Video Screencast Help

Oracle Dataguard Broker Agent.

Created: 02 Aug 2011 • Updated: 26 Aug 2011 | 3 comments
This issue has been solved. See solution.

Hi All

Just wanting some more information surrounding the Oracle Dataguard Broker Agent for Oracle RAC.  I'm reading page 16/17/18 of the VCS agent for Oracle Dataguard Installation and configuration guide.

We've 2 SFRAC clusters comprising 2 servers each.  On both, the database stays up all the time with dataguard replicating from one side to the other.  But how do i setup dataguard agent broker to monitor replication between the two and how does it handle failover? If both databases are up all the time how does the dataguard agent decide which one is the active and which is the standby? I've put it in a global service group so assume its achieved via that?

At the moment I've got two parrellel service groups on each cluster - the first is a local service group which has the diskgroups, mounts and oracle database reources.  The second has the Dataguard agent broker and is a parrellel global service group which is its parent.

Thanks

 

Comments 3 CommentsJump to latest comment

jdangelo_symc's picture

To properly setup the Data Guard Broker agent you will need to have the following configuration:

DGMGRL will need to be configured as the means to manage/configure DataGuard

A local parallel service group for the Storage and Oracle Resources The Oracle agent (Should you chose to use it) will also have to configured to use the following: 

StartUpOpt = SRVCTLSTART

Additionally, CRS will need to be configured to start the Database in "Mount" mode.

The Network Listener will have to be managed by CRS

This will allow for the OraDG agent to properly promote and demote each instance from Primary to Physical Standby (and vice versa).

The service group for the Oracle DataGuard Agent will have to be Global/Parallel and have an online-local-soft dependency on the Oracle Service group. Make sure that the SG name and resource name are identical across sites.

 https://sort.symantec.com/agents/detail/2373

Take a look at OraDG guide for more details.

 Hope this helps,

Joe D

SOLUTION
f1shboy's picture

Ah, Thanks Joe D - definitley a bit clearer.

So I'm assuming the following, but just want to check:

2 clusters - each with a local service group with all oracle agents in - they're online. then global service group for the oracle dataguard broker. Thats online at the Primary site only and offline at the Standby site.

Suppose I want to switch sites:

All I do is offline the dataguard broker agent service group, then online it at the other cluster.  The onlining action of the dataguard broker agent switches the Standby database to Primary and consequently, the cluster where the service group was online becomes standby?

 

 

 

jdangelo_symc's picture

You are correct.  The Oracle Data Guard service group is the only one that fails over.  Please keep in mind however that there are some additional considerations here.  When you initiated a "Switchover" the DG/DGBRoker agent will first promote your Physical Standby to Primary. VCS will then issue the DemotePri action to convert the original primary to a Physical standby.  This is done by recycling the database.  VCS will offline/on-line the Oracle agent. I would tail the alert logs in Oracle so that you can see the status changes within each database.  Also, you will want to make sure you have cluster authentication between each site.  An easy way to check on this is via the action command (VCS Gui) against the DataGuard agent.  You will want to select the DG status action and from there you will be able to determine whether or not the Database is a state that would  be "switchover Allowed"

Joe D