Oracle source endpoint unexpected behavior

[Ananguyen34]

Hi team,

Here is our configuration and info before continue:

- Qlik Replicate version:
Production server: November 2023
Development server: May 2024 (upgraded in June 3rd)
- Oracle:
Database version: 12.1.0.1.0
Compatibility version: 12.1.0.0.0
- Time zone mentioned in this post: GMT+7

On July 3rd, we noticed our Oracle to Log Stream CDC task had taken very long to start and couldn't be fully run before being stopped by the scheduler. On the same day and the next day, I tried to: 1. start the CDC task several times, then 2. do the full-load tasks to get data straight to the Snowflake target; the CDC tasks just keep starting for forever (see attached log file A), same as the full-load task but I saw it did load faster overtime. In afternoon of July 4th, the full-load tasks were able to load small tables, whereas transaction tables with ~1 million records were still stuck in a failed loading loop (see attached log file B). 

The same behavior occurs in both Prod and Dev replication servers; the only difference is that the dev server has the following warning added since Dev server was upgraded:

[ ASSERTION ]W: The Oracle Redo compatibility version 12.1.0.0.0 (0c100000) is not supported (oradcdc_redo.c:939)

On July 5th, the full-load tasks had run successfully, so I did reload target the CDC task, and everything has been back to normal since then.

We have been investigating the cause since the error occurred with our DBA and Qlik Replicate administrators, confirming that no action was taken on Oracle and Replicate servers at the time. I created this post with the purpose of finding out if there are any other reasons that could lead to this problem and the possible impact if we keep this Oracle version in the near future.

Thanks in advance!

[john_wang]

Hello @Ananguyen34 ,

Welcome to Qlik Community forum and thanks for reaching out here!

Please take note the old versions Oracle are suppressed.

Qlik Replicate 2024.5 support Oracle 19.x and 21c.

Qlik Replicate 2023.11 support Oracle 12.x, 18.x, 19.x, and 21c.

Hope this helps.

John.

[john_wang]

Hello @Ananguyen34 ,

I checked the 2 task log files:

A: Replicate 2023.11.0.159 + Oracle client 21.3.0.0.0 + Oracle server 12.1.0.1.0, no error no warning, so far so good.

B: Replicate 2023.11.0.159 + Oracle source + Snowflake target

    There are 2 errors:

    (1) target side error: Session no longer exists, please fix it by article How to set the CLIENT_SESSION_KEEP_ALIVE.

    (2) source side error: ORA-12161: TNS:internal error. This is a pure Oracle client and server communication issue, please work with your DBA try to fix it.

Hope this helps.

John.

[SushilKumar]

Hello @Ananguyen34 

The reason behind the Error may be related to post upgrade process. [ ASSERTION ]W: The Oracle Redo compatibility version 12.1.0.0.0 (0c100000) is not supported (oradcdc_redo.c:939)

In Normal scenario compatible parameter is changed post getting signoff from the application teams. as post Compatible change even Flashback did not work. 

so, the Archive logs generated during the whole upgrade process have both redo previous to upgrade and post upgrade 

Regards,

Sushil Kumar