From: Ethan Post <>
Date: Tue, 23 Dec 2008 09:20:28 -0600
Message-ID: <>

Thanks to those who provided some answers. I think I figured out the problem. I was able to write a procedure and recreate the issue. It appears oracle user was added to group "x" after the database was started. Membership in "x" is required to read the file. When I start a sql*plus session and execute the job this is a new process, when the scheduler starts a session to run the job it is a child process of the oracle server. I believe that the existing unix processes that were started before adding oracle to group "x" are not able to read the file. We are going to bounce the DB's tonight and I think that will solve the issue.

On Mon, Dec 22, 2008 at 1:06 PM, Ethan Post <> wrote:

> Seeing something a bit odd. I have a series of jobs which use DBMS_JOB,
> just migrated to 10G. The jobs kick off and attempt to run every 10 minutes
> until they are table to complete. There are some dependencies which they
> wait on. Trouble is that they are logging ORA-29283. However, if I log in as
> the user who owns the job and run DBMS_JOB.RUN to run the job they finish
> fine and report no error.

Received on Tue Dec 23 2008 - 09:20:28 CST

Original text of this message