How to unconfigure and configure a cpu/memory board using cfgadm on Sun Enterprise E6800/E6900
To un-configure a cpu/memory board from system
1. Unconfigure each bank separately
# cfgadm -c unconfigure ac#:bank#
To get ac#:bank#, run :
# cfgadm -s cols=ap_id:info:type
The above command will associate the address controller(ac) number with the slot number, because the device names for memory are assigned chronologically, not by board number.
2. Unconfigure the board
Unconfiguring the board will prompt solaris to remove all processes from the cpu’s of the unconfigured boards and reschedule them on different cpu’s. Therefore, it is not necessary to issue a command to bring the cpu’s offline.
# cfgadm -c unconfigure sysctrl#:slot#
3. Disconnect the board
The board may be removed from the system now. This should be performed in a “fluid motion” and take less then one second. Insertions or removals taking longer may stall the bus and crash the system.
# cfgadm -c disconnect sysctrl#:slot#
To configure a cpu/memory board into system
1. Connect board
# cfgadm -c connect sysctrl#:slot#
2. Configure board
# cfgadm -v -c configure sysctrl#:slot#
3. Test memory on board (you must do this for both banks of memory)
If you do not test memory before configuring memory, this may cause an “unknown” condition when you run “prtdiag” or “cfgadm -l”
# cfgadm -o configure ac#:bank#
# cfgadm -o configure ac#:bank#
4. Configure memory (you must do this for both banks of memory)
# cfgadm -c configure ac#:bank#
# cfgadm -c configure ac#:bank#
5. Bring cpu’s back online
psrinfo command will give you the cpu #’s that are offline.
# psradm -n # #
To confirm cpu’s came online :
Follow the procedure below in case you have forgotten or want to reset the existing XSCF user password. (This procedure needs the user to be physically present at the server)
1. Connect directly to the serial port using the serial cable and a laptop.
2. Configure the settings on the terminal program you are using to have the following settings:
Baud rate: 9600 bps Data length: 8 bit Parity: None Stop: 1 bit Flow control: None Delay: Except for 0
3. When the login prompt appears, enter default for the login name:
Default user account: default Default password:
The default password is not input directly on the keyboard. Instead, after the default user account is input, the mode switch of the operator panel is operated as follows.
Change the panel mode switch to Service and press return…
(Operation : Locked state -> Service -> Return)
Leave it in that position for at least 5 seconds. Change the panel
mode switch to Locked, and press return…
(Operation : Wait more than 5 seconds -> Service state -> Locked -> Return) XSCF>
This mode switch operation is done within one minute. When one minute is passed, the authentication timeout occurs.
4.Use the showuser -l command to confirm your current user info.
User Name: xxxxx UID: 100 Status: Enabled Minimum: 0 Maximum: 99999 Warning: 7 Inactive: -1 Last Change: Mar xx, 20xx Password Expires: Never Password Inactive: Never Account Expires: Never Privileges: useradm platadm fieldeng
5.Use the deleteuser command to delete your forgotten user or create a new user for XSCF
6. Using the commands below, set user account, password and privileges.
XSCF> deleteuser xxxxx XSCF> adduser xxxxx XSCF> password xxxxxxxx XSCF> setprivileges xxxxxx
7. You can use the exit command to logout and try your new user account.
The post includes front view, top view and rear view of SPARC T5-2, SPARC T5-4 and SPARC T5-8.
The post includes front view, top view and rear view of SPARC T4-1, SPARC T4-2 and SPARC T4-4.
The post includes front view, top view and rear view of SPARC T3-1, SPARC T3-2 and SPARC T3-4.
The post includes front view, top view and rear view of T1000 / T2000 / T5120 / T5140 / T5220 / T5240 / T5440.
The post provides a description of forcing a crash dump on T1000/T2000 servers from ALOM. Typically, when a system hang occurs, it is required to collect a crash dump. In such case, the ALOM break command is the command to drop the system to OBP (ok prompt), which then allows user to run “ok> sync” to save crash dump. However, if “break” command can’t drop the system to ok prompt, the ‘-D’ option of the command may allow the user to get a coredump directly. This is new option is introduced with system firmware 6.3.0 or later and also requires Solaris [TM] 10 Kernel Update 118833-24 or later.
Options for the ‘break’ command:
-D: Forces a panic coredump of the managed system OS (not supported by all OS versions). -y: Instructs ALOM CMT to proceed without first asking the confirmation question: Are you sure you want to send a break to the system [y/n]? -c: Instructs ALOM CMT to connect to the system console after performing the operation.
sc> break -Dyc SC Alert: SC Request to Dump core host. Enter #. to return to ALOM. 100% done: 53847 pages dumped, compression ratio 5.54, dump succeeded rebooting...
You will see the following panic message and FMA MSG-ID: SUNOS-8000-0G when the command is executed.
