We had an issue recently where in an attempt to build fairly small Join Indexes on a fairly large table (2TB), the build would blow out the spool limit. even after increasing the spool limit to 5TB, we still had issues. We finaly gave in and rebooted the 4 node 5500 system running 13.10.04.01b. We then ran the exact same Join Index build just fine, and it got the spool it needed.
Our assumption was that even though the system said there was space available for spool, something was still holding the spool space captive. Does anyone have any insight to this problem and what might be the cause?