SGI Origin 200 Troubleshooting Guide

Origin 200 front panel: status LED, front power switch, NMI button and reset button (SGI Owner's Guide, Figure 4-1)

Fault-finding on the SGI Origin 200, from SGI's Origin200 Owner's Guide (007-3415-003) and Origin 200 and Origin 200 GIGAchannel Maintenance Guide (007-3709-002). The Origin 200 has no graphics, so its two diagnostic channels are the three-colour status LED on the front panel and the module system controller (MSC), which answers on the AUX port at the rear even when the system is switched off. SGI's procedures stop at the field-replaceable unit: fan, power supply, logic carrier, SCSI backplane and MSC.

Status LED

In a two-module system only the master module's LED shows messages.

Status LED messages (SGI Maintenance Guide Table 4-1, Owner's Guide Table 9-1)

LED	Meaning	SGI's action
Red, amber, green, then off, about half a second each, when the master switch is turned on	MSC is working	None
Green, blinking	System booting	None
Green, blinking while running	One half of a redundant power supply has failed	Replace the redundant supply when convenient
Green, solid	System on	None
Amber, blinking, then shutdown	Normal power-off	None
Amber, solid	A fan has failed	Replace the fan
Red, blinking during power-off	Power supply fault detected	Replace the power supply
Red, blinking, then shutdown	Too hot; system shutting down	Check intake air is below 40 °C, airflow is clear and all fans run
Red, blinking during boot, then shutdown	Power supply or on-board voltage regulator failed	Replace the power supply; if that does not cure it, the logic carrier
Red, solid, then shutdown	Too hot, or the power supply or on-board regulator failed	Check the temperature with the MSC; replace the power supply, then the logic carrier

Using the module system controller

Connect a terminal or a computer with a terminal emulator to the DIN-8 AUX port, using the same cables as the Indy, Indigo, Indigo2 and CHALLENGE S. Set 9600 baud, 8 data bits, 1 stop bit, no parity, and no flow control or XON/XOFF. Type Ctrl+T and press Enter; the MSC answers MSC>. If nothing comes back, check the cable and settings, check the rear master power switch is on, and on a two-module system use the master module's AUX port.

Power, reset and NMI commands need supervisor mode: Ctrl+T pas xxxx, where the factory password is none. A command refused for want of supervisor mode returns err perm; an unknown command returns err cmd, and a bad argument err arg.

MSC commands useful in fault-finding (SGI Owner's Guide, Table 4-5)

Command	Reply or action
tmp	n normal, o over normal, h dangerously high
fan	n normal speed, h high speed; ok f xxx if a fan has failed
fan h, fan n	Force high or normal fan speed
pwr	u main supply on; d only the MSC's 5 V is on
pwr u, pwr d	Turn the main supply on or off (no IRIX shutdown)
rst, nmi	Hardware reset; nonmaskable interrupt
aut, aut 1, aut 0	Show, enable or disable power-on when the master switch is turned on
ver	MSC firmware revision
sel cpu, see cpu	Route MSC input to, or show output from, one CPU (for example 1a)

The three-digit code after ok f shows which fans have failed: 001 is fan 1, 010 fan 2, 100 fan 3, 011 fans 1 and 2, 101 fans 1 and 3, 111 all three. Fan 1 is behind the 5.25-inch bay and fan 3 is nearest the PCI backplane.

Will not power on

1. Check the rear master power switch. When it is turned on the status LED flashes green briefly; that shows the MSC has power.
2. Press the front power switch far enough to feel a click. The LED should turn red, then green.
3. If the LED does not change colour, turn the master switch off, wait at least 7 seconds, turn it on and press the front switch again. SGI warns that the MSC may fail to initialise if the master switch is turned back on too quickly.
4. From the AUX port, pwr shows whether the MSC thinks the main supply is on, and pwr u (in supervisor mode) asks it to turn it on.
5. On a two-module system, press the switch on the master module; the slave's front switch is inactive.
6. If the LED blinks red during boot or turns solid red, SGI's order is power supply first, then the logic carrier. The 1998 owner's guide also describes a power supply test: pressing "the button" lights a green LED if the supply is working, though it does not say where the button is. An owner on the IRIX Network forum described a push-button on the supply itself.

