Sun Microsystems UltraSPARC Prototype Restoration

4/12/23

We powered on the system with a monitor, keyboard, and mouse connected. The monitor said that there was no video signal. This was expected.

We could not get the ILOM to connect to a network, and my new laptop doesn't have a wired Ethernet connector. I tried an older HP laptop, but it did not see an Ethernet link. We were told by the donor that the ILOM was finicky and would not connect to many different types of Ethernet switches and routers.

I forgot to bring a Cisco DB-9 to RJ-45 console cable so a serial connection to the ILOM will have to wait until Saturday.

We spent a few hours inspecting the system and taking pictures for documentation.

4/14/23

ILOM Notes:

The /SYS/MB/CMP0/BR[1..4]/CH0/PRSNT and /SYS/MB/CMP0/BR[1..4]/CH0/T_AMB are the 4x Memory DIMMs

-> version

SP firmware 2.0.4.0

SP firmware build number: 38600

SP firmware date: Sun Mar 29 16:15:06 PDT 2009

SP filesystem version: 0.1.14

4/15/23

After a long hunt I found the bag of Cisco console cables I had at home. We connected the RJ-45 end of the Cisco cable to the SER MGT port on the back of the Wollemi, connected DB-9 end to a USB RS-232 serial cable, loaded the PuTTY terminal emulator, set the serial configuration to 9600, N, 8, 1, and powered on the Wollemi. We were rewarded with a U-Boot banner, and lots of console output.

We noticed that the power supply can make 220 Watts of 12V, and that the CMOS configuration says that the CPU can use up to 600 Watts.

4/19/23

We have another clue. The ILOM command "show /SYS" displayed:

        type = Host System

        keyswitch_state = Normal

        product_name = Sun Blade T6xxx Server Module

        product_part_number = 541-3395-01

        product_serial_number = 1005LCB-09139M002Y

        product_manufacturer = SUN MICROSYSTEMS

That confirms that the Willomi is based on the T6xxx series of rack mount servers. The T6320 also has an ILOM, so the documentation from that machine will help a little.

The ILOM log contains the log from an installation of Solaris 11 and NetBSD. The output from booting those operating systems give is some good clues about the architecture of the I/O part of the machine. It is similar, and also very different than the T6320 server.

We looked at the voltages when the ILOM is powered and the CPU is not;

/SYS/MB/V_VTT = 0 (should be 0.78V)

/SYS/MB/V_VDDIO = 0 (should be 1.1V)

/SYS/MB/V_VCORE = 0 (should be 1.135V)

/SYS/MB/V_+1V5_VDD = 0 (should be 1.5V)

/SYS/MB/V_VMEM = 0 (should be 1.5V)

/SYS/MB/V_+3V3_STBY = 3.322V (should be 3.35V)

/SYS/MB/V_+12V0_MAIN = 0 (should be 12.0V)


/SYS/IOB/V_+1V0_VDD = 0 (should be 1.0V)

/SYS/IOB/V_+1V2_VDD = 0 (should be 1.2V)

/SYS/IOB/V_+1V8_VDD = 0 (should be 1.805V)

/SYS/IOB/V_+3V3_MAIN = 0 (should be 3.35V)

/SYS/IOB/V_+5V0_VCC = 0 (should be 4.98V)

/SYS/IOB/V_+12V0_IO = 0 (should be 12.0V)


4/22/23

This looks like it might be a problem. The power supply can make 200W and the system is configured for a maximum of 600W.


 /SP/powermgmt

    Targets:


    Properties:

        actual_power = 0.00

        permitted_power = 600

        available_power = 600

        policy = performance


We tried disabling 48 of the 64 "CPU" to reduce the power requirements. After fiddling with the CPUs the Service Indicator on the front of the system is lit. Disabling the CPUs made no difference, so we reenabled all of the "CPUs" and the Service Indicator went out.


/SYS/ and /SYS/MB are listed as faulted.

-> show /SP/faultmgmt


 /SP/faultmgmt

    Targets:

        0 (/SYS/MB)


4/26/23

We Googled looking for ways to break out of the ILOM shell to the Linux shell. In the process, we listed the properties of /SYS/SP and found:

 /SYS/SP