Skip to main content
  1. Home >
  2. Products >
  3. Computing Products >
  4. Servers >
  5. Fujitsu SPARC servers >
  6. Downloads >
  7. User Manuals >
  8. Fujitsu M10/SPARC M10 Systems Product Notes for XCP Version 2410 >
  9. Notes on Maintenance of a CPU Memory Unit, Motherboard Unit, XSCF Unit, PSU Backplane, or Crossbar Backplane Unit

Notes on Maintenance of a CPU Memory Unit, Motherboard Unit, XSCF Unit, PSU Backplane, or Crossbar Backplane Unit


Notes on Maintenance of a CPU Memory Unit, Motherboard Unit, XSCF Unit, PSU Backplane, or Crossbar Backplane Unit
  1. Note on correcting the time (1)
    After the replacement of the FRU and the micro SD card, and after the system has been running for a long time, an XSCF reboot followed by PPAR startup may cause the Oracle Solaris time to shift.
    This problem occurs when all of the following conditions are true:
    - Both the FRU (CPU memory units, motherboard unit, XSCF unit, PSU backplane, or crossbar backplane unit) and the micro SD card are replaced.

    - System: SPARC M10-1, SPARC M10-4, or SPARC M10-4S (single-chassis configuration)

    - The setting specifying the XSCF as an NTP client is enabled.

    - XCP 2240 or later


    To avoid this problem, be sure to perform the following procedure after replacing the FRU and the micro SD card.
    [Workaround]
  1. 1. Execute the showlogs event command to check whether the XSCF time and NTP server time are successfully synchronized.
    If multiple event logs are registered, see the latest event log.
XSCF> showlogs event -rM
Date                       Message
Mar 25 11:45:31 JST 2017   NTP service found NTP server to synchronize.
  1. When "NTP service found NTP server to synchronize." appears, the time is correctly synchronized. Then, the following steps are not necessary.
  1. 2. Execute the setntp command to temporarily disable the XSCF NTP client function.
XSCF> setntp -s client -c disable
  1. 3. Reboot the XSCF.
XSCF> rebootxscf -a
  1. 4. Execute the setdate command to set the XSCF time to the current time.
    In the following example, the XSCF time is set to 2017/3/25 12:00:00.
XSCF> setdate -s 032512002017.00
  1. The XSCF automatically reboots when the time is set.
  1. 5. Execute the setntp command to re-enable the XSCF NTP client function.
XSCF> setntp -s client -c enable
  1. 6. Reboot the XSCF.
XSCF> rebootxscf -a
Note - Even if the XSCF time and NTP server time fail to synchronize and the following message appears, this workaround has been completed. 
 
XSCF> showlogs event -rM
Date                       Message
Mar 25 12:15:31 JST 2017   NTP service failed to reach appropriate NTP server.
  1. Note on correcting the time (2)
    Due to the problem reported in RTIF2-140711-001 and RTIF2-141204-001, the Oracle Solaris time may shift at PPAR startup.
    To avoid this problem, be sure to perform the following procedure after maintenance.
Note - Before performing the procedure in "Note on correcting the time (2)," check whether the conditions in "Note on correcting the time (1)" are met. If they are met, first perform the procedure in "Note on correcting the time (1)."
  1. [Workaround]
    1. Execute the showdateinfo(8) command to check the times of the control domain and guest domain.

    - For XCP 2350 or later
    Check the times of the control domain and guest domain.
    If the times are correct, the subsequent steps are unnecessary.
    If the times are not correct, go to step 2.

    - For XCP 2340 or earlier
    Go to step 2 because it is not possible to check the times with the showdateinfo(8) command.
Note - The times of the control domain and guest domain in the XSCF time zone are displayed.
  1. Example: Physical partition is PPAR#0
XSCF> showdateinfo -p 0
PPAR-ID    : 0
config_name: 10guest_config
 
