les Draper