Re: Materialized View Question

jkstill_at_gmail.com wrote:
> A MV uses an internal table MLOG_$<source_table_name> to log changes.
>
> This table is populated by an internal trigger when changes are made
> to the source table.
>
> This table is in turn used to populate the target table. That is done
> when a dbms_mview.refresh is requested, either manually or via
> dbms_job.
>
> When applying archive logs in a database recovery, the database is not
> open.
>
> The source table will change, but I do not believe the MLOG_$ table
> will be updated (database is not open). The changes will not then
> propogate to the target table when a refresh is requested.
>
> I would think that a complete refresh would be required in this case.
>
> The caveat is that I have never tried this particular scenario.
>
> Perhaps you should set up a test to see exactly what happens.
>
> HTH
>
> Jared
>

But in your opinion changes to MLOG_$ are not logged too? Remember that the redolog logs *every* change made to the database.
If your theory were correct, then every trigger defined on a table would not work and so why would someone trust such a standby database?

Umberto Received on Tue Sep 13 2005 - 05:13:58 CDT