RE: killing oracle processes
Date: Mon, 17 Aug 2009 13:18:44 -0400
Message-ID: <0684DA55864E404F8AD2E2EBDFD557DA033083D5_at_JAXMSG01.crowley.com>
We had a situation like that at a previous job. We created a user and gave it execute permission on a stored procedure. The Help Desk could then launch a .sql script that would take the username as a parameter and execute the procedure. Only the helpdesk knew the password.
Of course it could be modified to take an sid and serial# instead etc. depending on what the circumstances are between the developers and 'help desk'.
Alter user kill session where status != 'KILLED' ... etc. (pseudocode).
Joel Patterson
Database Administrator
904 727-2546
From: oracle-l-bounce_at_freelists.org
[mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Michael Schmitt
Sent: Monday, August 17, 2009 11:34 AM
To: 'oracle-l_at_freelists.org'
Subject: killing oracle processes
Hi All,
I had a quick question that I was hoping the list could help me out with. We have a group of developers who are requesting the ability to kill their own processes in the database (PRD/DEV/TST). For example, if a poorly written report gets kicked off, one of their jobs chooses a poor execution plan, or an OWB process gets left out there. The only reason they can really offer is that they do not have to wait for the DBA team to respond. I am trying to think of technical reasons why this would not work.
I can write a script to limit the process to be killed to their stuff, but something about this still makes me feel uneasy. Is there anything that I should worry about?
Any thoughts?
Thanks
-- http://www.freelists.org/webpage/oracle-lReceived on Mon Aug 17 2009 - 12:18:44 CDT