Unexpected Non-Update Behavior Due to Numeric Precision Difference in Qlik Replicate

[EE_]

Dear Qlik Support,

We are currently replicating data from DB2 iSeries to Oracle using Qlik Replicate.

We noticed the following warning keeps appearing:
"Source changes that would have had no impact were not applied to the target database. Refer to the 'attrep_apply_exceptions' table for details."

We checked and confirmed that no one has modified the target database directly.

The issue seems to happen on one column (let’s call it column A), defined as NUMBER(13,2) in Oracle.
For example:

• The update value is '-6642500.00', but the value in Oracle is '-66425'.

• Similarly, update value is '6720900.00' , but becomes '67209'.

So far, this only happens to this column. Other numbers are updated correctly. Because of the difference in values, Oracle is not being updated.

Can you help us understand why this is happening?

Thank you for your assistance.

[OritA]

Hi, 
The message that you get points to a situation of data error, where a row to be updated could not be found by Replicate and therefore, the task displayed the message: "Source changes that would have had no impact were not applied to the target database. Refer to the 'attrep_apply_exceptions' table for details."
In your target endpoint, Replicate has created a table named attrep_apply_exceptions.  This table includes details about these kind of data errors that were received by tasks that were applying changes to this endpoint. To get more detail about a specific message you should select this table and search for row related to the specific task  you were running (column name: TASK_NAME ) at the specific time of the message (column name is |ERROR_TIME) and there under the STATEMENT column you'll see the exact statement that got the error and under ERROR column the exact error message. 

Regards,

Orit

[EE_]

Hi @OritA ,

Thank you for your response.

As suggested, we have reviewed the attrep_apply_exceptions table and were able to locate the corresponding entries. We confirmed that the errors are indeed related to a specific table, and more specifically, to a single column within that table.

This column (let’s call it Column A) is defined as NUMBER(13,2) in Oracle.

From our investigation, the issue seems to stem from a mismatch in numeric precision or scale during updates. For example:

• The source update value is -6642500.00, but the value in Oracle becomes -66425

• Similarly, 6720900.00 becomes 67209

To clarify, the target database has not been modified manually, and this issue seems isolated to this particular column. Other numeric fields are replicating without any problems.

Could you please help us understand:

1. Why the value might be truncated or incorrectly scaled during replication?

2. Is there any known issue or setting in Qlik Replicate that might cause this behavior?

3. Are there datatype mapping or transformation settings we should review?

We appreciate your guidance to resolve this issue.

[john_wang]

Hello @EE_ ,

The behavior cannot be reproduced in our lab environment. Please open a support ticket and include the following information:

1. Set the SOURCE_CAPTURE/TARGET_APPLY to Verbose in the task, then reproduce the issue.

2. Decrypt the verbose task log files and attach them to the support ticket.

3. If possible, please provide a SAVE FILE of the DB400 physical file for our investigation — a sample with just a few rows will be sufficient.

Regards,

John.

[john_wang · Accepted solution]

Hello @EE_ ,

It looks like there was a mistake in the original description — the values were increased by a factor of 100, not decreased.

To resolve this issue, please upgrade Qlik Replicate to version 2024.5 SP3 or later.

thanks,

John.