12.5 Cluster-Enabling an Existing NSS Pool and Its Volumes

Cluster-enabling an existing pool allows it (and its volumes) to be moved or mounted on different servers in the cluster in a manner that supports transparent client reconnect. This makes the data highly available to users.

You might need to cluster enable an existing pool in at least the following situations:

You are converting stand-alone servers with SAN attached devices to a cluster.
You are migrating NSS pools between two separate clusters.

In this scenario, you must create a Pool object and Volume objects in the new cluster context by using the Update eDirectory option in NSSMU or the Storage plug-in for iManager.
You are creating a cluster resource from a storage-based snapshot.

In this scenario, you must create a Pool object and Volume objects for the activated pool snapshot and its volumes by using the Update eDirectory option in NSSMU or the Storage plug-in for iManager.

When you cluster-enable the exiting pool, its pool and volume information will be added automatically to the load, unload, and monitor scripts for the pool cluster resource. In addition, the volume entry is automatically removed from the /etc/fstab so that the pool cluster resource load script controls when the volume is mounted.

Before you cluster-enable the pool, we recommend that you deactivate it. The Online Resource after Creation option and pool state on the node when you begin determines the resource state and pool state after the resource is created, as described in the following table:

Pool State on the Node	Online Resource After Creation Setting	Resource State after Creation	Pool State after the Resource Is Created
Deactive	Selected	Online	Active as a resource on a preferred node.
Deactive	Deselected	Offline	Deactive. Modify the resource settings and scripts as needed, then bring the resource online on a preferred node.
Active	Deselected	Offline	Active locally on the same node, even if the node is a preferred node. Modify the resource settings and scripts as needed, deactivate the pool locally, then bring the resource online on a preferred node.
Active	Selected	Comatose	Active locally on the same node, even if the node is a preferred node. Take the comatose resource offline. Modify the resource settings and scripts as needed, deactivate the pool locally, then bring the resource online on a preferred node.

The default name of a pool cluster resource is the pool name plus the word SERVER, such as POOL1_SERVER. You can modify the resource name after the resource has been created by using the cluster rename command. See Section 10.15, Renaming a Cluster Resource. Changing the resource name does not modify the pool name or the virtual server name.

After the pool cluster resource is brought online, users can access the cluster-enabled volume by using the IP address or virtual server name of its pool cluster resource. The default virtual server name of the resource is the cluster name plus the default cluster resource name, such as CLUSTER1-POOL1-SERVER. You can specify the name of the resource’s virtual server name when you cluster-enable the pool.

The procedure in this section describes how to enable clustering for an existing pool and its volumes. It assumes the following:

The pool contains at least one volume.
The pool name meets the naming conventions supported by OES Cluster Services. See Naming Conventions.

If the pool name contains special characters, rename the pool by using NSS management tools. See Renaming a Pool in the Storage Services File System (NSS) Administration Guide for Linux.
The pool and its volumes have Storage objects in the eDirectory tree where you are setting up the cluster.

If the Pool and Volume objects are missing, you can create (or re-create) them in eDirectory by using the Update eDirectory option in the Storage plug-in for NSS or in NSSMU. You can do this from any node in the cluster. See Updating eDirectory Pool Objects with NSSMU and iManager and Updating eDirectory Volume Objects using iManager in the Storage Services File System (NSS) Administration Guide for Linux.
The pool resides on a disk that has been enabled as Sharable for Clustering. If it is not shared, mark the device as shareable. You can use NSSMU, the nlvm share command, or the Storage > Devices task in OES iManager to change the device’s share state.
Each pool resource requires its own unique static IP address that is in the same IP subnet as the cluster IP address.
To add CIFS as an advertising protocol when you cluster-enable an existing pool, you must install and configure OES CIFS on the server, and ensure that the CIFS daemon is running and functioning properly before you begin. The CIFS daemon should also be running before trying to online the resource.
If you use special settings for mounting the volume, you must manually add those settings to the load script before you bring the resource online.

To cluster-enable an existing NSS pool:

Ensure that the SAN device is attached to all of the nodes in the cluster.
Log in to iManager as a cluster administrator user.
Deactivate the shared pool on the current server.
1. In Roles and Tasks, select Storage > Pools.
2. Browse to select the cluster where the pool is currently assigned.
3. In the Pools list, select the pool, then click Deactivate.
  
  Wait until the page refreshes and confirms that the pool is deactive.
Verify that the device is marked as shareable for clustering. If the pool uses multiple devices, you must verify the shared setting of each device.
1. In Roles and Tasks, select Storage > Devices.
2. Browse to select the cluster where the pool is currently assigned.
3. Select the device to view its Details page.
4. If the Shareable for Clustering check box is deselected, select the check box, click Apply, read the warning, then click Continue to enable sharing.
  
  Marking the device as shareable automatically modifies the Shared attribute on the eDirectory objects for the pools and volumes on the device.
  
  Wait a couple of minutes before continuing to allow the attribute change to synchronize in eDirectory. You can verify the change in the shared state of the pool and its volumes by returning to the Pools page and Volumes page.
5. Repeat this process for each device that is used by the pool.
In Roles and Tasks, select Clusters > My Clusters, select the cluster, then click Cluster Options.
On the Cluster Options page, click the New link in the Cluster Objects toolbar.
On the Resource Type page, specify Pool () as the resource type you want to create by clicking the Pool radio button.
At the bottom of the Resource Type page, click Next.

