ServeRAID II Considerations

ServeRAID II Considerations

Terms and Definitions

It is important to understand the following ServeRAID II adapter considerations before you set up your high-availability cluster solution.

Note: The following ServeRAID II considerations are for the Microsoft Cluster Solution using Windows NT.
For similar considerations for the Novell IntranetWare Cluster Solution, refer to
'Vinca High Availability for NetWare and IBM ServeRAID II SCSI Adapter User's Guide.

Microsoft Cluster Server (MSCS) Configurations using the ServeRAID II adapters in each of the servers require a small computer system interface (SCSI) connection joining the two ServeRAID II adapters together.
This connection is known as a SCSI heartbeat . The SCSI heartbeat connection is required to manage the Microsoft Cluster Server quorum disk and, in the event of a server failure, to determine which is the surviving server.
The connection is a combination of any length 0.8-mm to 68-pin cable and a pigtail 68-pin to 0.8-mm cable.
For more details about this connection, see the 'IBM Shared Disk Clustering Hardware Reference', Part Number 33L3945.
The SCSI heartbeat connection must be connected to the third channel of the ServeRAID adapter pair that has the quorum drive connected to it. No disks can be installed on this heartbeat channel.
If you choose to move the quorum drive to another ServeRAID II adapter, you must also move the SCSI heartbeat cable on both servers to the new quorum ServeRAID adapter pair.
The quorum disk can be placed on any ServeRAID channel shared by the two servers,
except for the channel used by the SCSI heartbeat cable.
The quorum logical drive can be used to store customer programs and data.
The ServeRAID II adapter does not allow the failover of RAID level-5 logical drives that are in a critical state; that is, a drive in a RAID level-5 array has failed. For this reason, do not use RAID level 5 for the quorum disk.
The quorum drive must be configured as a RAID level-1 logical drive.
Note: You should define hot-spare drives (HSPs) in your array to minimize the time that logical drives remain in critical state.
Every logical drive that is shared by the two servers must have its cache policy set to write-through mode to ensure data integrity is maintained.
Logical drives that are not shared between the two servers can be configured for write-back mode for improved performance.
The SCSI-2 Fast/Wide Enhanced Repeater Card is required for configurations using the
IBM Enterprise Expansion Enclosure (Model 3518) or the IBM Model 3519 Rack Expansion Enclosure.
The previous version of the SCSI-2 Fast/Wide Repeater is not supported in these configurations.
For information on the latest version of the SCSI Fast/Wide Enhanced Repeater Card, contact your IBM remarketer or dealer.
Each ServeRAID II adapter must have a unique host ID assigned and the host ID of its cluster partner must be specified.
Do not attempt to change the host ID on both ServeRAID II adapters in a pair at the exact same time.
Doing so can cause a problem, such as a server lock-up.
Each ServeRAID II adapter must have a different SCSI Initiator ID assigned to the shared channels.
For example, one server, Server A, could be set to ID 6, and the other server, Server B, could be set to ID 7.
This will be described during the setup instructions.
The stripe unit size of each ServeRAID adapter in a pair must be set to the same value (8 KB, 16 KB, 32 KB, or 64 KB).
The SCSI channels in an adapter pair must be cabled so that the same channels are shared between adapters.
All disks that will be shared must be connected to the shared channels. This includes hot-spare (HSP) drives, as well as data drives.
Only one logical drive must be created for each physical array because only one server at a time can own a physical drive.
For example, only logical drives A0, B0, C0, and so on, are supported, not A1, B1, and so on.
Merge ID numbers in the range 1-8 must be assigned to each logical drive that will be shared. Merge ID numbers in the range 200-215, where the last two digits are the SCSI ID of a shared channel, must be assigned to the nonshared logical drives.
If you are starting (booting) the operating system from a shared adapter, define the first logical drive as the startup drive and assign a nonshared Merge ID, for example, 206 for Server A.
Do not connect nondisk devices, such as tape drives, to shared channels. They will be accessible to both servers and conflicts can arise.
Hot-spare drives are not shared between the adapters in a pair.
Each ServeRAID II adapter must have a hot-spare drive defined if hot-spare protection is desired, no matter which server is controlling a logical drive.
All hot-spare drives must be connected to shared channels so that, if the hot-spare drive is used to replace a failed shared disk, the disk can be accessed from either server in the case a failover occurs later.
RAID level-0 logical drives can failover in any state.
RAID level-1 logical drives cannot failover in critical state.
Use hot-spare drives to minimize the time that logical drives are in critical state and exposed to a second disk or system failure.
The total number of logical drives per adapter is eight, before or after a failover.
A failover will not complete if this number is exceeded.
Logical drives that are currently undergoing Logical Drive Migration (LDM) operations, will not be allowed to failover.
However, all other drives will be able to failover if necessary.
Drives that are configured by one adapter in the pair must appear as RDY (ready) drives to the other adapter in the pair.
The same must also be true in reverse. Be sure to verify this consideration when drives are added after the nodes are started.
After adding new drives to the server, you can use the RAID Administration and Monitoring Utility to restart the servers, scan for new devices, or set the drives to the RDY state.
All shared drives must be part of a logical disk or defined as a hot spare to one or the other servers.
All other drives should be removed from the shared disk enclosure to ensure proper operation.
If your configuration will contain multiple ServeRAID adapter pairs, each adapter Host Configuration ID must be unique.
If a failover occurs while a critical RAID level-1 logical drive is rebuilding to a spare disk, the rebuild automatically starts a few seconds after the failover is completed.
The cluster support software will initiate a synchronization of RAID level-1 and RAID level-5 logical drives immediately after a failover.
If a drive fails before this synchronization is complete, RAID level-5 logical drives are placed in the blocked state.
When a logical drive spans across multiple channels and a failure within the drive subsystem occurs that is unique to a channel (for example, a disconnected cable), the entire physical array will be identified as unsuccessful even though access from the surviving server can occur.
Therefore, you might want to consider, if you have small arrays, not spanning across multiple channels.

Back to

Please see the LEGAL - Trademark notice.
Feel free - send a for any BUG on this page found - Thank you.