We're running Oracle on RHEL 5.4 64-bit. We recently did an upgrade from 10.2.0.1 to 10.2.0.4. Many errors were generated during the upgrade (sample listed below from trace.log) but during application testing afterward everything seemed fine (clean EXP, inserts, updates, deletes, etc.). The errors look like they are all related to Advanced Queuing tables and views. We are not using replication at all, this is a simple single instance db.
ORA-24002: QUEUE_TABLE SYS.AQ_EVENT_TABLE does not exist
ORA-24032: object AQ$_AQ_SRVNTFN_TABLE_T exists, index could not be created
ORA-24032: object AQ$_ALERT_QT_S exists, index could not be created for queue
ORA-06512: at "SYS.DBMS_AQADM_SYSCALLS", line 117
ORA-06512: at "SYS.DBMS_AQADM_SYS", line 5116
Is this worth worrying about, and if so, how do I go about cleaning up/recreating the corrupted and/or missing objects?
My suggestion is to not ignore it.
I recently had AQ related issues after changing from Interpreted PL/SQL to Native Compiled PL/SQL. My AQ tables where munged and we had some data dictionary corruption. We do not explicitly use any AQ features for our products, but it looks like Oracle does use it for some of their feaures.
The main issue for us is that we were not able to use DataPump exports as they bombed out with AQ related errors. Also OEM access to this one database was very flaky. All user operations seemed to work OK.
If you check your alert log you may be getting some weird ORA-600's due to AQ subsystem not being available.
My suggestion is to open an SR with Oracle and go from there. They have an unpublished procedure they can give you that drops and recreates the AQ tables. If you don't have any data dictionary corruption it should be something simple... but make sure you have ability to absorb some downtime as their procedure can not be run while users are logged on. Also make sure you have some good backups - we ended up having to roll back 1 week because of all this (luckily just a Dev database).