For successful replacement of your controller module, carefully follow the procedures contained in this document. If you do not follow the steps precisely, you risk controller failure. The following CAUTION is one of several cautions which are critical to successful replacement.
Topics in this document include:
Note - These FRU components can be replaced by customers or by Sun-trained personnel. |
All user documentation is available on the following websites:
Documentation includes the following manuals and release notes.
As a minimum, be sure to review the latest product release notes on the website above.
SCSI controller modules are hot-swappable. In the event that it is impossible or impractical to halt I/O from hosts to the array, a controller can be replaced while the surviving controller is active and servicing I/O.
However, if your configuration enables you to halt I/O during the controller replacement procedure without disruption, it is a good idea to do so. With write-back cache enabled under heavy I/O, cache synchronization can take three hours or more to complete. During this time, your controller will display "preparing failback" status and the LED remains amber.
It is also possible that hardware errors or configuration errors may be present but undetected, resulting in an unserviceable condition that will only be revealed when the hot-swap is underway. For instance, it is possible that a hardware module might be improperly seated.
It is good practice before replacing a controller FRU to try to verify your system's health. Do not replace a working controller on a channel that is offline.
When a controller fails in a dual-controller configuration, the remaining controller automatically becomes the primary controller, if it is not already. If you replace a controller that has not failed, force-fail the controller to be replaced via one of the following methods.
For details on the fail, show redundancy, and show events commands, refer to the Sun StorEdge 3000 Family CLI User's Guide.
Before replacing a controller module, save the NVRAM configuration settings to disk. The saved NVRAM file can be used to restore configuration settings.
1. From the RAID firmware Main Menu, choose "system Functions controller maintenance save NVRAM to disks."
2. Select Yes to confirm. A message informs you that NVRAM information has been successfully saved.
1. Go to http://sunsolve.sun.com.
3. Use Patch Finder to locate the patch ID by entering the patch ID into the search field and clicking the Find Patch button.
4. Select the link for the format that you want, either HTTP or FTP next to Download Patch, or HTTP or FTP next to Download Signed Patch.
5. In the dialog box that is displayed, indicate the destination directory for the patch and proceed to download the file to that location.
6. To install the latest firmware, follow the instructions in the patch README file.
When a controller is replaced in a dual-controller configuration, the controller firmware of the remaining functional controller automatically overwrites the firmware of the new replacement controller to maintain compatibility. This is referred to as cross-loading. Cross-loading uses the NVRAM configuration settings to synchronize the firmware version of the newly installed controller to match the firmware version of the running controller.
To replace a SCSI controller module in a dual-controller configuration, perform the following steps.
Removing a SCSI Controller Module |
2. Turn the thumbscrews on the left and right sides of the controller module counterclockwise until the thumbscrews are disengaged from the chassis.
3. Hold the thumbscrews, and carefully pull out the controller module.
Installing a SCSI Controller Module |
Caution - DO NOT POWER OFF the chassis when you replace a controller module. Multiple problems can occur. If you power off the array and replace a controller module in a dual-controller configuration, the replacement controller could become the primary controller and overwrite any configuration settings previously set. Additionally, if the array is powered off incorrectly, data that is written to cache and that has not been completely written to the disks will be lost. If you powered off the array during replacement, see Restoring the Configuration Settings of a Powered-Off Array. |
2. Gently slide the controller module into the array.
Caution - Be sure that the module is properly inserted in the guide rails of the array and that you keep the power on. If you power off and then replace the module, you will have to continue with additional steps. See Restoring the Configuration Settings of a Powered-Off Array. |
a. Insert the controller FRU into the slot, and push forward until you begin to feel resistance as the connecting pins engage.
b. Push the controller FRU until the connecting pins are fully seated and the RAID controller fits flush against the back plate of the RAID array.
Failure to insert the controller carefully as described above can cause one of the following problems:
3. Turn the thumbscrews on the left and right sides of the controller module clockwise until they are finger-tight, to secure the module and to make the module's front panel flush with the chassis.
To ensure that a thumbscrew is finger-tight, tighten it with a screwdriver and then loosen the thumbscrew counterclockwise a quarter turn.
The new controller automatically becomes the secondary controller.
In a redundant controller configuration where a new controller FRU is installed, the controller status LED will remain amber until the controllers complete the redundant controller process, which can take more than 10 minutes. Identical firmware versions on both controllers are required for proper redundant controller operation.
The redundant controller process automatically cross-loads the firmware version of the newly installed controller FRU to match the firmware version of the other running controller. For example, if the running controller has firmware 4.12B and the new controller has 4.12E, the new controller will be cross-loaded with the 4.12B firmware of the running controller.
To monitor this process, see Monitoring the Firmware Update for a Recently Installed Controller FRU.
Caution - You must connect the hosts to the correct host channels on the controller module or your configuration will not work correctly. |
To monitor the status of the firmware update, use the Sun StorEdge CLI show redundancy command. The Sun StorEdge CLI will display the progression of "Failed," "Scanning," "Detected" and "Enabled" states.
If the array was inadvertently powered off during the controller replacement in a dual-controller configuration, you must perform the following steps to ensure successful operation of your array.
1. In a dual-controller configuration where both controllers have the same version number (such as 4.12B), power on the array and wait a minimum of 10 minutes for firmware cross-load to occur.
