Jump to content

SGI Onyx 350 Troubleshooting Guide

From RetroTechCollection
SGI Onyx 350 InfiniteReality rack (User's Guide Figure 1-1)
Onyx 350 InfiniteReality rack: each module has its own L1 display, and the L2 controller sits on the roof (SGI User's Guide, Figure 1-1)

Fault-finding on the SGI Onyx 350, from SGI's SGI Onyx 350 Visualization System User's Guide (007-4632-001) and owners' reports. The compute module is the SGI Origin 350's, and its rail faults, monitoring-chip faults and power supply faults are set out in full on SGI Origin 350 Troubleshooting Guide; the system-wide checks and the graphics are covered below.

Order of work

[edit | edit source]
  1. Read the L1 display on each module and note which modules show the service-required (yellow) or failure (red) LED.[1]
  2. Check the power supply LEDs at the front of each compute module (table below).[2]
  3. Connect a terminal to the L1 console port, or open the L2, and run env and log on the module concerned.[3]
  4. At the L2, run cfg to see which modules the controller can find.[4]

SGI's troubleshooting chart

[edit | edit source]
Troubleshooting chart (SGI User's Guide, Table 8-1)
Problem Action
The system will not power on Check that the PDU cord is seated and the PDU circuit breaker is on.
A module will not power on Check its rear power switch if it has one, read its L1 display, check that its L1 is running, and check its connection to its power source.
Service-required or failure LED lit Read that module's L1 display and look up the message.
NUMAlink port LEDs not lit Reseat the NUMAlink cable at both ends.
PCI slot PWR LED off Reseat the card.
PCI slot fault LED on Reseat the card; replace it if the LED stays on.
Amber LED on a disk drive Replace the drive.

[5] For anything else, including a system that will not boot IRIX, SGI's chart refers to its own engineers.[5]

Power supply LED

[edit | edit source]
Compute module power supply LED (SGI User's Guide, Table 7-2)
LED Meaning
Off No AC reaching the supply: the module is unplugged or a fuse has blown
Amber In a two-supply module one supply has no AC; or a voltage, temperature or current limit has been exceeded
Blinking green AC present, main DC output not yet on
Green Working normally

[2] kaigan saw a supply stay blinking green on an Onyx 350 whose interface board L1 had died, and took it to mean the supply was not being told to start.[6]

L1 messages

[edit | edit source]

The L1 message table in the Onyx 350 guide is the same as SGI's Origin 350 and Onyx 300 tables: a voltage more than 10% from nominal is a warning and more than 20% a fault, which starts a 30-second power-off; fan and inlet temperature messages follow, with advisory, critical and fault levels at 30, 35 and 40 °C.[5] The full table is on SGI Onyx 300 Troubleshooting Guide.

The L1 does not run

[edit | edit source]

If the L1 display stays blank and the module does not answer on its console port, suspect the L1's battery-backed Dallas chip first. hamei reports that interface boards in this family have stopped working after their Dallas battery went flat, and that a new chip did not bring them back for him or for others who reported it.[6] jwhat found that the Onyx 350's L1 initialises a cleared replacement chip itself.[7] See SGI Onyx 350 General Maintenance.

Rail faults

[edit | edit source]

A rail warning or fault on a compute module is investigated as on the Origin 350: read env, measure the rail with a meter to rule out a failed monitoring chip, and remove recently added cards. jwhat's Onyx 350 shut down with 1.8V low fault limit reached @ 1.396V after he moved an Adaptec FireWire card into it; with the card removed the rail read 1.777 V.[8]

InfiniteReality graphics

[edit | edit source]
  • Each pipe needs its own compute module, cabled from the module's XIO port to the Ktown2 board in the graphics module.[9]
  • The graphics module does not plug into the PDU. Its L1 starts only when its rear switch is on, and it must be powered from the same breaker box as the PDU.[10]
  • If the L2 cannot see the graphics module, check its USB link: directly to the L2 in the same rack, or through a USB hub when the module is in another rack.[10] jwhat notes that the module's USB port takes a cable with a DB-9 connector at the module end.[11]
  • Wrong colours on a monitor from another system: set its RGB inputs to 75 ohms.[10]

