About node eviction of Oracle RAC caused by heartbeat network (not instance eviction)

About node eviction of Oracle RAC caused by heartbeat network (not instance eviction)

Ask:
Suppose the following scenario: 4 nodes RAC, the heart jumper is the thousand m network switch, if the thousand m network switch power off, I want to know the CRS of the expulsion node algorithm is how?

Oracle Dalian GCS Reply:
If the heartbeat network is broken, Oracle's cluster will split into several small groups of itself, which we call cohort (you can find in Ocssd.log).
Oracle determines which subset of groups to leave based on the number of nodes that each cohort contains, and the basic principle is:
1. The number of nodes of their own group left, the node points less is expelled.
2. If the number of nodes for each subset group is the same, the node containing the minimum node number is retained.

That's why you see a cluster of two nodes, and if there is a problem with the heartbeat network, it must be the cause of Node 2 being restarted.

Other than that:
The expulsion of the node and the VF (which is what many people say voting disk) doesn't matter.
The main reason is that many people confuse instance expulsion at the database level with the definition of the cluster level, and the concept of expulsion.

So that's the wrong kind of understanding.

If the expulsion at the database instance level is indeed related to the control file, because the control file at the database level plays the role of VF

About node eviction of Oracle RAC caused by heartbeat network (not instance eviction)