Jump to content

SGI Origin 2000 Troubleshooting Guide

From RetroTechCollection
SGI Origin 2000 deskside MSC front panel
Origin 2000 deskside module system controller: reset and NMI buttons, AC OK, DC OK, fan high-speed and over-temperature LEDs, eight-character display, key switch and diagnostic port (SGI Deskside Owner's Guide, Figure 3-5)

Fault-finding on the SGI Origin 2000, mainly the deskside module, from SGI's Origin2000 Deskside Owner's Guide (007-3453-002), Origin2000 Rackmount Owner's Guide (007-3456-003) and Origin 2000 Redundant Power Supply Option Owner's Guide (007-3654-001), with owners' reports from the IRIX Network forum.[1][2][3] The SGI Onyx2 uses the same base module, power supply and system controller. The module system controller (MSC) display, its four LEDs and the three LEDs on the power supply are the main diagnostic tools. SGI's procedures stop at the field-replaceable unit.[4]

MSC LEDs

[edit | edit source]
MSC front panel LEDs (SGI Deskside Owner's Guide, chapter 6)
LED Meaning
AC OK (green) The system is plugged in, the rear circuit breaker is on and the MSC is receiving its 5 V auxiliary supply through the midplane
DC OK (green) Lights 3.5 seconds after the key is turned to On: the supply's +12 V, +5 V and +3.45 V and the remote 3.3 V, 2.4 V and 1.6 V are present with no error. It does not show that the voltages are accurate
Fan high speed (amber) The intake air is warmer than optimal, or a non-critical fan has failed and the others have gone to full speed
Over temperature (amber) The MSC has shut the system down for temperature or fan failure

[5]

The reset and NMI buttons work only with the key in the Diagnostic position.[5]

MSC display messages

[edit | edit source]
MSC messages (SGI Deskside Owner's Guide Table 6-1 and Redundant Power Supply Guide Table A-1)
Message Meaning
SYS OK Operating normally
POWER UP, R PWR UP Powering on from the front panel, or remotely through the MSC serial port
R PWR DN, PWR CYCL Powered off remotely; power-cycle command received
TEMP OK, PSTMP OK System temperature normal; power supply temperature normal
PS HITMP Power supply running hotter than normal
PS OT FL Power supply over temperature; system shut down
PS FAIL Power supply failed; system shut down
PFW FAIL Incoming power failed or dropped too low; system shut down
POK FAIL Power OK failure on an unidentified board
POK N 0 to POK N 3 Power OK failure on node board 1 to 4
POK RT 0, POK RT 1 Power OK failure on router board 1 or 2
OVR TEMP System over temperature; shut down
FAN FAIL A fan has failed. The system shuts down if it is fan 1, 2 or 3 (under the power supply), or one of fans 4 to 9 with a high system temperature
M FAN FL More than one of fans 4 to 9 failed; shut down
HBT TO Heartbeat timeout; an NMI then a reset follow
I2C FAIL The MSC cannot talk to the node boards; can appear while a multirack system powers on
KEY OFF, RESET, NMI Key turned to Standby; reset or NMI pressed with the key at Diagnostic
SP INT 1, SP INT 2 MSC firmware spurious timer interrupt or clock signal
RPWR FL, RPWR OK Redundant supply lost redundancy because a module failed; redundancy restored

[5][6]

SGI's redundant power supply guide says its message list supersedes the one in the owner's guides.[6]

Will not power on

[edit | edit source]

SGI's checks, in order:[7]

  1. The circuit breaker at the bottom right of the rear panel is up (on).
  2. The power cord is firmly in at both ends.
  3. With the front cover off, the cable from the power supply to the fan tray is secure.

Then read the three LEDs on the front of the power supply, visible with the front cover off:[7]

Power supply LEDs (SGI Deskside Owner's Guide, chapter 7)
LED State Meaning
Amber (AC OK) On AC is applied and the breaker is on. If it is off, check the outlet, power cord and switch, and the fan tray cable
Green (Power Good) On All outputs within specification
Green Blinking Supply overloaded, for example a 110 V system with too many node boards
Red (Fault) On Supply shut off for lack of airflow, or a system over-temperature shutdown
Red Blinking Undervoltage on +3.45 V, +5 V or +12 V. Power-cycling resets the supply

[7]

A deskside module with three or four node boards needs 208 V or more; on 120 V it is outside SGI's supported configurations.[8]

Shuts down or fans run fast

[edit | edit source]

The MSC shuts the system down and lights the over-temperature LED if two or more of the nine fans fail, if one fan fails and the intake air is hot, if a critical fan cooling the power supply or a router board fails, or if the intake air is simply too hot.[7] SGI's operating limit is 35 °C below 5,000 ft.[8]

  • With the fan high-speed LED on and no fan failed, check the room temperature and the clearances. jan-jaap, a collector on the IRIX Network forum, notes that a heavily loaded module can go to high speed on a warm day in a room without air conditioning.[9]
  • An owner whose system shut down at random with PS FAIL found that the green supply LED went out, and later saw FAN FAIL and the over-temperature LED; he moved his attention from the supply to the fan tray. Another owner suggested cleaning out dust between the supply and the midplane stiffeners.[9]
  • Do not run the system with the plastic covers off: SGI warns that the disturbed airflow can overheat it.[4]

