DSXchange

The keyword there is native interfaces where as for odbc, a generic connection handle is used.

Where do you want to store the Job end timestamp? You can use system timestamp, you can write a small routine that gets the current timestamp and stores it somewhere or as Craig mentioned, you could get the log file and get the end timestamp of your job.

chulett wrote:Or farm it from the log.

'farm it' huh, cool terminology

You can also get faster loads by increasing the array size.

In the after job subroutine you can get the system date and time. That will be the job end timestamp.

Get rid of the IPC stage. Go to the job properties , performance tab, uncheck 'project defaults', check Enable row buffer and choose Inter process. Run you job. See if you still have the rollback problem.

You cannot do that within a single transformer. Do it in two different transformers.
The result of the first lookup will be fed to the second transformer where in you will do the second lookup.

Give it your best shot, post your code here and then we can go from there. Its not that hard.

What is the value when you do StringToDecimal(). Also what happens when you send rtype as 'round_inf' for StringToDecimal() ?

Go to the columns tab, right click on the column that is nullable and hit "edit row". In the properties window you will see 'Nullable'. You can specify the Null Field Value there.

Get rid of the IPC stage.

Oracle has row level locking so you should be fine if you proceed. But a better design would be to unload the table into a flat file/dataset and proceed from there.

Use strstr() C funtion to locate your string and then do substrings or you can use sscanf() to get the literal you require.

Reset the job and see if any entry "From previous run..." pops up in your log. Post it here for more diagnostic info. Also need more info on your job design.
Is it possible that the database was not available at the time the job was run?

Try doing an exact search on keywords "not found in view-adapted input schema".