Manually Performing a Network Boot of an RS/6000 SP MCA Thin or Wide Node


Contents

About this document
    Related documentation
Booting an RS/6000 SP MCA thin or wide node via manual node conditioning

About this document

This document describes the procedure for performing a network boot via manual node conditioning on an RS/6000 SP MCA, thin or wide node. Manual node conditioning is a procedure that may be used when the automatic procedures to network boot a node in Perspectives or the command nodecond has failed. Manual node conditioning allows more effective monitoring and troubleshooting during the network boot.

The procedure detailed in this document is not AIX specific, so it applies to all versions of AIX.

Related documentation

The product documentation library is also available:
http://www.rs6000.ibm.com/resource/aix_resource/Pubs/index.html


Booting an RS/6000 SP MCA thin or wide node via manual node conditioning

  1. Configure the System Data Repository (SDR) for the appropriate bootp_response using one of the following procedures.
    1. From the command line on the control workstation, enter:
      	spbootins -r <bootp_response> -l <node_number, node_number, ...> -s yes
      

      NOTE: <bootp_response> = Choose one of the the following, install, migrate, maintenance, or diag (for diagnostic).
                   < node_number>= the number(s) of the node(s) separated by commas.

      If the preceding steps are not used, select one of the following procedures according to the appropriate Parallel System Support Programs (PSSP) level.

    2. From SMIT on the control workstation for PSSP 3.1.0, 3.1.1 or 3.2.0, enter smitty server_dialog.
      	---------------------------------------------------
      			Boot/Install Server Information
      	Type or select values in entry fields.
      	Press Enter AFTER making all desired changes.
      	Start Frame					[]	#
      	Start Slot					[]	#
      	Node Count					[]	#
      	OR
      	Node List					[]
      	Response from Server to bootp Request			+
      	Volume Group Name				[]
      	Run setup_server?				yes	+
      	---------------------------------------------------
      

      In the preceding screen, fill in the following:

      Node_List = The number(s) of the node(s) separated by commas.
      Response from Server to bootp Request = install, migrate, maintenance, or diag.
      Run setup_server = yes

    3. From SMIT on the control workstation for PSSP 2.2, 2.3, or 2.4, enter smitty server_dialog.
      	---------------------------------------------------
      			Boot/Install Server Information
      	Type or select values in entry fields.
      	Press Enter AFTER making all desired changes.
      	Start Frame					[]	#
      	Start Slot					[]	#
      	Node Count					[]	#
      	OR
      	Node Group					[]	+
      	OR
      	Node List					[]
      	Boot/Install Server Node Identifier			[]
      	Network Install Image Name			[]
      	Destination Hard Disk(s)				[]
      	Response from Server to bootp Request			+
      	LPP Source Name					[]
      	PSSP Level						+
      	/usr Server's Hostname or IP Address		[]
      	Gateway to /usr Server				[]
      	/usr Client Adapter Name				+
      	Run setup_server on the Control Workstation?	yes	+
      	---------------------------------------------------
      

    In the preceding screen, fill in the following:

    Node_List = The number(s) of the node(s) separated by commas.
    Response from Server to bootp Request = install, migrate, maintenance, or diag.
    Run setup_server on the Control Workstation = yes

    For all three of the preceding methods, setup_server should complete with a return code of 0, that is, rc = 0. If you do not see rc = 0, this could indicate possible NIM errors, which requires fixing before the network boot will work. Please contact customer support if you do not see rc = 0.

    NOTE: For steps 2-7, you may use either perspectives or the command line. Examples for both perspectives and the commmand line are included.

  2. Power off the node.
    1. If the node is still running, enter the following in the command line:
      	shutdown -f
      	spmon -p off node<#>
      

      <#> is the number of the node.

    2. From Perspectives on the control workstation:
      1. On the SP Perspectives Launch Pad, choose Hardware Perspective.
      2. Highlight the node you are working with.
      3. Press the Action Menu on the top menu bar and select the Power Off, Reset, Shutdown, or Fence Nodes option.
      4. In the Power Off options, choose Power Off.
      5. In the Shutdown options, choose Shutdown.
      6. In the Fence options, choose Fence with autojoin.
      7. Press Apply.

  3. Open the SP LED panels
    1. From the command line on the control workstation, enter:
      	spled &
      
    2. From Perspectives on the control workstation:
      1. On the SP Perspectives Launch Pad, choose Hardware Perspective.
      2. Click in the Nodes pane.
      3. Press the Action Menu on the top menu bar and select the LCD and LED Display option.

  4. Put the key in Secure mode.
    1. From the command line on the control workstation, enter:
      	spmon -k secure node<#>
      
    2. From Perspectives on the control workstation:
      1. On the SP Perspectives Launch Pad, choose Hardware Perspective.
      2. Highlight the node you are working with.
      3. Press the Action Menu on the top menu bar and select the Change Key Switch option.
      4. In the Key Switch Position, choose Secure.
      5. Press Apply.

  5. Power on the node.
    1. From the command line on the control workstation, enter:
      	spmon -p on node<#>
      
    2. From Perspectives on the control workstation:
      1. On the SP Perspectives Launch Pad, choose Hardware Perspective.
      2. Highlight the node you are working with.
      3. Press the Action Menu on the top menu bar and select the Power On or Cluster Power On Nodes option.
      4. In the Power On options, choose Power On.
      5. Press Apply.

