I've got a Derived Columns component as part of a data flow. On this I've set the error output for my columns to Redirect Row in all cases. I've set a data watcher on the error output and then ran the package.
There are several rows which I'm expecting to fail - about 3 of them. These fail but there are also another 697 which seem to have no problem. So I fixed one of the problem columns in the source data and then re-ran the package. I only updated one row in the source so this row no longer appeared in the error output, but neither did several hundred of the other rows.
Is it possible that the error output has been tripped for that one row but for some reason it sends several hundred more rows? The ids on these additional rows follow on from the erroneous row, and when I fix that row the rows following it no longer appear in the error output.
The Derived Column transform should giuve your error handling at a row level. One row that fails should not impact the following rows. This just sounds wrong, so there is either something unexpected you've missed when looking at what rows shoudl fail, or there is a bug. Can you reproduce this behaviour on a simple set of data that could be used to log a bug? Perhaps you could send me a sample package?|||Darren - I've managed to reproduce this with a simple package which reads data from a file. Sent to you via email.
Cheers,
Nick
|||no longer a problem once upgraded to SP2. Thanks Darren.sql
No comments:
Post a Comment