This content has been marked as final. Show 1 reply
ca107207 wrote:I can't say with 100% surety that our apply instance has always been 1 and we just happened to always be on 1, but it seems that way.
Oracle RAC 18.104.22.168 with DataGuard with DG Broker on 2-node clusters
To start the apply process, we use sqlplus and the command:
alter database recover managed standby database using current logfile disconnect ;
To stop the apply process, we use:
ALTER DATABASE RECOVER MANAGED STANDBY DATABASE CANCEL;
We usually these things from instance 1. Recently we notice that when the apply process is instance 2, we must cancel the apply process while sitting on node 2.
Now we are thinking that perhaps we should be setting PreferrredApplyInstance in dgmgrl, to start we should be doing an edit database 'db' set state='APPLY-ON' with apply instance='instance1', and the reverse to cancel.You are using DG Broker, Oracle Recommend do stop and start Apply process with DGMGRL.
Does anyone have any insights into the pros and cons of using dgmgrl or sqlplus?
PreferredApplyInstance: name of the instance you prefer to be the apply instance
DGMGRL> EDIT DATABASE "DB" SET STATE ='APPLY-ON' WITH APPLY INSTANCE ='DB1' ; --- Start Apply process on Instance 1 with name DB1 DGMGRL> EDIT DATABASE "DB" SET STATE ='APPLY-OFF' WITH APPLY INSTANCE ='DB2' ; --- Stop Apply process on Instance 2 with name DB2
ApplyInstanceTimeout: time period that Broker waits after detecting apply instance failure and before initiating the failover;
default apply instance failure and before initiating the failover; default = 0
How the Broker chooses the new apply instance
If PreferredAppylInstance indicates an instance that is currently running, Broker selects it as the new apply instance
Otherwise Broker picks a random instance that is currently running
Mahir M. Quluzade