Dear All,

Thanks for taking the time to raise and comment on this issue.

I believe this problem could be easily avoided if we follow the required Oracle DB setup document https://confluence.atlassian.com/doc/database-setup-for-oracle-173821.html when installing Confluence
I decided to close this as won't fix.

Regards
Minh Tran
Confluence Bugmaster
Atlassian

Hi guys we just upgraded to v5.6.3 and have 22 matches when running the log scanner.

Any news on this problem?

Hi Rodrigo

This issue is currently on our backlog and we should be addressing it in the near future.

Regards
Steve Haffenden
Confluence Bugmaster
Atlassian

Logs from upgrade from 3.1.2 to 3.5.5

Faced the same error when upgrading from 3.1.2 to 3.5.5.

When upgrading a Confluence instance configured with Oracle with the following scenario, the upgrade process is stopped and an error is thrown on the log file (attached).

• table (e.g. cwd_user) already exists on the same database but belongs to another user (e.g. jira)
• the confluence user on the database as grant to select any table.

It's been my experience with other projects using Hibernate and Oracle that this behavior is not necessarily a bug - it's caused by the Oracle user having access to more than one schema. For example, if the Confluence user has 'select any table' privileges, and there is more than one table on the database with the same name, and Hibernate does not explicitly specify the schema to query, then you'll get a namespace collision. You can work around this by setting the hibernate.default_schema property.

Alternatively, an error message might look like:

execute ORA-01430: column being added already exists in table

This appears somewhat infrequently, perhaps because the documentation on Oracle installation includes a warning, or perhaps because the chance of it happening are more or less rare. Nonetheless it does continue to occur, and when it does the error might look like:

(HibernateTemplate): data integrity violated by SQL ''; nested exception is java.sql.BatchUpdateException: ORA-00001: unique constraint (CONFLUENCE.SYS_C003601) violated

Based on what I have read on this issue and my knowledge of Oracle, this should not be an issue if the oracle account(schema) only has permissions to objects in it's own schema (a good security practice for this type of account anyway.)

Can anyone confirm this?

I am going to be trying next week, I try to remember to post my results.

A possible workaround:

I think I found a workaround. I created a view within the schema to filter the object queries. As long as hibernate doesn't access sys.all_objects table directly, it will see only what's owned by the schema.

CREATE VIEW wiki_dev.all_objects
AS
SELECT *
FROM sys.all_objects
WHERE owner = 'WIKI_DEV'
;

fixversion and affects version were incorrectly set

This issue may still be open. Here are the tables in 2.5.6:

I just tried confirmed that this is fixed in head. I created one Oracle user on a SID and that user created a CONTENT table. I then installed confluence using a different oracle user on the same SID with no difficulties.

We have been using the same hibernate jar throught 2.2 and 2.3.

Hi,

Does the latest Confluence Wiki still require a separate Oracle Instance for the configuration to Oracle database? Would a new Confluence Schema on an existing SID work? Assume the Confluence Wiki is going to be deployed on an existing Tomcat instance.

Thanks
-Perci Wong

It also appears that is is no longer a problem in 2.2.6a+ as a customer reported in his support issue. We will need to verify this and update the issue as well as the documentation.

Apparently this bug has been fixed. We will need to find out for which exact version of Hibernate and should consider upgrading.

Confluence 2.2.2 List of Tables Update: