This section contains additional information that is important in working with Microsoft Failover Clustering and Storage Foundation for Windows.
Note the following considerations:
When a cluster disk group resource is offline or a cluster disk group that is not a failover cluster resource is in a Deported state, it is not protected from access by other computers. For maximum data protection, keep Volume Manager Disk Group resources online. Note that the SFW disk group resources still retain the "Volume Manager" name.
When using the Windows Server's Failover Cluster Manager snap-in to create a disk group resource, the Volume Manager Disk Group Parameters screen might not list all the available Storage Foundation for Windows cluster disk groups in the drop-down list. If this happens, exit the New Resource wizard and use the Windows Server's Failover Cluster Manager snap-in to select the cluster group to which the resource is to be assigned. Next, move the cluster group to the cluster node where the Storage Foundation for Windows cluster disk group is currently online. Then create the Storage Foundation for Windows disk group resource.
Under the following circumstances, the VEA Disk View may not reflect the latest state of the disk(s) until a refresh is performed:
SFW support of the Microsoft Failover Clustering environment allows the selection of SCSI-2 reservation mode or SCSI-3 reservation mode. Selecting the type of SCSI support for the Microsoft Failover Clustering environment is done by using the System Settings portion of the SFW Control Panel.
When selecting the type of SCSI support in a Microsoft Failover Clustering environment, it is important to know if your storage arrays support SCSI-3. SFW SCSI-3 clustering support does not let you mix the storage arrays that support SCSI-3 with the storage arrays that cannot. In a situation of mixed storage arrays, you must use SFW SCSI-2 clustering support. Refer to the HCL for arrays that support SCSI-3.
After selecting the type of SCSI support, you must issue the following CLI commands to complete the setting on your system:
net stop vxsvc
net start vxsvc
Note: |
If a cluster disk group is imported on the system, you must deport or move the cluster disk group to another system before issuing these CLI commands. |
If SFW SCSI-2 clustering support is selected and Active/Active load balancing is desired, the SCSI-3 Persistent Group Reservations (SCSI-3 PGR) support mode must be enabled for the DMP DSM.
A cluster dynamic disk group that is part of the cluster resources cannot be a source disk group for a join command. However, it can be a target disk group for the command.
Change in Bringing a Two-Disk Cluster Group Online
In earlier versions of Volume Manager for Windows, it was possible to bring a two-disk cluster disk group online when only one disk was available. If a cluster were to lose all network communication, this allowed the disk group to be brought online on two cluster nodes simultaneously, with each node owning a single disk, possibly resulting in data loss or a partitioned cluster. Though the likelihood of this situation occurring is slim for most customers, the consequences if it does happen may be severe. However, this is no longer supported with recent versions of Volume Manager and it is not possible to bring a two-disk cluster disk group online in Volume Manager unless it complies with the normal majority algorithm which means both disks must be available.
The normal majority algorithm is (n/2 +1).
You are not allowed to deport a cluster disk group that is also a Volume Manager disk group resource for Microsoft Failover Clustering.
Connecting to a Cluster Node
If you connect to a computer from the VEA GUI using the virtual name or the virtual IP address, the VEA GUI displays the computer name of the cluster node that currently owns the virtual name and IP resources. Therefore, it is not recommended to use the virtual name or virtual IP address when connecting and administering a cluster node through SFW HA.
Instead, use the actual computer name or the IP address of the cluster node.
Dynamic Multi-Pathing (DMP) does not support using a basic disk as a cluster resource under Microsoft Failover Clustering.
Failover may not function properly when using Dynamic Multi-Pathing with a Microsoft Failover Clustering basic disk cluster resource. Refer to Tech Note 251662 on the Veritas Support site for details.
If you want to use Dynamic Multi-Pathing with SFW and Microsoft Failover Clustering, you must convert any Microsoft Failover Clustering basic disk cluster resources to dynamic disk cluster resources before activating Dynamic Multi-Pathing. The initial setup of Microsoft Failover Clustering requires that you use a basic disk as the quorum disk. Once InfoScale Storage is installed, you should upgrade the basic disk to dynamic by including it in a dynamic cluster disk group and then convert the quorum resource from a basic disk resource to a dynamic disk resource.
Cluster dynamic disk groups that contain iSCSI disks are not set up for persistent login on all nodes in the cluster.
SFW ensures that the iSCSI targets of cluster dynamic disk groups that contain iSCSI disks are configured for persistent login. If the persistent login is not configured for the target, SFW automatically configures it.
Cluster dynamic disk groups that contain iSCSI disks are only automatically configured for persistent login on the node where they were created. The other nodes in the cluster are not enabled for persistent login. You need to manually set up the persistent login for each of the other nodes in the cluster.
Copying the policy file, VxVolPolicies.xml, to Another Node
If the second node is configured the same as the first and if the first node's policy settings for Automatic Volume Growth are to be maintained on the second node, you need to copy the VxVolPolicies.xml file of the first node to the second node. Copy the VxVolPolicies.xml file to the same path location on the second node as its location on the first node. The default path of the VxVolPolicies.xml file is Documents and Settings\All Users\Application Data\Veritas.
More information about the policy file is available.
More information about using SFW and Microsoft Failover Clustering in a shared cluster environment with the FlashSnap off-host backup procedure is available.
If you install the Microsoft Failover Clustering feature on a server on which InfoScale Storage for Windows is already installed, then you must manually restart Veritas Enterprise Administrator Service (VxSvc) by running the following commands: