Oracle Bug on 6000/55 ?

We are experiencing problems with Oracle V6.0.0.27 with TPO on a Unisys 6000/55.

The symptoms are that the program invoking Oracle clocks up some cpu time, then remains static while the Oracle shadow process continues to accumulate cpu time.

This has been observed on both a C program and an import. In the case of an import, all rows from certain tables were deleted prior to the import starting, and a full file import was begun from an export file containing only those tables. The import hung on the first table (approx 2000 rows) and the shadow process continued to clock up time for 8 hours. Similar symptoms have been experienced with the C program. This also produces printed output, so that the time of last write to the output file, prior to the problem occurring, can be ascertained.

No unusual events are recorded in the Oracle trace files, and in most cases where this has occurred, Oracle has been shutdown for the backup and then restarted just prior to these jobs being run. These jobs are also run with no other users on the system.

Nothing suspicious can be seen in the Oracle monitor and there is sufficient rollback space (120 Mbytes).

Local Oracle technical support are baffled and this is severely impacting production (salaries).

Any help or ways to solve the problem would be gratefully appreciated. Thanks in advance.....
Post replies or email to address below.

Ray Bartho                              Phone    : (031) 816-2468
I.T. Consultant                         FAX      : (031) 816-2125
University of Natal                     Internet : bartho_at_cc.und.ac.za