SAS DI deployed and schedule jobs vs manual- diff results

[sameer112217]

I need one help.

We have one job and it is scheduled. The data transfer is from SQL to Oracle. We used data step and fetch it from SQL into work library. Then we truncate and append the data every time in Oracle. The issue is with numeric variables difference in data.

When the job runs on schedule time, we often notice that we get 0 values for numeric in some rows and the original values of numeric in some. For example

The data source rows can be 7,80,000 but we get the actual numeric values in 1,50,000 and we get null values in others. It keeps on changing every day. And on occasion, we get no null values. But when we run it manually in DI, we never get the null values. We tried to investigate and change the format of source data in SQL, SAS code and the output format in Oracle and we are not understanding how can the results be different when it runs in schedule. I noticed a similar case earlier when we scheduled for some other use. How can it be resolved or investigated further. 

[LinusH]

Assuming you are exiting in the same hosts.
The boring tip is try to find out everything that differs between the two executions, including (but not limited to):
- userid on SAS server, in Oracle and SQL Server
- environment variables
- SAS (and perhaps even in databases) global and line options in effect
- the actual data that is processes (rows)
...