Tags

, , , ,

You know there’s a cluster and that means that an HA system exists, but OEM won’t discover it.  Of course that means that your RAC databases can’t be discovered either and all of this is making you cranky.  Take a deep breath.  We can fix this.

The agent can only recognize an HA system when it knows of the clusterware home.  For some reason the agents weren’t aware of the CRS_HOME on some of our Solaris hosts.

Here’s my solution.

  1. Determine the CRS_HOME with a quick ps -ef | grep lsnr
  2. Add that home to your environment settings with export CRS_HOME=/my/path/to/crshome
  3. Stop and start the agent to pick up the new information.

The agent will discover the clusterware home and report it back to the the management servers.

Run through your typical process in Add Targets Manually to save the HA system as a monitored target.

add_ha_service

Come right back to Add Targets Manually to discover, configure, and add the cluster database using the Guided Process.

In some cases the Guided Process didn’t discover any database targets so I  ran through the dreaded Add Targets Declaratively by Specifying Target Monitoring Properties.  Web page titles like that are good indicators of their unpleasantness.  This one is no exception.

 

add_ha_service2

I found that configuration on that page is quicker if you limit yourself to just the RAC database and one of its instances.  Once that pair is saved, open the Monitoring Configuration page for your newly discovered RAC database and hit the Sync Cluster Database Instances button.  OEM will add the other instances to the list with no typing on your part.  Test the updated connections and save your work.

Reward yourself – you just earned it.