Great answer - a little more detail for the alert/metric
Carla Bell, January 16, 2017 - 3:20 am UTC
Thank you Connor, this is great information. However, I'm not totally undersatnding the recommendation for the alert/metric.
Can you explain in more detail performing a count(*) for the alert/metric?
Thank you very much.
January 17, 2017 - 1:17 am UTC
In Enterprise Manager, you can define your own metric/alerts.
Just like (for exampe) the default tablespace usage metric is "more than 85% = raise an alert", you can define your own metrics, which is defined by your own SQL statement.
So you could have a metric:
"RMAN excessive duration"
defined by (say):
select count(*) from v$rman_status where status like 'RUNNING%' and start_time < sysdate - 1/24;
with a threshold set to: "> 0" = raise alert