IBM TotalStorage NAS 300, User Reference

Page: 56 / 200

those resources. Even though a disk group has a preferred owner, its resources
can run on the other node in the cluster following a failover. If you restart a cluster
node, resources that are preferentially owned by the restarted node switch to the
standby system when the cluster service detects that the node is operational, and
provided that the defined failover policy allows this to occur. If you have not defined
the node as the preferred owner for the resources, then they do not switch to the
standby system.
Note: You must reboot before you can see changes made to the cluster resource
balancing.
Setting up failover
The failover of resources under a disk group on a node enables users to continue
accessing the resources if the node goes down. Individual resources contained in a
group cannot be moved to the other node; rather, the group it is contained in is
moved. If a disk group contains a large number of resources and any one of those
resources fails, then the whole group will perform a failover operation according to
the group’s failover policy.
The setup of the failover policies is critical to data availability.
To set up the failover function:
1. Open the Properties panel for the disk group.
2. Select the Failover tab to set the Threshold for Disk Group Failure.
For example, if a network name fails, clustering services attempts to perform a
failover operation for the group 10 times within six hours, but if the resource
fails an eleventh time, the resource remains in a failed state and administrator
action is required to correct the failure.
3. Select the Failback tab to allow, or prevent, failback of the disk group to the
preferred owner, if defined.
In allowing failback of groups, there is a slight delay in the resources moving from
one node to the other. The group can also be instructed to allow failback when the
preferred node becomes available or to perform a failover operation during specific
off-peak usage hours.
Each resource under each disk group has individual resource properties. The
properties range from restart properties, polling intervals to check if a resource is
operational, to a timeout to return to an online state. The default settings for these
properties are selected from average conditions and moderate daily use.
Creating users
The creation of users is performed through normal procedures. Users do not need
to be created exclusively for use on the cluster resources. You must define
properties of the resources for users to access the resources within the domain
policies. All user-accessible cluster resources have the same properties as standard
Microsoft Windows resources, and should be set up following the same policies.
Note: If your storage will be accessed by UNIX or UNIX-based clients and servers,
continue with “Defining UNIX users and groups” on page 44. The NAS
Gateway 300 is on a Windows domain and inherits those Windows users,
eliminating the need to define local Windows users and groups. Also, shares
are created in the clustering setup.
42
NAS Gateway 300 User’s Reference