See SGI Origin 200 Capacitor Replacement Guide before opening the supply.

Shuts down hot, or fans loud

• Solid amber means one fan has failed. Below 28.5 °C intake air the other two run at high speed; above 28.5 °C with a failed fan the MSC shuts the system down.
• Fans at high speed with none failed means intake air between 28.5 and 40 °C. tmp returning o means over normal temperature; h means shut down at once.
• Above 40 °C intake air the system shuts down automatically. Below that, look for blocked vents, dust (there are no filters) and missing PCI blanking plates, which the PCI cooling path relies on.
• The temperature sensor is on the MSC.

Boot and console

• The system console is serial port 1. Use 9600 baud, 8 data bits, 1 stop bit, no parity, local echo off, and a VT100, VT102, Wyse 60, ANSI or dumb terminal type.
• Power-on diagnostics take longer with more memory; SGI gives over two minutes for 512 MB. The System Maintenance Menu then appears.
• The reset button reboots without the power-on diagnostics. The NMI button saves the system state; from the PROM it drops the CPU into power-on diagnostics (POD) mode, and a second NMI forces POD mode if a kernel debugger fails.
• The serial ports use an IBM-compatible DB9 pinout, which differs from earlier SGI DB9 ports. Appendix B of the owner's guide gives the wiring of console cables for the Indy and Indigo2, the O2 and Onyx2, the Onyx, PCs and the Macintosh.

Memory

• If hinv -t memory reports less than is fitted, shut down and reseat the DIMMs.
• DIMMs must be in matched pairs, and every module needs a pair in bank 0; on a two-module system both modules need memory.

Drives

• 3.5-inch drive IDs are set by slot position, ID 1 at the bottom (the system disk) to ID 6 at the top. On power-up the drives are started in order from slot 1, then the 5.25-inch devices.
• 5.25-inch devices are on SCSI channel 1 with IDs set by jumpers on the drives.

Two-module systems

• The master and slave roles are set by a DIP switch on each MSC. Only the master has a console, AUX port and working front switch.
• Both modules must be on the same grounded power source. SGI warns that a difference in ground potential damages the main system board.

GIGAchannel expansion cabinet

GIGAchannel LEDs (SGI Maintenance Guide, Tables 4-2 and 4-3)

LED	Meaning	SGI's action
Status, green blinking	Normal power-on	None
Status, amber blinking	Half of a redundant supply has failed	Replace the supply when convenient
Status, red blinking, cabinet shuts down	Blower, power supply or overheat fault	Read the three diagnostic LEDs
Diagnostic, top or left	Blower failed	Turn off at once and replace the blower
Diagnostic, middle	Power supply or midplane voltage regulator failed	Replace the supply, then the midplane
Diagnostic, bottom or right	Overheated	Keep ambient between 5 and 35 °C and clear the airflow

The diagnostic LEDs are inside the cabinet and visible through the rear vent at the far right (or top). On the Crosstown board, each XIO port has link and data LEDs, which should all be on and steady. If they are dark, check that CPU module connector A goes to cabinet connector A and B to B, that the thumbscrews are tight, then swap the two XIO cables. Two fuses in the cabinet's power inlet housing protect its supply.

Ethernet

The green LED on the RJ-45 port flashes when the server transmits; the yellow LED flashes on any network traffic. 100Base-TX needs Category 5 cable.

Related pages

• SGI Origin 200
• SGI Origin 200 General Maintenance
• SGI Origin 200 Capacitor Replacement Guide
• SGI Origin 200 Documentation
• Capacitor Failure Symptoms