Joan, You will still want to check with support to see if there are any recommendations about these issues. While you may have gotten a successful run a time or two with these adjustments, these inconsistencies may come back to bite you and your users at inopportune times. You need "plan stability", which may be achieved via a psoft patch, or an oracle outline or profile. I would not be satisfied nor settle for these inconsitent results. ----- Original Message ----- From: "Joan Hsieh" <joan.hsieh@xxxxxxxxx> To: "Michael Fontana" <michael.fontana@xxxxxxxxxxx> Cc: oracle-l@xxxxxxxxxxxxx Sent: Monday, November 23, 2009 1:45:17 PM GMT -06:00 US/Canada Central Subject: Re: ora-4030 error Mike, After I reduced the ulimit of psoft user, the job ran successfully via process scheduler. We ran twice to make sure it run. However, I ran the same sql via sqlplus, it failed again with 4030 error. I got explain which is differently from what I got from sqr (via ps). The explain plan I got for oem. I used tkprof to generate the plan for sqlplus. Another interesting thing is after tax upgrade and cpu patch, I immedailty refresh 2 testing database. One used the old file cp method to clone the database, the other one used rman hot backup to clone the database. Same sql ran successfully on the rman refreshed database. But run out of temp space on the old fashion cloned database. Of course, the execution plan are different from everywhere. prod (via ps and via sqlplus), test (via rman duplicate and cold file cp) Joan -- Michael Fontana Sr. Technical Consultant Enkitec M: 214.912.3709 enkitec oracle_certified_partner -- //www.freelists.org/webpage/oracle-l