Hmmmm curious this ...

The ORA error messages posted above were captured by the Application and recorded in an Error_log table in the schema. Hey - a developer doing something sensible! However, I can't find any ORA errors at all in the alert log. This database (call it 'c') links to DB 'b' and DB 'a'.

A - is the master database.
B - has a DB link and uses Views to replicate data at B based on data at A.
C - has a DB link and uses Views to replicate data at C based on data at B. Some queries against C will pull data from A via B via these views.

Make sense?

I can't find any ORA-04031 or ORA-00604 errors in the alert logs of DBs A, B or C.

So where can they be coming from. The Developer doesn't know - he says he just catches them! Is it possible that somehow one of these DBs has a messaging level that doesn't allow the messages to appear in the alert logs? (I very much doubt this).

Or can the messages be written somewhere else than the udump or udump destinations?

To further complicate things:
DB A is Ora 9.2.0.4

DB B is Ora 9.2.0.5
DB C is Ora 8.1.7

A is on a Windows 2003 server.

B and C share an NT4 server, and the contractors installed multiple Oracle homes and migrated each Db up from 8.1.5 in order to get some of the functionality between B and C working. Why they stopped upgrading C at 8.1.7 I don't know.

The errors are timed at 20 minute intervals. At almost the exact time as the system appears to archive redo logs (10Mb logs - the original DB had 1Mb logs and were sufficient for this fairly small system)


Anyone have any clues??????