Qlik Replicate - Task stops abnormally when reading online redo log from Oracle source

[NakulanR]

Hi,

We are seeing an issue with 2 Replicate tasks in which the task stops abnormally. The task has an Oracle source and a Snowflake target. The Replicate log files always have the line "Start processing [online/archived] Redo log sequence xxxxx ...." as the last line of the log file before the task stops abnormally. It then retires 5 times before coming to a complete stop. We were able to determine that the task was failing every time it reached the same log sequence.

This first occurred after an Oracle quarterly patch and the issue was able to be mitigated by restarting the task from a timestamp such that the particular log sequence that was appearing at the end of the Replicate logs was being missed. Oracle DBA's were able to confirm that the log was still available in the Oracle source system.

The Replicate logs don't indicate any warnings nor do they say the redo log file is missing/corrupted, the task just fails once it starts reading that particular log sequence. Enabling Trace and Verbose on Source capture didn't provide any further information on what caused the error.

What we'd like to understand is what potential issues with the particular Oracle log sequence could cause Replicate to provide no further information before it crashes. Is there any other place we should be looking in order to diagnose the issue, aside from the task logs? And just wondering if you have seen this issue before.

*Note: The Replicate instance is November 2022 (V2022.11.0.394), and the Oracle patch in question is Oracle Critical Patch update - January 2024

Any assistance/insight is greatly appreciated.

Thanks,

Nak

[john_wang · Accepted solution]

Hello Nak, @NakulanR ,

Thank you so much for the detailed information. The current running Replicate version is V2022.11.0.394 (SP02). There are several Replicate patches which are related to Oracle Patch installation especially about " Oracle 19c July CPU Patch (19.20.0.0.0)" and "Oracle 19c Oct 2023 patch". Not very sure what's the Oracle version you are running now.

However please take note that upgrading Replicate to 2023.11 latest patch is required as it contains the above fixes.

Looks to me the task crash after the Oracle patch, it maybe caused by some special DDLs (which were generated during the Oracle Patch installation). If you want to dig it deeper:

1. Check Windows Event Viewer to see if any clue

2. If any clue in Oracle alert log

3. Compare between Oracle LogMiner and Oracle LogReader access mode, if Logminer is supported

4. How about if target endpoint replaced by NULL Device, to make sure the crash is target endpoint irrelevant

5. Run Replicate 2023.11 SP02 and 2022.11 in parallel for comparison (with NULL Device target)

Hope this helps.
John.

[NakulanR]

Thanks @john_wang for the feedback.

We will try some of the points you mentioned to dig deeper, and we will also look at the possibility of the upgrade. Regarding the fixes in subsequent releases of Replicate, are there any particular Jira/Salesforce IDs that we can refer to in the release notes which are relevant to this issue?

Regards,

Nak

[john_wang]

Hello @NakulanR ,

Yes, some of the are:

Oracle 19c Oct 2023 patch

Oracle 19c July CPU Patch (19.20.0.0.0)

Thank you so much for your great support!

John.