The controller with the higher serial number becomes the primary controller.
2. Check that the secondary controller is active on the array by entering one of the following commands.
The Redundant Controller: Enabled status indicates an initialized secondary redundant controller.
The Redundancy Status: Enabled indicates an initialized secondary redundant controller.
If one controller fails in a dual-controller configuration, you might want to run a single controller for an extended period of time so that the array does not display as degraded.
For instructions on converting a dual-controller configuration to a single-controller configuration, refer to the section titled "To Convert a Dual Controller Array to a Single Controller Array" in the Sun StorEdge 3000 Family Configuration Service User's Guide.
Note - SCSI single-controller arrays ship with a blanking panel covering the secondary controller slot. This must be removed in order to install a secondary controller. |
1. Keep the array powered on and make sure that the connected hosts are inactive.
2. Turn the thumbscrews on the left and right sides of the blanking panel and remove it.
3. Gently slide the new controller module into the array.
Caution - Be sure that the module is properly inserted in the guide rails of the array and that you keep the power on. |
a. Insert the controller FRU into the slot, and push forward until you begin to feel resistance as the connecting pins engage.
b. Push the controller FRU until the connecting pins are fully seated and the RAID controller fits flush against the back plate of the RAID array.
Failure to insert the controller carefully as described above can cause one of the following problems:
4. Turn the thumbscrews on the left and right sides of the controller module clockwise until they are finger-tight to secure the module and to make the module's front panel flush with the chassis.
To ensure that a thumbscrew is finger-tight, tighten it with a screwdriver and then loosen the thumbscrew counterclockwise a quarter turn.
The new controller automatically becomes the secondary controller.
When the new controller FRU is installed, the controller status LED will remain amber until the controllers complete the redundant controller process, which can take more than 10 minutes. The same firmware versions must be installed on both controllers for proper redundant-controller operation.
The redundant-controller process automatically cross-loads the firmware version of the newly installed controller FRU to match the firmware version of the other running controller. For example, if the running controller has firmware 4.12B and the new controller has 4.15, the new controller will be cross-loaded with the 4.12B firmware of the running controller. To monitor this process, see Monitoring the Firmware Update for a Recently Installed Controller FRU.
5. If you want the most current version of firmware on your controllers, download the latest firmware patch as described in the release notes for your array.
6. Set up the host channels for the new controller module.
For host channel set up information, refer to the "Connecting Your Array" chapter in the Sun StorEdge 3000 Family Installation, Operation, and Service Manual for your array.
Caution - You must set the hosts to the correct host channels on the controller module or your configuration will not work correctly. |
To replace a SCSI controller module in a single-controller configuration, perform the following steps.
1. If possible, make a record of the firmware version and configuration settings before replacing the controller.
a. Use the show configuration CLI command to output the configuration settings to a file. Refer to the Sun StorEdge 3000 Family CLI User's Guide for more information.
b. Save NVRAM configuration settings to disk.
From the RAID firmware Main Menu select "system Functions Controller maintenance Save nvram to disks," and choose Yes to save the contents of NVRAM to disk.
c. Record the Unique Controller Identifier which combines the serial number and MAC address for each chassis and is used for network connections.
From the RAID firmware Main Menu, select "view and edit Configuration parameters Controller Parameters Controller Unique Identifier <hex>."
a. Keep the array powered on and make sure that the connected hosts are inactive.
b. Turn the thumbscrews on the left and right sides of the controller module counterclockwise until the thumbscrews are disengaged from the chassis.
c. Grasp the handle and carefully pull out the controller module.
3. Insert the replacement controller.
b. Gently slide the controller module FRU into the slot, and push forward until you begin to feel resistance as the connecting pins engage.
c. Slowly push the controller FRU the rest of the way until the connecting pins are fully seated and the RAID controller fits flush against the back plate of the RAID array.
Caution - Be sure that the module is properly inserted into the guide rails of the array. |
d. Turn the thumbscrews on the left and right sides of the controller module clockwise until they are finger-tight, to secure the module and to make the module's front panel flush with the chassis.
To ensure that a thumbscrew is finger-tight, tighten it with a screwdriver and then loosen the thumbscrew counterclockwise a quarter turn.
4. Restore the NVRAM settings.
If the old controller and the new controller have the same firmware version number, from the RAID firmware Main Menu, select "system Functions Controller maintenance Restore NVRAM from disks," and choose Yes to confirm.
5. Confirm that the "Controller Unique Identifier" parameter is set to the value recorded in step 1c.
a. From the RAID firmware Main Menu, select "view and edit Configuration parameters Controller Parameters Controller Unique Identifier <hex>."
b. Type the value 0 (to automatically read the chassis serial number from the midplane) or type the hex value for the original serial number of the chassis (used when the midplane has been replaced).
The Controller Unique Identifier is used to create Ethernet MAC addresses and worldwide names. The value 0 is immediately replaced with the hex value of the chassis serial number. A nonzero value should be specified only if the chassis has been replaced, but the original chassis serial number must be retained; this feature is especially important in a Sun Cluster environment, to maintain the same disk device names in a cluster.
c. To implement the revised configuration settings, choose "system Functions Reset controller" from the RAID firmware Main Menu, then select Yes to confirm.
Copyright © 2006, Dot Hill Systems Corporation and others. All rights reserved.