SP Network Backup, Boot and Restore with Sysback


Contents

About this document
How to set up the CWS for Sysback reinstallaton of your nodes
Configuring and backing up your node for possible recovery from CWS
How to recover the SP2 node with this setup
Appendix A: Manual netbooting of an SP node
Appendix B: Special procedures for high nodes (SMP)
Appendix C: Special procedures for silver nodes (CHRP)
Appendix D: SP post-install considerations

About this document

This document explains how to back up and restore SP nodes for reinstallation with Sysback. It applies to in-service versions of Sysback and any supported AIX level.


How to set up the CWS for Sysback reinstallation of your nodes

  1. Check the filesets on the boot server (CWS).
  2. Install the proper filesets for device support. See cloning faxes if an exact fileset list is required.
  3. Reboot CWS to incorporate these changes into the running config.
  4. Rebuild the Sysback boot images. Execute smitty sysback and proceed through the following panel sequence.
    1. Configuration options
    2. Network boot configuration
    3. Add or change a network boot client
      Fill in the hostname and platform type. Set the parameters as follows:
      • Thin = rs6k (not MP)
      • Wide = rs6k (not MP)
      • High = rs6k/MP
      • Silver = chrp/MP (not rspc)
      NOTE: This is not RSPC and must be CHRP/MP or the boot will fail with an LED E1F7, indicating an invalid boot image!
    4. Update network boot image

    5. Select platform and network type.

  5. Configure the CWS for a Sysback backup and recovery server. Execute smitty sysback and proceed through the following panel sequence:
    1. Configuration options
    2. Remote services
    3. Server options
    4. Configure remote services
    5. Add or change a client
      1. Client hostname (per reverse name resolution)
      2. User access (root, all or a specified user's name)
      3. At the menu, fill in the tape devices if they apply (all, /dev/rmtX or blank for none).
      4. Fill in the directories if applicable.

Configuring and backing up your node for possible recovery from CWS

  1. Make sure the node is set up to back up to the CWS. Execute smitty sysback and proceed through the following panel sequence:
    1. Configuration options
    2. Remote services
    3. Client options
    4. Configure remote services
    5. Add or change a server and enter the CWS's hostname per forward name resolution. Disk options should be no for tape and yes for disk (directories).

  2. Back up the node on a regular basis (from SMIT). Execute smitty sysback and proceed through the following panel sequence:
    1. Backup and recovery options
    2. Full system backup (Install Image)
    3. Back up the system
    4. Choose remote device for backup
    5. Select any additional (non-rootvg) volume groups to be included.

  3. Back up the node on a regular basis (from command line or script).
  4. sysback -f rmtX -h $cws -N $net -T $cpu -k $krn vg1 vg2
    

How to recover the SP2 node with this setup

  1. Execute spmon -g.
  2. Power down the node. Use the GUI or issue the following command:
  3. spmon -p off node1
    
  4. Use netboot:
    1. Global controls
    2. Netboot
    3. Select node number

  5. If netboot fails or if you need to boot from a node other than CWS, proceed through the following steps.
  6. Open a tty with the GUI or with the command.

    See Appendix A for the manual netboot process.

  7. When Sysback comes up:
    1. Change installation device
    2. Select tape on CWS
    3. Change volume group & logical volume info
    4. Verify information
    5. Install the system with current settings

  8. See Appendix D (a Sysback manual excerpt) for post-installation considerations.

Appendix A: Manual netbooting of an SP node

  1. Open the Node Front Panel via spmon as described in the section "How to recover the SP2 node with this setup", step 4.
  2. Logically power off the node with Node Controls. Use the GUI or issue the command spmon -p off node1.

  3. If this node is an SMP (high node), go to Appendix B and proceed through the steps there. Then, as indicated, skip to step 10 in this appendix.
  4. If this node is a silver node (CHRP), go to Appendix C and proceed through the steps there. Then, as indicated, skip to step 21 in this appendix.
  5. Put the key in Secure mode. Use the spmon GUI or issue the command (for node slot 1):
    spmon -key secure node1
    
  6. Power on the node with Node Controls. Use the GUI or issue the command:
    spmon -p on node1
    
  7. When the LED reaches 200, put the key in Service mode. Use the spmon GUI or issue the command (for node slot 1):
    spmon -key service node1
    
  8. Reset the node with Node Controls.
  9. When the LED reaches 260 or 262, select Open TTY.
  10. You should see the following menu.
  11. NOTE: If the menu is not displayed, press the Enter key.

