avoiding ora-1652 on temp tablespace

[Seley, Linda]

We recently released a new version of our software and suddenly we're blowing out our temp tablespace.  The previous version never complained at 1000m.  The new version is choking on 2500m.  I don't want to keep adding space, although that's the conventional wisdom.  I want to find the query or queries that are being pigs and get the developers to fix 'em!    By the time I get notification that there's been an ora-1652 I can't find much info in the system.  All of the currently running queries are small, we only have 20 users connected at any time, v$sort_usage is empty, and v$sort_segment shows one segment taking the full 2500m.  I also can't turn on tracing, these 20 users are persistent and trace files will fill up in no time.  Our entire database is VERY small, only about 60m!    Any advice on how to catch these queries?  The ora-1652 errors are occuring about a week apart so it isn't practical for me to personally babysit the database.    Thanks!   Linda