Other: HVR Adapt enhancement
The hvradapt command line program is a great tool for staying on top of new tables, changes to tables, dropped tables, etc. We recently discovered an issue (using HVR with Postgresql) where a table was dropped and recreated, resulting in a new OID in Postgres. Replication stopped on the table because it was still enrolled with the original table OID in its channel. However, HVR didn't complain about this. Since the enrollment file is binary, there's not an easy way to discover such discrepancies, especially in big databases with lots of DML activity. It would be really great and very much appreciated if hvradapt could verify OID when it runs and report discrepancies in its output - even better if it could correct and re-enroll tables in these cases.
Thanks!
-
Official comment
Hi Cary,
Please note that HVR does not support AdaptDDL on PostgreSQL (https://fivetran.com/docs/local-data-processing/capabilities/capabilities-for-postgresql#capture). I.e. HVR does not take any action when there is a DDL operation in PostgreSQL.
We do however consider creating a report to validate enrollment information with current information from the data dictionary. This is tracked as ticket T-320941.
FYI you can look into the enrollment file using the hvrrouterview utility (https://fivetran.com/docs/local-data-processing/command-line-interface/command-reference/hvrrouterview). E.g.:
hvrrouterview a b <file name>.enroll
Hope this helps.
Mark.
Please sign in to leave a comment.
Comments
1 comment