On the Cluster Pool Information page, specify the following information:

Parameter	Action
Pool Name	Specify the Pool Name by browsing to select the Pool object () of the pool you want to cluster-enable, such as servername_POOLC1_POOL. After the pool is cluster enabled, the Pool object name is changed to clustername_POOLC1_POOL, such as clus1_POOLC1_POOL.
Virtual Server Name	(Optional) Change the default name of the Virtual Server for the cluster resource. The default virtual server name for the resource is the cluster name plus the cluster resource name. The default name of a pool cluster resource is the pool name plus the word SERVER, such as POOL1_SERVER. For example, if the cluster name is clus1 and the pool cluster resource name is POOL1_SERVER, then the default virtual server name is CLUS1-POOL1-SERVER.
CIFS Server Name	If CIFS is enabled as an advertising protocol, specify the name of the CIFS virtual server that CIFS clients see when they browse the network. The name can be up to 15 characters, which is a restriction of the CIFS protocol. By default, the NCP virtual server name is suggested as the CIFS virtual server name. For more information about how the default name is determined, see Section 12.1.7, OES CIFS for Linux. You can use the suggested name or specify a custom name for the CIFS virtual server name. If OES CIFS is not installed and running, this field value is NOT_SUPPORTED.
IP Address	Specify an IP address for the pool cluster resource. Each pool cluster resource requires its own unique IP address. The IP address assigned to the pool remains assigned to the pool regardless of which server in the cluster is accessing the pool.
Advertising Protocols	Select the check boxes of the advertising protocols (CIFS, NCP) that you want to enable for data requests to this shared pool. NCP is required to support authenticated access to data via the OES Trustee model. Selecting a protocol causes commands to be added to the pool cluster resource’s load and unload scripts to activate the protocol for the resource. This lets you ensure that the cluster-enabled pool is highly available to users via the specified protocol. If the OES CIFS protocol is not installed and running, selecting the corresponding CIFS check box has no effect.
Online Resource after Creation	The check box is deselected by default. Select this option only if the pool is known to be currently in a deactive state; otherwise the resource goes comatose on creation because it will not automatically deactivate the pool from the local node.
Define Additional Properties	Select the Define Additional Properties check box. This allows you to configure the resource policies for the start, failover, and failback modes, and to configure the preferred nodes.

At the bottom of the Cluster Pool Information page, click Next.
On the Resource Policies page, configure the resource policies for the start, failover, and failback modes.

See Configuring the Start, Failover, and Failback Modes for Cluster Resources.
At the bottom of the Resource Polices page, click Next.
On the Resource Preferred Nodes page, assign the preferred nodes to use for the resource.

By default, a new resource is configured to run on all nodes, including future nodes. Add the nodes you do not want to use to the Unassigned list.

IMPORTANT:For Assigned nodes, ensure that you prepare the node for the services in the resource before you migrate or fail over the resource to it.

See Configuring Preferred Nodes and Node Failover Order for a Resource.
At the bottom of the Resource Preferred Nodes page, click Finish.

The pool cluster resource appears in the Cluster Objects list on the Cluster Options page, such as POOL1_SERVER.

The load, unload, and monitor scripts are automatically created with information about the pools and volumes. The volume entries are removed from the /etc/fstab file so that the resource load script can be used to control when the pools and volumes are mounted. The resource is offline.
Verify that the pool cluster resource was created and that it is offline.
1. In Roles and Tasks, select Clusters > My Clusters, select the cluster, then click Cluster Options.
2. View the pool cluster resource in the Cluster Objects list, such as POOLC1_SERVER.
3. Select Cluster Manager, then view the pool cluster resource state as Offline.

Verify the cluster resource scripts for the pool cluster resource, and modify them if desired.

In Roles and Tasks, select Clusters > My Clusters, select the cluster, then click Cluster Options.
Click the pool cluster resource link to open its Properties page.

Configure the following resource settings:

Property	Action
Monitoring	(Optional) Select the Monitoring tab, select Enable Resource Monitoring, and configure monitoring as described in Section 10.7, Enabling Monitoring and Configuring the Monitor Script.
Load script	Select the Scripts tab, select Load Script, then modify the script as described in Section 12.6, Configuring a Load Script for the Shared NSS Pool.
Unload Script	Select the Scripts tab, select Unload Script, then modify the unload script as described in Section 12.7, Configuring an Unload Script for the Shared NSS Pool.
Monitor Script	Select the Scripts tab, select Monitor Script, then modify the unload script as described in Section 10.7, Enabling Monitoring and Configuring the Monitor Script.

At the bottom of the page, click Apply or Ok to save your changes.

The changes are not applied until the next time the resource is brought online.

If you did not deactivate the pool before you began, manually deactivate the pool on its currently assigned node.

You must deactivate the pool from the server before attempting to bring the pool cluster resource online.
1. In Roles and Tasks, select Storage > Pools.
2. In the Server field, browse to select the Cluster object () of the cluster.
  
  The Cluster object is automatically associated with the node that is currently the master node in the cluster.
3. In the Pools list, select the pool, then click Deactivate.
  
  Wait until the page refreshes and confirms that the pool is deactive.
Bring the resource online:
1. In Roles and Tasks, select Clusters > My Clusters, select the cluster, then click Cluster Manager.
2. Select the check box next to the pool cluster resource, then click Online to bring the resource online.
  
  The pool is activated and its volumes are mounted on the primary preferred node that is configured for the pool cluster resource.
  
  If the pool is comatose, take the resource offline, then bring it online. If it goes comatose again, check the changes you made to the cluster scripts, then try again.