    	MAIN MENU
    	1.	Select BOOT (Startup) Device
    	2.	Select Language for these Menus
    	3.	Send Test Transmission (PING)
    	4.	Exit Main Menu and Start System (BOOT)
    

    Type the number for your selection and then press Enter. (Use the Backspace key to correct errors.)

  12. Select option 1 from the MAIN MENU to get to the menu below.
  13. 	SELECT BOOT (STARTUP) DEVICE
    	Select the device to BOOT (Startup) this machine.
    	WARNING:  If you are using Token-Ring, selection of an
    	incorrect data rate can result in total disruption of the
    	Token-Ring network.
    	"==>" Shows the selected BOOT (startup) device
    		1. Use Default Boot (Startup) Device
    		2. Ethernet:  Slot 0/1, 15-pin connector
    		3. Ethernet:  Slot 0/1, BNC connector (1-pin)
    		4. Ethernet:  Slot 0/4, 15-pin connector
    	Page 1 of 2
    	88. Next Page of Select BOOT (Startup) Device Menu
    	99. Return to Main Menu
    
    Type the number for your selection and then press Enter. (Use the Backspace key to correct errors.)

  14. This menu will always have Use Default Boot (Startup) Device as the initial selection. Choose the correct boot device from the menu.
  15. Fill in the appropriate IP addresses.
    1. Be sure to use leading zeroes.
    2. You will be prompted to enter:
      1. IP addresses for the client node to be installed
      2. The server IP address boot server, which should be at the same level of AIX as the node's backup
      3. A gateway IP address (leave empty if same subnet)
      	SET OR CHANGE NETWORK ADDRESSES
      	Select an address to change
      	Currently selected BOOT (startup) device is:
      	Ethernet:  Slot 0/1, BNC connector (1-pin)
      	Hardware address .............................. 02608C2D4BDF
      	   1.	Client address                            009.019.173.027
      		(address of this machine)
      	   2.	BOOTP server address                      009.019.173.023
      		(address of the remote machine you boot from)
      	   3.	Gateway address                           000.000.000.000
      		(Optional, required if gateway used)
      	   97.	Return to Select BOOT (Startup) Device Menu
      		(Saves addresses)
      	   99.	Return to Main Menu (Saves addresses)
      
      Type the number for your selection, then press Enter. (Use the Backspace key to correct errors.)

  16. Select 99 to return to the Main Screen.
  17. At the MAIN MENU, select 3 to ping the server (test network).
  18. 	SEND TEST TRANSMISSION (PING)
    	A test to see if the machine at the origin
    	address can communicate, thru the network, with the
    	machine at the destination address.
    	   Currently selected BOOT (startup) device is:
    	   Ethernet:  Slot 0/1, BNC connector (1-pin)
    	   Hardware address ............................ 02608C2D4BDF
    	   Select an address to change or select "4" to begin the test.
    	   1.	Origin address                          009.019.173.027
    	   2.	Destination address                     009.019.173.023
    	   3.	Gateway address                         000.000.000.000
    		(Optional, required if gateway used)
    	   4.	START PING TEST
    	   99.	Return to Main Menu
    
    Type the number for your selection, then press Enter. (Use the Backspace key to correct errors.)

  19. Select 4 to start the ping test.
    1. If the ping test is unsuccessful you have a network problem or you have entered an invalid IP.
    2. Do not proceed until you correct the problem.

    A successful test will display the following:

    	TEST TRANSMISSION (PING) RESULTS
    	SUCCESSFUL TEST.  Transmission sent and received.
    	   97.	Return to Send Test Transmission screen.
    	   99.	Return to Main Menu
    
    Type the number for your selection, then press Enter. (Use the Backspace key to correct errors.)

