Support for multiple clusters attached to the same SAN device (304415)

SUMMARY

A single cluster that supports several nodes can be qualified and put on a Microsoft Hardware Compatibility List (HCL). This configuration requires storage interconnects and switch technology. There are many examples of complete configurations on the HCLs. However, this does not represent a storage area network (SAN) multiple cluster server or a heterogeneous clustered configuration where multiple clusters or non-Windows hosts are attached to the same storage.

Microsoft supports having multiple clusters (or multiple Clusters and stand-alone servers) attached to the same SAN. For Windows NT, only the storage subsystem name must be on the static text list on the Windows NT 4.0 HCL under the following category:

Cluster/Multi-Cluster Device

Note There are no further requirements.

For Microsoft Windows 2000 or Microsoft Windows Server 2003, the storage subsystem must be on the Windows Server Catalog (WSC) under the following category:

Storage, RAID Storage, Multi-Cluster

Note The Storage, RAID Storage, Multi-Cluster category has the storage subsystem that has been tested with specific HBAs and drivers. The listing in the Storage, RAID Storage, Multi-Cluster category indicates that the storage subsystem has been tested with multiple clusters that have been attached to the same storage subsystem. The storage subsystem must be configured as specified in the Storage, RAID Storage, Multi-Cluster category, including drivers, switches, and HBAs. iSCSI storage solutions cannot be listed in the Storage, RAID Storage, Multi-Cluster category. iSCSI storage solutions are not required to have listings in this category.

MORE INFORMATION

Example of a supported SAN-based cluster

Storage, RAID Storage, multi-cluster

Storage subsystem SD1 HBA H1
Storage subsystem SD2 HBA H2

Storage Cluster HCL

2 node advanced server cluster AS1
- Server 1: Server Box S1, 256 MB, 700Mhz PIII, HBA H1
- Server 2: Server Box S2, 256 MB, 700Mhz PIII, HBA H1
- Storage: Storage subsystem SD1
4 node advanced server cluster AS2
- Server 1: Server Box S5, 512 MB, 1.2Ghz PIV, HBA H1
- Server 2: Server Box S6, 512 MB, 1.2Ghz PIV, HBA H1
- Server 3: Server Box S7, 512 MB, 1.2Ghz PIV, HBA H1
- Server 4: Server Box S8, 512 MB, 1.2Ghz PIV, HBA H1
- Storage: Storage subsystem SD1
2 node advanced server cluster AS3
- Server 1: Server Box S10, 256 MB, 700Mhz PIII, HBA H2
- Server 2: Server Box S11, 256 MB, 700Mhz PIII, HBA H2
- Storage: Storage subsystem SD2

In this case, the 2 node AS1 and the 4 node AS2 configurations can both be put on the same SAN and can share the same storage subsystem. It is also possible to have AS3 in the same SAN as long as it uses storage subsystem SD2 (and not SD1) and is completely isolated from all other clusters. This is because it is using a different HBA than was tested with the multi-cluster submission.

A key issue that needs to be addressed is the isolation of the different clusters that are attached to the SAN. The SAN must be configured so that a set of cluster disks are only accessible by one cluster and are completely hidden from the rest of the clusters. One or more mechanisms may be employed to accomplish this isolation such as defining zones on the switch or enabling LUN masking on the storage subsystems. The use of Fiber Channel Arbitrated Loops (FC-AL) is not supported because hubs typically do not allow for the clusters to be isolated from each other.

The best way to configure multiple clusters that are attached to the same storage subsystem is to involve the storage subsystem vendor. A clustered SAN environment can be very complex, and if zones or LUN masking are not configured properly, a cluster failure on one of the clusters may affect a totally separate cluster.

The Cluster HCL is available on the following Microsoft Web site:

http://www.microsoft.com/hcl

The Windows NT 4.0 clusters and multi-cluster submissions are in a static text list. The Windows Server 2003 and Windows 2000 clusters and multi-cluster submissions are part of the Windows Server Catalog (WSC).

Error Messages

Typical System log errors that maybe seen on the Cluster nodes if there is a configuration issue are as follows. This is not a complete list but does indicate some of the common errors:

Event ID: 51
Event Type: Warning
Event Source: FTDISK
Description: An error was detected on device \Device\Harddisk0\DR0 during a paging operation.

Event ID: 11
Source: %HBA_DRIVER_NAME%
Description: The driver detected a controller error on Device\ScsiPort0.

Event ID: 9
Source: %HBA_DRIVER_NAME%
Description: The device, \Device\ScsiPort0, did not respond within the timeout
period.

Event ID: 50
Event Type: Warning
Event Source: Disk
Description: {Lost Delayed-Write Data} The system was attempting to
transfer file data from buffers to \Device\HarddiskVolume1. The
write operation failed, and only some of the data may have been
written to the file.

Event ID: 26
Source: Application Popup
Description: Windows - Delayed Write Failed : Windows was unable
to save all the data for the file \Device\HarddiskVolume1\MFT$.
The data has been lost. This error may be caused by a failure
of your computer hardware or network connection. Please try to
save this file elsewhere.