InfinitePerformance graphics

[edit | edit source]
  • The InfinitePerformance board appears in env as ODYSSEY, with its own temperature sensor; on a module without the board the entry reads not present.[8][12]
  • On a rackmount SGI Tezro, which sgistuff.net places in the same hardware family, an owner's board disappeared from the system after the L1 reported FAN 1 warning limit reached @ 0 RPM. The dead fan was the blower on the underside of the graphics board; indy99 explains that it plugs into a three-pin header on the system board with SGI's own pin order, so a replacement fan's wiring has to be matched to SGI's. jan-jaap adds that such a board can also vanish when its Dallas I2C chip fails, and that the same L1-controlled I2C bus monitors the fans.[12][13]
  • Because of where the graphics board sits, SGI does not offer the redundant cooling fan option for the InfinitePerformance module.[14]

Before calling it a hardware fault

[edit | edit source]

Racks joined by NUMAlink or Xtown2 cables must not differ in ground potential by more than 0.25 V.[10]

[edit | edit source]

References

[edit | edit source]
  1. ↑ SGI, 007-4632-001, chapter 3, pp. 60–61. Front panel LEDs.
  2. ↑ 2.0 2.1 SGI, 007-4632-001, chapter 7, p. 137, Table 7-2. Power supply LED.
  3. ↑ Silicon Graphics, SGI L1 and L2 Controller Software User's Guide, 007-3938-006, chapter 3, pp. 66–78 (File:SGI L1 and L2 Controller Software User's Guide (007-3938-006).pdf).
  4. ↑ Silicon Graphics, SGI Onyx 300 Graphics System User's Guide, 007-4509-002, May 2002, chapter 2, pp. 90–91 (File:SGI Onyx 300 Graphics System User's Guide (007-4509-002).pdf). L2 cfg command.
  5. ↑ 5.0 5.1 5.2 SGI, 007-4632-001, chapter 8, pp. 141–145, Tables 8-1 and 8-2. Troubleshooting chart and L1 messages.
  6. ↑ 6.0 6.1 kaigan, hamei and weblacky, "O350 Interface Board L1 Issues", IRIX Network forum, 5–20 June 2020, forums.irixnet.org.
  7. ↑ John Hartley (jwhat), "SGI NVRAM & Time Keeper Chips", Just Graphica, October 2020, updated March 2022, just.graphica.com.au (accessed 22 October 2026).
  8. ↑ 8.0 8.1 jwhat and weblacky, "O350 Voltage Fault:ATTN: 1.8V low fault limit reached @ 1.396V.", IRIX Network forum, 16–21 November 2020, forums.irixnet.org.
  9. ↑ Silicon Graphics, SGI Onyx 350 Visualization System User's Guide, 007-4632-001, June 2003, chapter 1, pp. 3–5 (File:SGI Onyx 350 Visualization System User's Guide (007-4632-001).pdf). One compute module per InfiniteReality pipe.
  10. ↑ 10.0 10.1 10.2 10.3 SGI, 007-4632-001, chapter 2, pp. 23–45. Monitor, power connections, L2 connections, ground potential caution.
  11. ↑ John Hartley (jwhat), "SGI Onyx 350 - Racking and Stacking", Just Graphica, 12 July 2021, just.graphica.com.au (accessed 22 October 2026).
  12. ↑ 12.0 12.1 "Tezro Rack - FAN 1 failure", IRIX Network forum, posts by the original poster, HAL, jan-jaap and indy99, 5–10 April 2019, forums.irixnet.org.
  13. ↑ Gerhard Lenerz, "Origin 350 / Onyx 350 / Onyx 4 (Chimera)", sgistuff.net, sgistuff.net (accessed 22 October 2026).
  14. ↑ SGI, 007-4632-001, chapter 4, p. 73. No redundant fan in the InfinitePerformance module.


Retrieved from "https://retrotechwiki.com/index.php?title=SGI_Onyx_350_Troubleshooting_Guide&oldid=29645"