Due to my previous issue of not being able to create a simple SQL metric extension against a pluggable database , my workaround currently is to create a shell script that does the work.
it seems shell scripts attached to OEM metric extensions deploy as the oemagent user and as such, using other credentials for those cause problems.
So i have manually deployed the scripts to the target server (and can add these to source control and add to formal release process once i have the final one defined)
So now i have a script that i have deployed to a single node, defined the metric extension and all works fine.
Except when i test it on two hosts of a RAC node.
When i try to define a monitoring host credential for the hosts, again "computer says no".
it says "cannot set monitoring credential at instance"
But this is the unix host so each host has it's own credentials, not the same ones for both.
Now my aim is to run a SQL statement from the shell script (to workaround earlier issue) but seems i am stuck again.
I could already foresee a problem with this approach as both hosts would provide data from the same database but duplicates would be better than no monitoring.
So i guess i am looking for guidance on how to setup OEM13.2 for RAC environments.
If one node is down, the other should still provide metrics. If both nodes are up, only one should provide metrics (and alerts)
Any pointers on setting it up for DR (where the DR nodes are down most of the time but come up if the primaries failover would also be helpful.
At the moment, i assume if setting up at the virtual host level, then this would not be a problem but OEM discovery finds the actual host names
Is there a way to configure it with virtual hostnames so in a DR of the target servers, the monitoring still works ?
Thanks in advance