Level 0 Archive Required

Hi IIUG Folks, We started experiencing an On-BAR issue after upgrading from IDS 9.40.UC4 to 9.40.UC9. We upgraded four instances on one server and four on another server. Both servers are Sun boxes and are running Solaris 8. We use BMC's SQL-BackTrack (now known as Recovery Manager for Informix), version 3.7.00, as our storage manager. Three instances on each box must back up directly to a Legato tape system because we cannot get enough disk space to backup to local disk. This has been our scenario for quite a few years (as our IDS version indicates). Before anyone suggests it - I have tried in vain to upgrade for many years. We run a Level 0 each weekend and then a Level 1 every other night. After the upgrade, some of our Level 1 archives required a Level 0 to be performed - even though I ran a fake Level 0. That is not too bad and really should have been expected. However, the largest instance (1.2 TB) insists on performing a Level 0 every night - even though we've had successful Level 0's. Due to the fact that it has to back up directly to the tape system and that it is competing with other Legato backup jobs, a Level 0 can take anywhere from 8 hours to over 24 hours. The error appearing in the bar_act.log file is: (-43185) A change in physical configuration has occurred. A level 0 backup of ndw_db11 is being performed. I thought perhaps it required a Level 0 to be explicitly requested, rather than a forced Level 0. However, after a Level 0 was explicitly requested and successfully run on the weekend following the upgrade, the instance still insists on doing a Level 0 every night. In an earlier post, I saw where John Miller suggested running "select sum( sysmaster:bitval(flags,'0x20000') ) as level_0_required from sysmaster:sysdbstab" to see what needs the Level 0. The result is 0. I also saw a post in the same chain where Art Kagel said if the transaction activity is high enough, the engine will determine that a Level 0 would be faster than anything less. No known increase in transaction activity has occurred. There has been no change to the application code, nor has there been any structure changes. The only recent change was to upgrade from 9.40.UC4 to 9.40.UC9. The upgrade was required to get around a bug causing one of the instances to crash. The bug was fixed in 9.40.UC5, but we decided to get as high on the 9.40 tree as we could get. Does anyone have any suggestions as to how we can a Level 1 to run? Thanks in advance Rob Schmitz