Assume we have a service with preferred node as 1 and available node as 2.
Scenario 1:
We are shutting down the instance 1 manually what will happen to the service? Will it failover to available node?
No it won’t and if we query the status of the service it will report that it is not running.
Always we have to relocate service if it is running on single node. More than one node no issues.
or
In 11.2, you need to specify -f option if you want service failover (start in another instance)) to available instance (ie, srvctl stop instance -d xxx -i xxx1 -f) when stopping instance using srvctl and the service is running on only one node.
Scenario 2:
We are killing pmon for a instance to test or node eviction happens due to contention or high memory swapping and node 1 evicts. What will happen to the service? Will it failover?
Yes service failover happens only when the instance or node evicts based on the service configuration. Always ensure service is configured in more than one node.
While creating a service it creates the service as below
Management policy: AUTOMATIC which means it will start the service when it comes up if we stop and start the instance manually.
we have a service which is running on both the nodes as preferred. When I bring down the instance 1 service will run on node 2. After patching or node eviction if we bring up the instance or node 1 up the service will start running on both the nodes 1 & 2.
Any suggestion or corrections to this post are welcome.
Also would like to know how cluster decides whether to fail over or to start back a service? where can i get more readings on this?