Jump to content

SGI Origin 3000 Troubleshooting Guide

From RetroTechCollection
SGI Origin 3000 C-brick rear panel
Origin 3000 C-brick rear panel: power switch, PWR connector, 48 VDC and 12 VDC LEDs, LINK and XIO connectors, console and L1 ports, and the CPU activity LEDs (SGI Owner's Guide, Figure 4-2)

Fault-finding on the SGI Origin 3000 series, from SGI's SGI Origin 3000 Series Owner's Guide (007-4240-002).[1] Each C-, I-, P-, X- and R-brick has its own L1 controller with a display; tall racks add an L2 controller with a touch display on the front door, which shows the messages from every L1 in the rack.[2] The guide treats most repairs as work for SGI's engineers, so what it offers an owner is the controllers' messages and LEDs.[3]

Reaching the controllers

[edit | edit source]
  • Without an L2 controller (a small Origin 3200), connect a terminal to the DB-9 console port of the lowest C-brick that has an I-brick attached.[4]
  • With an L2 controller, connect a terminal to the L2's console port (38.4 kbaud) or an SGIconsole to its Ethernet port.[4][5]
  • At an L1 prompt such as 003c01-L1>, Ctrl+D switches to the system console and Ctrl+T back to the L1. Console mode from the L1 is not available in a system with an L2 controller.[5]
  • config lists the bricks a controller can see. A command prefixed with * goes to every brick; cti, ctc and ctci send it to the attached I/O brick, the attached C-brick, or that C-brick's I/O brick.[5]
  • Each L1 keeps a log of local events, shown by log.[5]

Brick LEDs

[edit | edit source]
  • On/off switch LED, green: the brick's internal components are on.
  • Service-required LED, orange: something has failed, for example a fan, and the brick is still running.
  • Failure LED, red: a failure has occurred and the brick is down.

[6] At the rear of the C-brick, the 12 VDC LED lights green when the L1 controller is powered (the rear power switch must be at 1), and the 48 VDC LED when the rest of the brick is on.[7] The LINK and XIO connector LEDs light yellow when the cable is seated and green when the link is established.[8]

Power supply LEDs

[edit | edit source]
Power bay supply LEDs (SGI Owner's Guide, Table 10-1)
Condition Power (green) PFAIL (amber) FAIL (amber)
No AC to any supply Off Off Off
No AC to this supply Off Off On
AC present, standby voltage on Blinking Off Off
DC outputs on On Off Off
Supply failure Off Off On
Current limit reached on 48 V output On Off Blinking
Predictive failure On Blinking Off

[9]

The L1 controller of a connected brick can read the power bay's status but cannot control it.[9] The fault reset switch at the rear of the bay resets it.[9] SGI's guide reserves supply replacement for its own engineers.[9]

L1 messages

[edit | edit source]

SGI counts a voltage more than 10% from nominal as a warning and more than 20% as a fault.[6]

L1 controller messages (SGI Owner's Guide, Table C-1)
Message Meaning
ATTN: x.xV high (or low) fault limit reached @ x.xxV A voltage is 20% out; 30-second power-off of the brick, or of the system if there is no backup
ATTN: x.xV high (or low) warning limit reached @ x.xxV A voltage is 10% out
ATTN: x.xV level stabilized @ x.xV The voltage is back within limits
ATTN: FAN # x fault limit reached @ xx RPM A fan has reached its maximum speed; the intake air may be too hot, or a fan has failed
ATTN: FAN # x warning limit reached @ xx RPM A fan has speeded up; check the room temperature and whether the fan settles
ATTN: TEMP # advisory / critical / fault temperature reached Intake air above 30, 35 or 40 °C (27, 31 or 35 °C at the high-altitude setting)
Auto power down in xx seconds The L1 has registered a fault and is shutting down; repeated every 5 seconds
Brick appears to have been powered down The L1 has shut the brick down after a fault

[6]

env shows each monitored voltage, fan and temperature with its limits. On a C-brick SGI's example lists DIMM, 2.5V, 1.5V, 3.3V aux, 3.3V L1, 5V aux and 12V bias rails, and SRAM, CPU, 1.5V, 3.3V aux, 5V aux and 12V bias rails for each PIMM.[10] env reset clears current warnings and faults.[10]

