Managed disks (MDisks)

A managed disk (MDisk) is a logical disk (typically a RAID array or partition thereof) that a storage subsystem has exported to the SAN fabric to which the nodes in the cluster are attached. A managed disk might, therefore, consist of multiple physical disks that are presented as a single logical disk to the SAN. A managed disk always provides usable blocks of physical storage to the cluster even if it does not have a one-to-one correspondence with a physical disk.

Each managed disk is divided into a number of extents, which are numbered, from 0, sequentially from the start to the end of the managed disk. The extent size is a property of managed disk groups. When an MDisk is added to an MDisk group, the size of the extents that the MDisk will be broken into depends on the attribute of the MDisk group to which it has been added.

Access modes

The access mode determines how the cluster uses the MDisk. The possible modes are:

Unmanaged
The MDisk is not used by the cluster.
Managed
The MDisk is assigned to an MDisk group and is providing extents that virtual disks (VDisks) can use.
Image
The MDisk is assigned directly to a VDisk with a one-to-one mapping of extents between the MDisk and the VDisk.
Attention: Attention: If you add a managed disk that contains existing data to a managed disk group, you will lose the data that it contains. The image mode is the only mode that will preserve this data.

The figure shows physical disks and managed disks.

Figure 3. Controllers and MDisks

The status of a managed disk consists of four settings. The following table describes the different states of a managed disk:

Table 1. Managed disk status
Status Description
Online The MDisk can be accessed by all online nodes. That is, all the nodes that are currently working members of the cluster can access this MDisk. The MDisk is online when the following conditions are met:
  • All timeout error recovery procedures complete and report the disk as online.
  • LUN inventory of the target ports correctly reported the MDisk.
  • Discovery of this LUN created successfully.
  • All of the managed disk target ports report this LUN as available with no fault conditions.
Degraded The MDisk cannot be accessed by all the online nodes. That is, one or more (but not all) of the nodes that are currently working members of the cluster cannot access this MDisk. The MDisk may be partially excluded; that is, some of the paths to the MDisk (but not all) have been excluded.
Excluded The MDisk has been excluded from use by the cluster after repeated access errors. Run the Directed Maintenance Procedures to determine the problem. You can reset an MDisk and include it in the cluster again by running the svctask includemdisk command.
Offline The MDisk cannot be accessed by any of the online nodes. That is, all of the nodes that are currently working members of the cluster cannot access this MDisk. This state can be caused by a failure in the SAN, the storage subsystem, or one or more physical disks connected to the storage subsystem. The MDisk will only be reported as offline if all paths to the disk fail.
Extents

Each MDisk is divided into chunks of equal size called extents. Extents manage the mapping of data between MDisks and virtual disks (VDisks).

Attention: Attention: If your fabric is undergoing transient link breaks or you have been replacing cables or connections in your fabric, you might see one or more MDisks change to the degraded status. If an I/O operation was attempted during the link breaks and the same I/O failed several times, the MDisk will be partially excluded and will change to a status of degraded. You should include the MDisk to resolve the problem. You can include the MDisk by either selecting the Include MDisk task from the Work with Managed Disks - Managed Disk panel in the SAN Volume Controller Console, or issue the following command:
svctask includemdisk <mdiskname/id>

Managed disk path Each managed disk will have an online path count, which is the number of nodes that have access to that managed disk; this represents a summary of the I/O path status between the cluster nodes and the particular storage device. The maximum path count is the maximum number of paths that have been detected by the cluster at any point in the past. Thus if the current path count is not equal to the maximum path count then the particular managed disk may be degraded. That is, one or more nodes may not see the managed disk on the fabric.

(C) Copyright International Business Machines Corporation 2004