XSCF information:
-----------------

  XSCF
    Date    : Aug 03 19:56:16 JST 2017
 
Logical domains information:
----------------------------

  primary
    Date    : Aug 03 19:56:16 JST 2017
  guest 0
    Date    : Aug 03 19:56:16 JST 2017
  guest 1 
    Date    : Aug 03 19:56:16 JST 2017
  guest 2 
    Date    : Aug 03 19:56:16 JST 2017
XSCF>
  1. 2. Specify false for the OpenBoot PROM environment variable auto-boot?, and execute the setpparparam(8) command to disable the autoboot function of the control domain. Also execute the setpparmode(8) command to disable startup of guest domains.
    Example: Physical partition is PPAR#0
XSCF> setpparparam -p 0 -s bootscript "setenv auto-boot? false" -y
XSCF> setpparmode -p 0 -m guestboot=off
  1. 3. Execute the poweron(8) command to power on the target physical partition.
XSCF> poweron -p 0 -y
  1. 4. Execute showdomainstatus -p 0 -g primary, and confirm that "OpenBoot Running" appears under Status.
  1. 5. Execute the console(8) command to switch to the control domain console.
XSCF> console -p 0 -y

Omitted
{0} ok
  1. 6. Execute the boot -s command to start Oracle Solaris in single user mode.
{0} ok boot -s
  1. 7. Execute the date command of Oracle Solaris to set the control domain time.
    Example: Setting for 12:20:00 on March 25, 2017
primary# date 0325122017.00
Note - Execute the showdate(8) command to display the XSCF time, and confirm that the control domain time of the target physical partition is the same as the XSCF time.
  1. 8. Switch to multi-user mode.
 primary# exit
  1. 9. For a logical domain configuration, set the correct time for guest domains.
    For details on how to check for a logical domain configuration, see "Checking between a logical domain configuration and the factory-default configuration."
    Perform the following procedure on every guest domain to set the time.
a. Check the CONS value of the guest domain.
primary# ldm list-domain
NAME             STATE      FLAGS   CONS  VCPU  MEMORY  UTIL  NORM  UPTIME
primary          active     -n-cv-  SP    8     3968M   0.2%  0.1%  18h 37m
guest            bound      ------  5000  8     2G
b. Disable the autoboot function of the guest domain.
primary# ldm set-variable auto-boot?=false guest
c. Start the guest domain.
primary# ldm start-domain guest
d. Open the guest domain console.
primary# telnet localhost 5000
:
{0} ok
e. Execute the boot -s command to start Oracle Solaris in single user mode.
{0} ok boot -s
f. Execute the date command of Oracle Solaris to set the guest domain time.
Example: Setting for 12:30:00 on March 25, 2017
guest# date 0325123017.00
g. Switch to multi-user mode.
guest# exit
h. If the guest domain is an NTP client, execute the ntpq -np command, and confirm synchronization with the NTP server.
The asterisk ("*") at the beginning of a line below indicates that the NTP server in the "remote" column is synchronized with the guest domain.
guest# while [ 1 ]; do ntpq -np; sleep 60; done
     remote           refid      st t when poll reach   delay   offset  jitter
==============================================================================
*203.0.113.156   203.0.113.70     4 u   46   64    3    0.458    0.754   7.243
     remote           refid      st t when poll reach   delay   offset  jitter
==============================================================================
*203.0.113.156   203.0.113.70     4 u   39   64    7    0.478    1.487   5.664
     remote           refid      st t when poll reach   delay   offset  jitter
==============================================================================
*203.0.113.156   203.0.113.70     4 u   35   64   17    0.450    2.164   5.077
^C
guest#
 
i. If necessary, re-enable the autoboot function of the guest domain.
primary# ldm set-variable auto-boot?=true guest
  1. 10. If necessary, re-enable the functions that were disabled in step 1.
    For the autoboot function of the control domain, specify true for the OpenBoot PROM environment variable auto-boot?, and execute the setpparparam(8) command to re-enable the function. For startup of guest domains, execute the setpparmode(8) command to re-enable it.
    Example: Physical partition is PPAR#0
XSCF> setpparparam -p 0 -s bootscript "setenv auto-boot? true" -y
XSCF> setpparmode -p 0 -m guestboot=on 
  1. 11. For a logical domain configuration, save logical domain configuration information.
    For details on how to check for a logical domain configuration, see "Checking between a logical domain configuration and the factory-default configuration."
    Example: Saving logical domain configuration information under the name of "config_A"
primary# ldm list-domain -l
Omitted
primary# ldm list-spconfig
Omitted
primary# ldm remove-spconfig config_A
primary# ldm add-spconfig config_A
Checking between a logical domain configuration and the factory-default configuration
You can check whether the configuration is a logical domain configuration by using the showdomainconfig command.
  1. Logical domain configuration
    If a value other than factory-default is set, as indicated by underline below, it is a logical domain configuration.
XSCF> showdomainconfig -p 0
PPAR-ID     :0
Booting config
(Current)   :config_A
(Next)      :config_A
-----------------------------------------------------------------------------