Node board LEDs

[edit | edit source]

Each node board has two red LEDs at the top that light only on a voltage problem on that board; if the red LEDs light on every node board, SGI reads it as a system-wide power problem. The sixteen yellow LEDs are two columns of eight, one per processor. The bottom LED of each column should flicker as long as that processor is alive; the others light with load.[10]

Boot problems

[edit | edit source]
  • The system disk must be in bay 1, next to the MSC; each bay is wired to one SCSI ID, and a system disk in another bay will not boot. Put every drive back in the bay it came from.[4]
  • Pressing either reset button during boot aborts the boot.[4]
  • If some XIO slots do not work, check that the module has the node boards those slots depend on and that they are in the right slots. XIO slots 7 to 12 work only with a router board and a second node board fitted.[10][11]
  • To reach the System Maintenance Menu, press Esc when Starting up the system... appears.[7]

Diagnostic port

[edit | edit source]

The MSC has a 9600-baud alternate console on the 8-pin mini-DIN port behind the front door, mirrored by a DB-9 at the lower right of the rear panel. Use only one of them at a time. Through it the MSC accepts reset, NMI and soft power-off commands, and can move the supply's 3.45 V and 5 V outputs 5% up or down for margin testing.[5] Serial cables from older SGI deskside systems do not work on the Origin 2000's ports.[12]

Rack systems

[edit | edit source]

Each module in a rack has its own MSC, and the multimodule system controller (MMSC) in each rack talks to them. Its colour display on the primary rack, with Menu/Cancel, Execute and four cursor buttons, can power up, power down, power-cycle, reset or send an NMI to the whole configuration, so warn users before using it.[13] If one module shows I2C FAIL during power-on, SGI's procedure is to shut the configuration down, power off that module with its own rear switch, wait 30 seconds until all LEDs are dark, switch it back on and power up again from the MMSC.[3]

[edit | edit source]

References

[edit | edit source]
  1. ↑ Silicon Graphics, Origin2000 Deskside Owner's Guide, 007-3453-002, 1997 (File:SGI Origin2000 Deskside Owner's Guide (007-3453-002).pdf).
  2. ↑ Silicon Graphics, Origin2000 Rackmount Owner's Guide, 007-3456-003, 1997 (File:SGI Origin2000 Rackmount Owner's Guide (007-3456-003).pdf).
  3. ↑ 3.0 3.1 Silicon Graphics, Origin 2000 Redundant Power Supply Option Owner's Guide, 007-3654-001, 1999, pp. 5-8 (File:SGI Origin 2000 Redundant Power Supply Option Owner's Guide (007-3654-001).pdf). Failure messages and power-cycling a multirack configuration.
  4. ↑ 4.0 4.1 4.2 4.3 SGI, 007-3453-002, chapter 3, pp. 29-38. Parts reserved for SGI engineers, covers and airflow, drive bays and SCSI IDs, reset during boot.
  5. ↑ 5.0 5.1 5.2 5.3 SGI, 007-3453-002, chapter 6, pp. 59-68. MSC LEDs and switches, features, diagnostic ports, voltage margining, Table 6-1 messages.
  6. ↑ 6.0 6.1 SGI, 007-3654-001, appendix A, pp. 9-10, Table A-1. System controller messages.
  7. ↑ 7.0 7.1 7.2 7.3 7.4 SGI, 007-3453-002, chapter 7, pp. 69-77. Checklists, power supply problems and LEDs, System Maintenance Menu, MSC shutdown.
  8. ↑ 8.0 8.1 Silicon Graphics, Site Preparation for SGI 2000-Series, Origin Family, Onyx2, Octane and O2, 007-3452-004, pp. 62-63, Tables 6-3 and 6-4 (File:SGI Site Preparation for SGI 2000-Series, Origin, Onyx2, Octane and O2 (007-3452-004).pdf).
  9. ↑ 9.0 9.1 "O2000 / onyx2 PSU fail (?)", IRIX Network forum, posts by coldbug and Noguri (12-13 June 2021) and jan-jaap (8 November 2023), forums.irixnet.org/thread-2958.html. Owners' reports.
  10. ↑ 10.0 10.1 SGI, 007-3453-002, chapter 2, pp. 17-24. Node board LEDs, router and node combinations Table 2-2, XIO slots.
  11. ↑ SGI, 007-3456-003, chapter 8, p. 118. XIO board slots not functioning.
  12. ↑ SGI, 007-3453-002, p. 46. Serial cables.
  13. ↑ SGI, 007-3456-003, chapter 7, pp. 99-103. MMSC, display panel and rack power commands.


Retrieved from "https://retrotechwiki.com/index.php?title=SGI_Origin_2000_Troubleshooting_Guide&oldid=29262"