15k 25k servers

Today I want to cover 15K and 25K Sun Fire Servers.

=================================

Reliability, availability, and serviceability (RAS) are critical requirements of

customers who deploy business-critical applications. The Sun Fire 15K/12K systems build upon the industry-leading RAS capabilities. 

Sun Fire 15K server, a high-level Solaris-based server designed to complete with mainframes and other high-end servers. The Sun Fire 15K server has a baseline configuration of 106 processors, 576 gigabytes of RAM, and 18 I/O hubs for networking and storage connectivity, making it the world's largest single-cabinet UNIX server system..

Sun Fireplane interconnect that offers throughput of 172 Gigabytes per second. Sun Fire 15K can have up to 106 64-bit UltraSPARC III processors with the ability to scale beyond 1,000 CPUs. With "hot" CPU upgrades, the Sun Fire 15K server enables resources to be added easily while the system is up and running.

System Management Services 1.6 

support solaris 10 for platlform operating system 

UltraSPARC™ IV+ 1800 MHz processors

System Firmware version 5.2

Dual system controller boards are supported within the Sun Fire high-end systems. One board is designated as the primary or main SC, and the other is designated as the spare SC. If the main system controller fails, the failover capability automatically switches to the spare system controller.

On the SC for a Sun Fire high-end system, a board can be in one of four states:

unavailable, available, assigned, or active. You can use the showboards(1M) command to view the state of a specific board.

15K/25K

======

18 CPU/Memory boards and 18 I/O boards in the Sun Fire 15K system

Bulk power supply (3 in front, 3 in rear)

12K/20K

=====

9 CPU/Memory boards and 9 I/O boards in the Sun Fire 12K system

(with 9 CPU and I/O filler panels in the rear of the system)

The CPU/Memory board holds four CPUs. Each CPU has an associated memory

subsystem of eight DIMMs, so memory bandwidth and capacity are both scaled up as CPUs are added. The memory capacity of the board is 32 Gbytes using a 1-Gbyte DIMM. The maximum memory bandwidth inside a board is 9.6 Gbytes per second.

The CPU/Memory board has a 4.8 Gbyte per second connection to the rest of the system.

The Sun Fire 15K system can have up to 18 domains. The Sun Fire 12K system can have up to 9 domains. Domains are isolated from each other by the interconnect application-specific integrated circuits (ASICs).

There are four fan trays above and four fan trays below the system boards. In Sun Fire 15K/12K systems, each fan tray contains two layers of six-inch fans. The fans have three speeds and normally run at high speed. If any of the sensed components in the system overheat, all fans are set to super-high speed. If a single fan fails, the redundant fan in the corresponding layer of the tray switches to super-high speed.

The fans are N+1 redundant, enabling the system to run with a failed fan. The fan trays can be hot-swapped while the system is running.

Performance:

========

Sun is claiming a performance for the Sun Fire 15K of nearly 2.5 times the MIPS rating (millions of instructions per second) of the largest commercially available mainframe.



Use hpost -?<level> to see the differences.

sms-svc> hpost -?

Following are some interesting values of level: 
7       The minimum level:  hardware probing and configuration, no testing. 
16      Default level: basic testing of processor, memory, and I/O. 
24      Adds some additional locations for some tests (e.g., functional 
        and memory structure testing). 
32      Some CPU Functional tests added, and all locations are tested 
        in all structures except for Memory & ECache.  Memory adds basic 
        MATS+ test of all of memory using As/5s only. 
64      The highest "normal" diag level:  adds all locations and patterns 
        for all tests except extensive Memory and ECache tests.  Includes
 advanced CPU functional tests, MP Testing and error detection tests. 
96      Adds all patterns/locations for Memory & ECache tests. 
127     The maximum level:  Memory & ECache Data Validation, ECache stress.       

Question:
How can I change my post run level to run higher level diagnostics during a setkeyswitch on?
Answer:
The post level can no longer be modified on the command line. To run a higher level post, you will need to add “level ” into the .postrc file, and then run “/opt/SUNWSMS/bin/setkeyswitch -d a on”.

NOTE: The .postrc file can be located in the current working directory, /etc/opt/SUNWSMS/config//.postrc, or /etc/opt/SUNWSMS/config/platform/.postrc.  The .postrc will be searched for in that order.