There are cases where log pages get locked by a transaction that is not
completing and the log tail (the log is a circular buffer) cannot be advanced
past that page.  When the database backup is finished, it tries to advance the
log tail to the point in the log where the database ended, but if there's a
locked page it can't; the % util of the log remains above the dbbackuptrigger
and immediately another DB backup starts.

Recycling the ADSM server at that point rolls back the transaction and unlocks
the page; the log tail jumps forward to its proper position.

Cheers...
Bruce

Richard Sims wrote:

> >Time after time, a database backup was triggered almost immediately after
> >the last one completed. This happened 12 times before the system (probably)
> >ran out of scratch tapes:
> >Why does this happen?
>
> The Recovery Log having reached the DBBackuptrigger value, especially
> repeatedly, is obviously a symptom of a LOT of activity.  What sessions
> were active?  What was going on?  Look at your Activity Log and
> accounting records for client activity.
>     Richard Sims, BU

--
Bruce Elrick, Ph.D.
Bruce Elrick, Ph.D.
mailto:belrick AT home DOT com
http://members.home.net/belrick/