On 2017/12/07 16:19, Rich J wrote:
I had your exact problem, in addition to EM complaining about "blocking"
locks between PQ slaves. For LGWR, the problem happens when there is
over-committing, like doing 1M inserts with a commit between each row. I
could not force this problem using SQL/PLSQL, but I did create a short Perl
program which inserts rows from a significantly large table into a new one.
I used sys.wrm$_snapshot_details, that had ~1.5M rows in it. Running the
program triggered our alert.