  20. Select 99 to return to the main menu.
  21. At the MAIN MENU, select 4 to start the System Boot.
  22. 	STARTING SYSTEM (BOOT)
    	To get a NORMAL boot, turn the key on your system unit
    	to "NORMAL" and press "ENTER" to continue booting.
    	99. Return to Main Menu
    
    Type the number for your selection, then press Enter. (Use the Backspace key to correct errors.)

  23. Put the key in Normal mode (via the spmon Node Front Panel).
  24. Press Enter in the tty screen.
  25. You should see one bootp packet sent and received.
  26. You should see many tftp packets sent and received.
  27. Once tftp is completed, the system will run the kernel.
  28. After all devices are configured, your system should come up in Maintenance Mode.

Appendix B: Special procedures for high nodes (SMP)

Do not use this procedure for non-HIGH nodes.

  1. Put the key for the high node in the Service position. Use the spmon GUI or issue the command (for node slot 1):
    spmon -key service node1
    
  2. Open a tty on the node. Use the GUI or issue the command:
    spmon -open node1
    
  3. At the ">" prompt, enter sbb and hit Enter.
  4. Choose option 1, Set Flags.
  5. Ensure that the following two flags are set correctly:
    1. Autoservice IPL should be set to Disable.
    2. Fast IPL should be set to Enable.

  6. Exit to the ">" prompt and break the tty connection.
    1. Type x until the ">" prompt appears.
    2. Type Ctrl-x to break the tty connection.

  7. Power the node on. Use the GUI or issue the command:
    spmon -p on node1
  8. Open a tty on the node. Use the GUI or issue the command:
    spmon -open node1
  9. If you see the following prompt, hit Enter:
  10. DO YOU WANT TO UPDATE FLASH FROM LINE S2 [y/n] ? n
    
  11. On the MAINTENANCE MENU screen, choose option 6, SYSTEM BOOT.
  12. On the SYSTEM BOOT screen, choose option 1, BOOT FROM NETWORK.
  13. Continue with step 10 in Appendix A.

Appendix C: Special procedures for silver nodes (CHRP)

  1. Power on the node. Use one of the following:

  • Open a tty on the node. Use the Node Front Panel in spmon -g or execute:
    spmon -open node1
  • Enter SMS.
    1. Wait for the system logo.
    2. Press 1 at 1-second intervals until the message entering SMS appears.
    3. You will see the following menu:
          1  Display Configuration
    	2  Multiboot
    	3  Utilities
    	4  Select Language
    	===>
    
  • Enter the Bootp menu.
    1. Wait for SMS to come up.
    2. Choose Utilities.
    3. Choose Remote Initial Program Load.
    4. Select your network adapter from this menu.

  • Fill in your TCP/IP parameters. Do not use leading zeroes.
  • Perform a network ping.
    1. If the ping test is unsuccessful you have a network problem or you have entered an invalid IP.
    2. Do not proceed until you correct the problem.

  • Exit back to the SMS main menu. You will see the following menu:
  • 	1  Display Configuration
    	2  Multiboot
    	3  Utilities
    	4  Select Language
    	===>
    
  • Choose option 3, Multiboot. You will see Please Wait and then the following:
    	RS/6000 Firmware
    	Version WIL97072
    	(c) Copyright IBM Corp. 1996  All rights reserved.
    	--------------------------------------------------------------------
    	Multiboot
    	1  Select Software
    	2  Software Default
    	3  Install From
    	4  Select Boot Devices
    	5  OK Prompt
    	6  Multiboot Startup 
    	===>
    
  • Choose Install From, Option 3. You will see a menu similar to the following:
    	Install Operating System
    	Device Number  Device Name
    	1               SCSI 4512 MB Harddisk id=0 ( Integrated )
    	2               SCSI 4512 MB Harddisk id=1 ( Integrated )
    	3               Ethernet  ( Integrated )
    	4                                 None
    	===>
    
  • Select the network interface by typing its number and pressing Enter.
  • Return to Appendix A, step 21.

  • Appendix D: SP post-install considerations

    The SP may be configured in many ways, using thin, wide or high nodes. Each frame may also have an optional high-speed switch for very fast network data transfer between nodes. The nodes are independent machines, each with its own operating system installed. The nodes do not have a diskette drive or front panel key switch or LED, and in basic configurations, external devices are not attached.