Feb 8 17:35:27 eslab63 unix: [ID 760255 kern.warning] WARNING: Panic - Error Descriptor 0x5 invalid in non-resumable error handler Feb 8 17:35:27 eslab63 genunix: [ID 843051 kern.info] NOTICE: SUNW-MSG-ID:
SUNOS-8000-0G, TYPE: Error, VER: 1, SEVERITY: Major Feb 8 17:35:29 eslab63 unix: [ID 836849 kern.notice] Feb 8 17:35:29 eslab63 panic[cpu0]/thread=2a10001fcc0: Feb 8 17:35:29 eslab63 unix: [ID 400509 kern.notice] Unrecoverable hardware error Feb 8 17:35:29 eslab63 unix: [ID 100000 kern.notice] Feb 8 17:35:29 eslab63 genunix: [ID 723222 kern.notice] 000002a10001f6e0 unix:process_nonresumable_error+224 (2a10001f8d0, 0, 107c000, 40, 0, 5)Feb 8 17:35:30 eslab63 genunix: [ID 179002 kern.notice] %l0-3: 0000000000000040 0000000003000000 0000000000000001 0000000000000000 Feb 8 17:35:30 eslab63 %l4-7: 000000000180c5c0 0000000100000000 00000000ffffffff 0000000000000001 Feb 8 17:35:30 eslab63 genunix: [ID 723222 kern.notice] 000002a10001f820 unix:ktl0+64 (0, 0, d77e, ffffffffffffffff, 0, 12) Feb 8 17:35:30 eslab63 genunix: [ID 179002 kern.notice] %l0-3: 000000000180c000 0000000000000000 0000000000001406 0000000001023534 Feb 8 17:35:30 eslab63 %l4-7: 0000000000000000 0000000000000000 0000000000000000 000002a10001f8d0 Feb 8 17:35:31 eslab63 genunix: [ID 723222 kern.notice] 000002a10001f970 unix:cpu_halt+b8 (0, 0, 300013c8000, 16, 180c000, 1) Feb 8 17:35:31 eslab63 genunix: [ID 179002 kern.notice] %l0-3: 000000000184ca08 0000000000000001 0000000000000002 0000000000000000 Feb 8 17:35:31 eslab63 %l4-7: 0000000000000000 0000000000000000 0000000000000000 000000000103af04 Feb 8 17:35:31 eslab63 genunix: [ID 723222 kern.notice] 000002a10001fa20 unix:idle+128 (1819c00, 10, 180c000, ffffffffffffffff, 1, 1818800) Feb 8 17:35:32 eslab63 genunix: [ID 179002 kern.notice] %l0-3: 0000000001846420 000000000000001b 0000000000000000 ffffffffffffffff Feb 8 17:35:32 eslab63 %l4-7: 0000000000000000 0000000000000000 0000000000000000 000000000103af04 Feb 8 17:35:32 eslab63 unix: [ID 100000 kern.notice] Feb 8 17:35:32 eslab63 genunix: [ID 672855 kern.notice] syncing file systems... Feb 8 17:35:32 eslab63 genunix: [ID 733762 kern.notice] 1 Feb 8 17:35:34 eslab63 genunix: [ID 904073 kern.notice] done Feb 8 17:35:35 eslab63 genunix: [ID 111219 kern.notice] dumping to /dev/dsk/c1t0d0s1, offset 429654016, content: kernel Feb 8 17:36:18 eslab63 genunix: [ID 409368 kern.notice] 100% done: 53847 pages dumped, compression ratio 5.54, Feb 8 17:36:18 eslab63 genunix: [ID 851671 kern.notice] dump succeeded
The post includes front view, top view and rear view of Mx000 series servers – M3000, M4000, M5000, M8000 and M9000.
What is XCP?
XCP stands for XSCF (eXtended System Control Facility) Control Package. The XCP software is composed of :
1. eXtended System Control Facility : software which includes specific features implemented for OPL, including features such as the Fault Management Architecture, Dynamic Reconfiguration, Capacity On Demand, XSCF CLIs, etc and POST, This includes the base linux kernel and the associated linux packages (aka ELSC),
2. OpenBoot PROM.
XCP version format
XCP version format is XCP XYYZ (for example XCP 1060 – X(1), YY(06), Z(0)) where :
X : Major version 1-9
YY : Minor version 00-99
Z : Micro version 0-9
Each XCP release number is obviously unique.
How to check current version
In order to display the information about
XSCF> version -c xcp -v -t XCP: 1060 OpenBoot PROM : 01.30.0000 XSCF : 01.06.0001
XSCF> version -c xcp XSCF#0 (Active ) XCP0 (Current): 1060 XCP1 (Reserve): 1060 XSCF> version -c xscf XSCF#0 (Active ) 01.06.0001(Current) 01.06.0001(Reserve)
XSCF> version -c cmu DomainID 0: 01.30.0000 DomainID 1: 01.30.0000
The XCP firmware procedure exclusively runs on the
XSCF> version -c xcp XSCF#0 (Standby) ---> XSCF#0 is Standby XCP0 (Current): 1100 XCP1 (Reserve): 1100 XSCF#1 (Active) ---> XSCF#1 is Active XCP0 (Reserve): 1100 XCP1 (Current): 1100
1. updating firmware requires a user with either
2. run getflashimage, to import the new XCP firmware onto the main XSCF unit.
XSCF> getflashimage -u root ftp://10.169.233.60/M9000_fw/DCXCP1116.tar.gz
Verify the download of image :
XSCF> getflashimage -l
3. To check if firmware can be updated with domains running, run:
XSCF> flashupdate -c check -m xcp -s 1116
4. To actually update the firmware, run
XSCF> flashupdate -c update -m xcp -s 1116
5. Wait until the XSCF serial port or showlogs monitor output indicates the update has completed:
XSCF> showmonitorlogs .... Aug 13 07:25:48 Mx000_1-1-0 monitor_msg: SCF:XCP update has been completed (XCP version=1116)
Final XSCFs configuration is as follows; please note that Active/Standby roles and Active/Reserve banks are now switched:
XSCF> version -c xcp XSCF#0 (Active ) ---> XSCF#0 is Active XCP0 (Reserve): 1116 XCP1 (Current): 1116 XSCF#1 (Standby) ---> XSCF#1 is Standby XCP0 (Current): 1116 XCP1 (Reserve): 1116