Loosing data with Data Guard
I hope somebody can help us here.
We have Active Data Guard running in Maximum availability mode with sync transport and real-time redo apply to a physical database.
We are using 11.1.0.6 as that is our current production version. Will be moving to 11.2.0.1 soon
OS is OEL 5 u 4
We have a 10GB network between 2 Dell 610 blade servers.
We did some test where we have a script inserting into a table on the primary database and will commit after every insert, this script sits in a loop just inserting all the time.
We can see the data on the standby as while the script is running we will do a select count on the standby side and the count will increase telling us that the redo is being applied on the standby side.
We then pull the power from the primary server simulating a disaster on the primary database side. We then do a select count on the standby side.
We then restart the primary and do a select count from the table. We always loose 1 record. The standby will then also sync and have the correct count after
the primary has restarted. We got the same results running max protection mode as well.
So my question is this. Why do we lose records on the standby side as Oracle should not return the commit if the redo has not been shipped to the other side. This is what the documentation tells us. We also made the NetTimeout attribute 720 and then 10 with the same results.
We are loosing data on the standby side no matter what mode we are in. Can sombody who knows their Data Gaurd is set up correctley do such a test and tell me if you loose any data.
Regards.
Jaco