    Sysback is particularly useful on the SP for its ability to back up and restore over a network and to perform network installs from a network-attached tape drive or disk image files. This enables nodes to be backed up and even reinstalled from a tape drive or disk image file located on another node, the control workstation, or any other host on the network with which the node can communicate.

    When one performs a backup or restore from the control workstation, the data will be transferred across the Ethernet (default) network interface. This high amount of data transfer may impact other applications that communicate over the same network. If a high-speed switch is installed in a frame and the backup or restore is performed from one node to another node in the same frame, the network is generally not impacted due to its very high throughput rate.

    Any backup or restore operation performed with Sysback on the SP is identical to any backup or restore performed over a local-area network. The Sysback installation process on an SP is identical to network installs on other types of hosts, except that the nodes to be installed do not have a system console directly attached. It is therefore necessary to use the PSSP software, provided on the control workstation, to assist in a network boot operation.

    In most SP configurations, the nodes do not have tape drives, so the Sysback network installation process is the primary means of system recovery. Although the system backups can be performed over the high-performance switch (HPS), a network installation is limited to a Sysback-supported network (Ethernet, token-ring or FDDI).

    The spbootins and setup_server commands, provided with the PSSP software, may be used to "customize" a node. When this occurs, however, the node will begin pointing to the PSSP network boot image, rather than the Sysback network boot image. If you wish to perform another, later Sysback network boot of the same node, you will need to reconfigure the network boot client using the Sysback option Add or Change a Network Boot Client. Likewise, if after performing a Sysback network boot, you wish to perform a PSSP "customize / install", you must again use the spbootins and setup_server commands.

    Control Workstation Recovery

    PSSP software, provided for the control workstation, modifies the control workstation's AIX device configuration database to add an entry for the node_number in the parallel cluster. For the control workstation, this number is set to zero (0).

    Since this information is added to the device configuration database by the PSSP, and not AIX, Sysback is unable to recreate this information when it rebuilds the device configuration database at the end of the system installation of the control workstation. After the system installation, the command /usr/lpp/ssp/bin/install_cw must be executed on the control workstation. Running the install_cw program will restore the ODM device configuration specific to the SP control workstation.

    Node Recovery

    As with the control workstation, Sysback is unable to recover the node-specific AIX device configuration that PSSP applies to the nodes (SP and CSS attributes). Therefore, additional steps must be taken to recover this information for the node using commands provided with the PSSP software.

    NOTE: Make sure the node is set to disk in the SDR during the Sysback installation, since Sysback will be completing the installation process and not the PSSP code.

    After the node is installed from the Sysback backup, you must perform the following steps to reconfigure the node-specific device configuration information:

    1. On the control workstation, enter the command:
           spbootins -r customize-l #
      # is the node number. This will set the node to customize mode.
    2. On the control workstation, enter the command:
           splstdata -b
      This confirms that the node is set to customize mode.
    3. Perform this step's substeps on the node only if the system was installed from a backup image created from a different node:
      1. Enter:
             rm -f /etc/niminfo
      2. Edit the files /etc/ssp/server_name and /etc/ssp/server_hostname to point to the node's boot/install server.
      3. Edit the file /etc/ssp/reliable_hostname to reflect the node's reliable network IP address and hostname.
      4. Edit the file /etc/ssp/cw_name to ensure that it points to the control workstation.
    4. On the CWS, run the command:
           lslpp -h ssp.basic
    5. If the value returned shows 2.1.0.13 or later, enter the following on the node:
           export NIM_CUSTOM=sysback
    6. On the node, enter:
           /usr/lpp/ssp/install/bin/pssp_script
      If this file does not exist on the node, you can copy it from the /SPDATA/SYS1/INSTALL/PSSP/PSSP_SCRIPT file on the control workstation.

    It is important to remember that you must boot from the same AIX-version operating system that you will be installing. Therefore, if you plan to install from a tape drive attached to the control workstation, the control workstation must either have the same level of operating system installed as the node you are installing, or you must use a different server to perform the network boot. After booting from the correct level of operating system, you may select a different server on the network from which to retrieve the backup data.




    [ Doc Ref: 92783321814224     Publish Date: Oct. 17, 2000     4FAX Ref: 4396 ]