Broker(dgmgrl) convert snapshot standby in RAC - Failed to convert database

User_A9J4H

Hi,

I have a 19c Physical Standby DB (let us call it StdDB) in a RAC (19C) with two instances.

What I usually do to export data with PITR on standby with EXPDP, is to stop REDO (apply-of) im Broker(DGMGRL) and then I do FLASHBACK to an early point in time. This works fine.

Then I want to convert to SNAPSHOT Standby (to be able to perform EXPDP since the DB is then OPEN) but in the process I get the error:

DGMGRL> convert database 'STDBY_DR' to snapshot standby;

Failed to convert database "STDBY_DR"

Then I get in Broker: Warning: ORA-16782: instance not open for read and write access

Also I can confirm in Alert.log ORA-00600 later on , after these steps are over:

CONVERT TO SNAPSHOT STANDBY: Complete - Database mounted as snapshot standby

ALTER DATABASE OPEN

Then I check, there are TWO instances in RAC. The "apply" instance number 2 is opened in READ/WRITE mode but the instasnce number 1 is still in MOUNT mode.

When I login with SQLPLUS / as sysdba on the Instance 1 DB Node, I do ALTER DATABASE OPEN, and then after a white the ERROR in DGMGRL DISAPPEARS (ORA-16782) and I see succes.

Ist this normal behaviour for DG not to be able to OPEN both Dataguard instances while converting to Snapshot standby? I guess I could of perform EXPDP on the other instance 2 (which were in OPEN mode) , but I just want to narrow down, if this is an issue to look into, or normal behaviour.

When I take a loot at the documentation, I see that it should say CONVERTED succesfully, but it does not in my case (worked before).

https://docs.oracle.com/en/database/oracle/oracle-database/12.2/dgbkr/examples-using-data-guard-broker-DGMGRL-utility.html#GUID-05313396-B80D-4F53-BB44-AC7B0FF5E8B4

Thanks in advance.

User_A9J4H

Hi. Since the error happened after the CTWR process wanted to start on the one instance, I disabled the block change tracking and then it worked fine.

User_A9J4H

Hi. Since the error happened after the CTWR process wanted to start on the one instance, I disabled the block change tracking and then it worked fine.