  6. When the LED reaches 200, put the key in Service mode and reset the node.
    1. From the command line on the control workstation, enter:
      	spmon -k service node<#>
      	spmon -reset node<#>
      
    2. From Perspectives on the control workstation:
      1. On the SP Perspectives Launch Pad, choose Hardware Perspective.
      2. Highlight the node you are working with.
      3. Press the Action Menu on the top menu bar and select the Change Key Switch option.
      4. In the Key Switch Position, choose Service.
      5. Press Apply.

    3. From Perspectives on the control workstation:
      1. On the SP Perspectives Launch Pad, choose Hardware Perspective.
      2. Highlight the node you are working with.
      3. Press the Action Menu on the top menu bar and select the Power Off, Reset, Shutdown, or Fence Nodes option.
      4. In the Power Off options, choose Reset.
      5. In the Shutdown options, choose none.
      6. Press Apply.

  7. When the LED hits 260 or 262, open an slterm/tty connection to the node.
    1. From the command line on the control workstation, enter:
      	slterm -w <frame#>  <node_slot#>
      

      <frame#> is the number of the frame and <nodeslot#> is the number of the slot the node is in within the frame.

    2. From Perspective on the control workstation:
      1. On the SP Perspectives Launch Pad, choose Hardware Perspective.
      2. Highlight the node you are working with.
      3. Press the Action Menu on the top menu bar and select the Open TTY option.

  8. Select option 1 from the Main Menu.
    	-------------------------------------------
    	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 press "ENTER"
    	(Use the "Backspace" key to correct errors)
    	-------------------------------------------
    
  9. Select the correct boot device.

    On the thin nodes, the choice is usually the built-in Ethernet. On the wide nodes, the choice is usually the High Performance LAN Adapter or the BNC Ethernet.

    	-------------------------------------------
    	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, High Performance LAN adapter
    		3. Ethernet: Slot 1/1, 10/100 Mbs Ethernet TX MC Adapter
    		(Autosense)
    		4. Ethernet: Slot 1/1, 10 Mbs half-duplex Ethernet TX MC
     		Adapter
    		Page 1 of 2
    		88. Next Page of Select BOOT (Startup) Device Menu
    		99. Return to Main Menu
    	Type the number for your selection, then press "ENTER"
    	(Use the "Backspace" key to correct errors)
    	-------------------------------------------
    
  10. Set the client address and the bootp server address.
    1. Select option 1 as shown in the example below and enter the node's IP address leading with zeroes.
    2. Select option 2 and enter the control workstation's IP address leading with zeroes.
    3. Leave the gateway path address as zeroes.
    4. Press 99 to return to the Main Menu.
    	-------------------------------------------
    	Set or Change Network Addresses
    	Select an address to change
    	Currently selected BOOT (startup) device is:
    	Ethernet: Slot 0/1, High Performance LAN adapter
    	Hardware address...................................08005A75A45B
    	1. Client address			010.001.000.005
    	(address of this machine)
    	2. Bootp Server address			010.001.000.254
    	(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)
    	------------------------------------------
    
  11. Select option 3 from the Main Menu.
    	-------------------------------------------
    	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 press "ENTER"
    	(Use the "Backspace" key to correct errors)
    	-------------------------------------------
    
  12. Select option 4 from the Send Test Transmission (PING) Menu.
    	-------------------------------------------
    	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) devices is:
    	Ethernet: Slot 0/1, High Performance LAN adapter
    	Hardware address ...................................08005A75A45B
    	Select an address to change or select "4" to begin the test.
    	1. Origin address			010.001.000.005
    	2. Destination address			010.001.000.254
    	3. Gateway address			000.000.000.000
    	4. START PING TEST
    	99. Return to Main Menu
    	Type the number for your selection, and press "ENTER"
    	(Use the "Backspace" key to correct errors)
    	-------------------------------------------
    

    If you see the following screen, select 99 to return to the Main Menu. If the ping test failed, you may have a network or hardware problem that requires assistance from customer support.

    	-------------------------------------------
    	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, and press "ENTER"
    	(Use the "Backspace" key to correct errors)
    	-------------------------------------------
    
  13. Select option 4 from the Main Menu.
    	-------------------------------------------
    	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 press "ENTER"
    	(Use the "Backspace" key to correct errors)
    	-------------------------------------------
    

    After selecting option 4 from the preceding screen, the following should display:

    	-------------------------------------------
    	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, and press "ENTER"
    	(Use the "Backspace" key to correct errors)
    	-------------------------------------------
    
  14. Put the key in Normal mode.
    1. From the command line on the control workstation, enter:
      		# spmon -k normal node<#>
      
    2. From Perspectives on the control workstation:
    1. On ths SP Perspectives Launch Pad, choose Hardware Perspective.
    2. Highlight the node you are working with.
    3. Press the Action Menu on the top menu bar and select the Change Key Switch option.
    4. In the Key Switch Position, choose Normal.
    5. Press Apply.

  15. Press Enter. The following screen should display:
    	-------------------------------------------
    	STARTING SYSTEM (BOOT)
    	Booting ........ Please Wait
    	Ethernet: Slot 0/1, High Performance LAN adapter
    	Hardware address ...................................08005A75A45B
    			Packets sent 	Packets received
    	Bootp		  00003		    00003
    	--------------------------------------------
    

    The values for Packets sent and Packets received should increase at a similar rate. Next you should see TFTP packets also increasing. Once this is complete, the node will appear in the mode to which you have set the bootp_response in the SDR. If you do not begin receiving the requested output from the bootp_response, contact customer support.




[ Doc Ref: 96325043518454     Publish Date: Aug. 10, 2001]