Index       :1
config_name :factory-default
domains     :1
date_created:-
-----------------------------------------------------------------------------

Index       :2
config_name :config_A
domains     :1
date_created:'2014-03-05 17:33:17'
  1. factory-default configuration
If factory-default is set below, as indicated by underline, it is not a logical domain configuration but in the factory-default configuration.
XSCF> showdomainconfig -p 0
PPAR-ID     :0
Booting config
 (Current)  :factory-default
 (Next)     :factory-default
-----------------------------------------------------------------------------

Index       :1
config_name :factory-default
domains     :1
date_created:-
  1. [SPARC M10-1/SPARC M10-4/SPARC M10-4S]
    The setting information of the CPU Activation and the CPU Activation key may be deleted when the CPU memory unit lower (CMUL) or the motherboard unit (MBU) is replaced. To restore the CPU Activation setting information and the CPU Activation key, it is necessary to save the CPU Activation setting information and CPU Activation key beforehand, using the dumpconfig(8) command, and restore them with the restoreconfig(8) command.
  2. [SPARC M10-4S]
    To replace the CPU memory unit lower (CMUL) or to add the SPARC M10-4S, execute the replacefru(8) or addfru(8) command and follow the maintenance menu.
    Suppose that the system contains a crossbar box or consists of multiple SPARC M10-4S units and that the firmware is XCP 2044 or later. If the CPU memory unit lower (CMUL) is replaced or if the SPARC M10-4S is added without using the maintenance menu, the XSCF of the replaced/added device may not start.
    Should this occur, turn off the input power to the replaced/added SPARC M10-4S. Subsequently, using the maintenance menu, replace the CPU memory unit lower (CMUL) or install the SPARC M10-4S again.
  3. [SPARC M10-4S]
    From XCP 2090 onward, XCP firmware versions are automatically synchronized between chassis in the following situation: the CPU memory unit lower (CMUL) or XSCF unit is replaced or the SPARC M10-4S or crossbar box is added without using the maintenance menu and with the input power turned off.
    After replacement or expansion of components, if the "XSCF firmware update now in progress. BB#xx, please wait for XSCF firmware update complete." message is output after logging in to the master XSCF, the XCP firmware is in the process of being updated automatically.
    Automatic synchronization of the XCP firmware versions can be confirmed by referring to the "XCP firmware version synchronization completed" message, which is output by the showlogs monitor, showlogs event, or showmonitorlog command.
    Do not execute the following until the firmware update is complete. The same also applies to the XSCF Web operations corresponding to XSCF commands.
    - Turning off the input power

    - Executing the poweron, testsb, diagxbu, or reset command, or operating the POWER switch on the operation panel

    - Executing the setdate, switchscf, rebootxscf, initbb, restoreconfig, or restoredefaults command, or operating the RESET switch on the rear panel

    - Executing the getflashimage -d command

    - Executing the flashupdate -c update command
  1. If cold replacement of CPU memory unit lower (CMUL) or a motherboard unit (MBU) is done in the following procedure, the physical partitions (PPAR) may not start.
    1. Execute the "restoredefaults -c factory" command.

    2. After XSCF has stopped, turn off the input power of the system (AC OFF).

    3. Replace the CMUL or MBU without replacing the microSD card.

    4. Turn on the input power of the system.

    [Workaround]
    Replace CMUL or MBU in the following procedure.
    1. Execute the "restoredefaults -c factory" command.

    2. After you have stopped the XSCF in step 1, turn off/on the input power of the system (AC OFF/ON), and then confirm that the XSCF has started.

    3. Replace the targeted CMUL or MBU.

    [How to restore]
    Restore with the following procedure:
    1. Turn off the input power of the system (AC OFF).

    2. Temporarily mount the microSD card that was mounted on the CMUL or MBU, when the "restoredefaults -c factory" command was executed, to the new unit.

    3. Confirm that XSCF has started after turning on the input power (AC ON).

    4. Turn off the input power of the system (AC OFF).

    5. Extract the microSD card that was temporarily mounted and mount the original microSD card.

    6. Turn on the input power of the system (AC ON).