OGGBD OGG-03533 - Conversion Error From CESU-8 To UTF-8 For A Special Character (Doc ID 2803812.1)

Last updated on APRIL 17, 2023

Applies to:

Goal

Setup:
1. Oracle downstream mining DB 11.x
2. Oracle GoldenGate 19.x (191004) - extract and pump
3. Oracle GoldenGate for Big Data (OGG_BigData_Linux_x64_19.1.0.0.5) - Java based replicat that consumes the stream of data

Steps to reproduce:
1. We encountered OGG-03533 on one of our replicats during CDC (change data capture) which caused the replicat to go in ABENDED state. Unfortunately we did not capture the data that caused the issue (at this point it might be still retrieved but the timeline is not clear).

Workaround:
REPLACEBADCHAR with ESCAPE option

Experiments:
We tried inserting part of a surrogate pair (e.g. '\uD83C') that caused OGG-03533 issue. We observed that if we used the NOCHARSETCONVERSION, then the issue would happen in the Oracle Golden Gate for Big Data Java code, though this code doesn't seem to allow any pluggable functionality that could have let us implement certain behavior to handle the conversion issue. We also observed that the '\uD83C' character was extracted without any issues using JDBC.

Possible solutions we are looking for:
1. If OGG-03533 might be caused by a bug - assuming that a patch or a newer version of software would solve the issue.
2. Alternative options to handle the conversion issue
 

Solution

My Oracle Support provides customers with access to over a million knowledge articles and a vibrant support community of peers and Oracle experts.