Will not power on

[edit | edit source]
  1. Check the cables from the PDU or PDS to the wall, and that the PDU or PDS circuit breaker is on.[4]
  2. Check that the power switch at the rear of each brick is at 1, and that the brick's 12V LED is lit. If it is not, the L1 controller has not started.[4]
  3. For a D-brick, check that its RUN/SERVICE key switch is at RUN.[4]
  4. Check the power supply LEDs in the power bay against the table above, and that at least two supplies are fitted.[9]
  5. At the L2 prompt run config to see which bricks the controller can find; the L1 display of each brick should show L1 running once power-on starts.[4]
  6. Look up any message on the display in the table above.[4]

Shutdowns and false readings

[edit | edit source]

A brick that shuts down on a voltage or temperature fault reports it on its L1 display and in its log; read both, and run env to see which reading is out.[6][10] jan-jaap, a collector on the IRIX Network forum, recalls that faulty Dallas DS1780 monitoring chips were already a problem in Origin 3000 bricks when the systems were new.[11] On the related SGI Origin 350 such chips give readings that cannot be real; see SGI Origin 350 Troubleshooting Guide for how owners have traced them.

PCI slots

[edit | edit source]

The P-brick's power board supplies about 17.5 W per slot, while a card can draw up to 25 W. The L1 controller powers the slots in order from the lowest number until the power is used up, and prints a console message if there is not enough for every card; look for that message before condemning a card in a high-numbered slot.[12]

Booting

[edit | edit source]
  • The reset switch on a C-brick resets its processors and ASICs and loses memory contents; the NMI switch resets them without losing memory.[6]
  • The power-on self-tests automatically deallocate processors and memory that fail them.[2]
  • Each brick's failure LEDs show where in the PROM code the system stopped during boot; once IRIX is running they show processor activity.[2]
[edit | edit source]

References

[edit | edit source]
  1. ↑ Silicon Graphics, SGI Origin 3000 Series Owner's Guide, 007-4240-002, December 2001 (File:SGI Origin 3000 Series Owner's Guide (007-4240-002).pdf).
  2. ↑ 2.0 2.1 2.2 SGI, 007-4240-002, chapter 1, pp. 6 and 23-24. Controllers, RAS features: brick shutdown on overheating, failure LEDs, deallocation after self-test failures.
  3. ↑ SGI, 007-4240-002, chapter 11, pp. 211-212. Maintenance by owners and by SGI engineers.
  4. ↑ 4.0 4.1 4.2 4.3 4.4 4.5 4.6 SGI, 007-4240-002, chapter 2, pp. 25-47. Console connection options, preparing to power on, powering on from the L2.
  5. ↑ 5.0 5.1 5.2 5.3 SGI, 007-4240-002, chapter 3, pp. 80-81 and 105-108. L2 controller ports, operating the L1, command targeting, log, console mode.
  6. ↑ 6.0 6.1 6.2 6.3 6.4 SGI, 007-4240-002, appendix C, pp. 287-290, Table C-1. L1 controller panel, LEDs and messages.
  7. ↑ SGI, 007-4240-002, chapter 4, pp. 118 and 123-125. C-brick rear panel LEDs and power switch.
  8. ↑ SGI, 007-4240-002, chapter 12, p. 214. C-brick connector LEDs.
  9. ↑ 9.0 9.1 9.2 9.3 9.4 SGI, 007-4240-002, chapter 10, pp. 199-210, Table 10-1. Power bay, power supply LEDs, power supply configuration, fault reset switch.
  10. ↑ 10.0 10.1 10.2 SGI, 007-4240-002, appendix B, pp. 239-241, Example B-18. env output and env reset.
  11. ↑ "Origin 350 - PCI 5V missing", IRIX Network forum, post by jan-jaap, 27 September 2023, forums.irixnet.org/thread-4064.html.
  12. ↑ SGI, 007-4240-002, appendix A, p. 227. P-brick slot power.


Retrieved from "https://retrotechwiki.com/index.php?title=SGI_Origin_3000_Troubleshooting_Guide&oldid=29518"