Follow up
Ram Dittakavi, September 15, 2017 - 5:42 am UTC
Thanks Connor.
Can you please clarify my 1st question as well? How am I able to start the RO service in standby without starting it ever on primary in the first place?
Follow up
Ram Dittakavi, October 04, 2017 - 7:08 am UTC
Thanks Connor.
Can you please clarify my 1st question as well? How am I able to start the RO service in standby without starting it ever on primary in the first place?
October 06, 2017 - 5:38 am UTC
The issues was that adding a service to CRS did not execute the DBMS_SERVICE.CREATE_SERVICE package to put the service definition into the database's metadata until the service was started for the first time.
So as long as you started the service at least *once*, then it will be in data dictionary.
Follow up
Ram Dittakavi, October 12, 2017 - 6:29 am UTC
Hi Connor,
We cannot start a service on a database unless an entry is available in service$ ?
But, as I mentioned earlier, i was able to start the _RO service on standby even before it was started once on primary.
October 13, 2017 - 3:43 am UTC
But, as I mentioned earlier, i was able to start the _RO service on standby even before it was started once on primary.
I tried a few internal environments and I could not start a service on a RO standby (unless I had done it at least once at some stage in the past on the primary).
If you're seeing different behaviour, I think you'll need to log a call with Support with a full test case - because that should not be possible.
Follow up
Ram Dittakavi, October 19, 2017 - 6:13 am UTC
Thanks a lot, Connor.
Suppose, we have flashback logs pertaining to a guaranteed restore point (GRT) and then FRA gets 100% full.
Suppose, FRA is 2G.
Case1: GRT logs occupy 1.5G
Now, the DMLs are allowed after purging the logs that are there in the remaining 0.5G (ofcourse 200MB at a time) and this repeats?
Case2: GRT logs occupy 2G
Now, what happens?