In summary: AWS DMS Oracle to Oracle
Source database on prem : Oracle 19c, 1 table size 597Million records
Target database RDS : Oracle, 1 table fails at the 207Million mark
Scenario: multiple failures after loading 207M plus records, the task fails with :
2023-06-13T23:07:59 [SOURCE_UNLOAD ]I: Sent unloaded record 216414002 to internal queue {operation:LOAD_DATA (33), tableName:ISPCUSTOMERCHANGE, schemaName:ISPPOWER, connectionId:1031} (streamcomponent.c:2928)
2023-06-13T23:09:10 [SOURCE_UNLOAD ]W: Oracle error code is '29275' ORA-29275: partial multibyte character (oracle_endpoint_unload.c:241)
I upgraded DMS repl instance to 3.5 beta as this reported a fix on multibyte issues - No success
I've tested the following:
- remove records around the max id inserted into the target from the source
- I've copied the 'offending records' into another table and these load successfully
In a nut shell I've not been able to identify the source of records causing the multibye error
Questions / ponderings:
- Is it to do with the volume of records ?
- Any suggestions in identifying the records on source ? e.g. wide table 30 plus columns
- Is this a DMS bug. I upgraded to 3.5beta as I read there was a potential fix
Any suggestions / help appreciated.
Thanks.
Thanks for your time reviewing and reply. I'm performing further testing e.g.
This shouldn't be the case as I could select into a table no issue. So I'm digging deeper enabling all logging on at DMS task level to get and glean more information