A P P E N D I X  E

Best Practices

This appendix provides information about how to best use the Sun StorageTek SAS RAID HBA to improve your overall user experience. The appendix contains the following sections:


Best Practices For Placing a New HBA in a System Running the Solaris OS

If you install a new HBA into a system that is running the Solaris OS, the system might not detect the newly installed HBA.

To confirm whether the system has detected the HBA, do the following:

1. Run the arcconf getversion command at a terminal window.

Assuming the newly installed HBA is the sole HBA installed in the system, the output of the arcconf getversion command will display 1 adapter.

2. Review the output of the arcconf getversion command and do one of the following:

3. Repeat Step 1 to confirm that the system has detected the HBA.


Best Practices For Switching Cables and Making New Connections

This section contains the following topics:

Switching a Cable From One Port To the Other Port

If you need to switch a cable from one port to the other port on the same HBA or on the same JBOD, wait long enough after the initial cable pull for all the physical hard drives to be removed from the displays of the graphical user interface (GUI) and command-line interface (CLI). Waiting after the initial pull prevents the controller from trying to remove drives at the same time as it is re-adding those same drives on the other port. If no display is available, wait at least two minutes between pulling the cable and reconnecting it to the other port.

Pulling a Cable and Reconnecting It To the Same Port

Wait at least two minutes between pulling a cable and reconnecting it to the same port on an HBA or JBOD. To avoid confusing the controller, wait until the physical drives are removed from the display of the GUI and CLI.

Connecting a JBOD With SATA Disks

When connecting a JBOD with SATA disks to an HBA, connection problems might occur if the JBOD was previously connected to a different HBA. This is due to a SAS feature called, affiliation. An affiliation is used by the SAS protocol to prevent multiple SAS initiators (HBAs) from walking on each other when communicating with SATA drives. To avoid these connection problems, remove the affiliation by power cycling the JBOD prior to connecting the JBOD to the new HBA.


Best Practices For Cabling to Disk Enclosures

The HBA supports using a single connection, or path, between a RAID card and a JBOD. Therefore, when connecting a RAID card to a disk enclosure, be sure to use only one cable. Using more than one cable does NOT provide redundancy.


Best Practices For Testing Hard Drive Failure Conditions

When hot-plugging hard drives to test failure conditions, wait at least one minute between removing the drive and re-inserting it. Waiting one minute gives the controller time to recognize the removal of the hard drive. In a true failure situation, a removed drive is replaced with a different hard drive. In testing situations where the same drive is re-inserted, a one-minute waiting period before re-insertion is required.


Best Practices For Deleting Logical Volumes Without Deleting Partitions

When attempting to delete a logical volume, the Sun StorageTek RAID Manager graphical user interface (GUI) might require you to remove any partitions on the logical drive before it allows drive deletion. This is a safety feature designed to prevent accidental deletion of drives that contain valid data.

If you need to remove a logical volume without first deleting partitions from the OS, you can do so by using the arcconf command-line interface (CLI). The CLI allows the deletion of logical volumes without requiring the removal of partitions first.

To delete a logical volume through the CLI without deleting partitions, type the following at the command prompt:

arcconf DELETE controller-number LOGICALDRIVE logical-drive-number

You will receive a warning message, as shown in the following example.

 


WARNING: Logical device z may contain a partition.
All data in logical device z will be lost.
Delete the logical device?
Press y, then ENTER to continue or press ENTER to abort:


Best Practices For Testing Physical Drive Failures

If, for testing purposes, you use the Sun StorageTek RAID Manager GUI to fail a physical drive, the drive will not return to a ready state without you replacing the drive. In a test scenario, you may not need to physically replace the drive as you are only performing a test.

In this scenario, you can use the arcconf CLI to return the drive to its ready state without actually having to replace the drive.

To do so, type the following at the command prompt:

arcconf SETSTATE controller-number DEVICE drive-channel-number drive-ID-number RDY

Typing this command causes the failed drive to return to a ready state in the GUI.


Best Practices For Rescanning or Discovering Drives

If you add a new enclosure to an existing card, or if you initiate a card rescan and the rescan finds a new physical disk drive on the card, the newly discovered drive might be displayed as failed in the Sun StorageTek RAID Manager GUI (with a red X over the drive). When a drive is displayed as failed in the GUI, it causes multiple messages to be generated in the system messages file. These messages may indicate there is a failed drive, that the drive must be removed, or that the drive is not responding.

If this occurs, keep in mind that this is normal behavior for the GUI and you do not need to perform any further action. Treating the drive as failed is the safest way in which the GUI can handle newly discovered drives that are in an unkown state. After some time passes, and the GUI determines nothing is wrong with the drive, the failed marker will be removed and the drive can operate normally.

In the event that the newly added enclosure is a complete enclosure, it may take several minutes for the GUI to remove the failed marker from each drive. This is because the time for discovery is proportional to the number of drives added by the enclosure.


Best Practices For Controlling the Boot Order of Logical Drives

After installing a RAID HBA and creating at least one logical drive, the BIOS on the host system might insert that drive into the hard disk drive boot order in front of existing system disks. If the newly created logical drive does not have a boot sector, you will be unable to boot the host system upon subsequent reboots. This is not an issue if you are installing an internal RAID HBA, assuming you plan to boot off logical drives. However, if you are installing an external RAID HBA, the drive boot order might adversely affect other system boot drives.

To control the boot order of the logical drives on the host system, do the following:

1. Enter the host system BIOS, select boot, and review the Hard Disk Drives entry.

2. Verify whether the logical drive of the RAID card is the first drive in the list.

3. If the first drive listed is the original system boot drive, you do not need to perform any further action. If the first drive listed is the logical drive of the RAID card, change the list so that the original boot drive is listed first.

4. To save the settings, do one of the following:

The system can now boot as expected.


Best Practices For Selecting Members of RAID Logical Devices

When selecting physical drives to be members of RAID logical devices, make sure that the drives you select are from a single enclosure in a single logical device. This is because any action that removes an entire enclosure temporarily or permanently (firmware upgrades, switching off a single enclosure, dual power supply failures, and so on) could result in rebuilds, degraded states, and complete failures, depending on how many drives from a given RAID logical device are contained in one enclosure. Do not build logical devices that span enclosures unless you are an advanced RAID user who understands the risks of doing so.


Best Practices For Replacing an HBA

This section contains the following topics:

Replacing an HBA On a SPARC System

Upon replacing an existing HBA card with a new card on a SPARC system, the new card automatically detects the existing configuration upon installation and no further action is required. However, if the new HBA is not seeing one or more volumes in the existing configuration, rescan those volumes to enable their detection.

Replacing an HBA On an x64 System

Upon replacing an existing HBA card with a new card on an x64 system, the BIOS RAID Configuration utility sends a warning message alerting you to a configuration change during the initialization process of the new card. The existing drives and volumes in the configuration will not be activated on the new card until you accept the configuration change indicated in the warning message.

To activate the existing drives and volumes on the new card, do one of the following: