Critical resource protection methods on AIX and Linux

If critical resources are active in an AIX® or Linux® subcluster that has lost quorum, the ConfigRM uses the "CritRsrcProtMethod" attribute of each node in the subcluster to determine in which way the system must be terminated. The protection methods are based on immediate system shut down by means of kernel panic simulation. There are seven protection methods:

Table 1. CritRsrcProtMethod settings
Meaning	Value
Hard reset and reboot operating system (default)	1
Halt operating system	2
Hard reset and reboot operating system with Sync	3
Halt with Sync	4
No protection. System continues operating	5
Exit and restart RSCT subsystems	6
Disable all protections	7

Examples:

To change the domain-wide critical resource protection method, enter chrsrc command using -c flag.
```
chrsrc -c IBM.PeerNode CritRsrcProtMethod=3
```
To override the default domain-wide critical resource protection method on a single node, enter chrsrc command. using -s flag and a selecting string to identify the node.
```
chrsrc -s"Name==’node1’" IBM.PeerNode CritRsrcProtMethod=3
```

A protection method with sync sends file system buffers to disc before the operating system is halted or reset. Thus the probability of data loss or data inconsistency on the file system is reduced. Protection methods with sync could be an unsafe solution in certain situations. For example, data can be accessed during a file system flush from another application constituent that has just started in the subcluster which gained operational quorum. You must determine if this situation is likely to occur in a given system and application environment.

Regardless of the protection method being used, you must use journaling file system. This can prevent the corruption of the file system itself and greatly enhances file system recovery after a system reset.

The protection methods on the nodes may be different. The same protection method is set for each node in the entire cluster.