Jump to content

SGI Onyx 3000 Troubleshooting Guide

From RetroTechCollection
SGI Onyx 3000 rack with G-brick (Owner's Guide Figure 1-1)
Onyx 3000 rack: each brick has its own L1 display, and the L2 controller sits on the roof with its touch display on the door (SGI Owner's Guide, Figure 1-1)

Fault-finding on the SGI Onyx 3000, from SGI's SGI Onyx 3000 Series Graphics System Hardware Owner's Guide (007-4264-002), the SGI L1 and L2 Controller Software User's Guide (007-3938-006) and owners' reports. Most faults show up first on a brick's L1 display or in the L2 controller's warnings and logs, so start there.[1]

Order of work

[edit | edit source]
  1. Read the L1 display on every brick, and look at the LEDs on its front panel.[2]
  2. On the L2 touch display, open Warnings/Faults and L1/L2 Logs.[3]
  3. Connect a terminal to the L2 console port, or to a C-brick console port if there is no L2, and run env and log on the suspect brick.[1][4]
  4. Check the rear 12 VDC and 48 VDC LEDs and the cable LEDs on the brick and the bricks it connects to.[2]

Lights

[edit | edit source]
Brick indicators (SGI Owner's Guide, chapter 2)
Indicator Meaning
On/Off LED, green (front) The brick's internal components are powered
Service required LED, orange (front) Something has failed, for example a fan, but the brick is still running
Failure LED, red (front) A failure has occurred and the brick is down
12 VDC LED (rear) The L1 controller is powered and running; it comes on when the rear power switch is On
48 VDC LED (rear) The rest of the brick is powered
PWR LED on a G-brick board The board is seated and receiving power
Fault LED on a G-brick board, yellow A fault on that board
Ktown2 or XIO connector LED, yellow The cable is connected at both ends
Ktown2 or XIO connector LED, green The link has been negotiated with the brick at the other end

[2]

L1 messages

[edit | edit source]

SGI counts a voltage more than 10% from nominal as a warning and more than 20% as a fault.[1]

L1 controller messages (SGI Owner's Guide, Table 6-2)
Message Meaning
ATTN: x.xV high (or low) fault limit reached @ x.xxV A voltage is 20% out; the brick (or the system, if there is no backup) starts a 30-second power-off
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 cannot reach its minimum speed and the room is above 30 °C, or another fan is already at minimum; a 30-second shutdown starts
ATTN: FAN # x warning limit reached @ xx RPM A fan cannot reach its minimum speed in a room below 30 °C; the other fans speed up
ATTN: FAN # x stabilized @ xx RPM The fan is running at the right speed again, usually after replacement
ATTN: TEMP # advisory / critical / fault temperature reached Inlet 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 the brick down; repeated every five seconds
Brick appears to have been powered down The L1 has shut the brick down after a fault

[1]

Reading the environment

[edit | edit source]

At the L1 prompt of a brick, or with the brick selected at the L2:[4]

  • env lists every monitored voltage, fan and temperature with its warning and fault limits; the item that tripped is marked Fault.
  • env check gives a short status.
  • env reset clears the current warnings and faults.
  • log shows the L1 log, and log clear empties it.
  • env off and env temp off turn monitoring off. SGI does not recommend this in normal use.[4] weblacky, who repairs these systems, would only use it briefly, to see whether a machine reaches its PROM while a fault is being chased, and then turn monitoring back on.[5]

Will not power on

[edit | edit source]
  • Check that the PDU and PDS breakers are on and the cords seated. Each G-brick has its own cord to a wall outlet, separate from the PDU.[3]
  • Check the rear switch of every brick. SGI warns that the system can malfunction or fail to boot if one is left off.[3]
  • A brick whose rear 12 VDC LED stays dark after its switch is turned on has no power to its L1 controller.[2] SGI's chart for the related Onyx 300 says to check the rear switch, read the L1 display, and check the connection between the module and its power source.[6] On the Onyx 3000 that source is the power bay, or the wall outlet for a G-brick.[7]
  • If the 12 VDC LED is lit but not the 48 VDC LED, the L1 is running but has not powered the brick; read its display and log.[2][4]
  • Any D-brick must have its key switch at RUN.[3]

False readings and monitoring chips

[edit | edit source]

If env shows readings that cannot be right, or a brick shuts itself down while its rails measure correctly with a meter, suspect the Dallas DS1780 monitoring chips. jan-jaap, a long-time collector, says these chips were already a known problem in Origin 3000 bricks when the systems were new.[8] On the related Origin 350, owners have found failed DS1780s reporting impossible temperatures and voltages; the reports, and the chips' locations on that machine, are on SGI Origin 350 Troubleshooting Guide.[9] No owner has published the DS1780 locations for Onyx 3000 bricks.

No L1 comm

[edit | edit source]

A brick display showing No L1 comm has been reported on a C-brick. jan-jaap reads it as the brick's L1 unable to talk to the rest of the system, and names a dead NVRAM chip or battery as one possible cause; jpstewart has seen the same message on an Origin 300 whose Dallas battery had died.[10] Connect a terminal to the brick's console port with the system off and read the L1 status before replacing anything.[10]

Graphics faults

[edit | edit source]
  • No picture: check that the G-brick's 48 VDC LED is lit and that the Ktown2 connector LEDs for the pipe are green. Yellow means the cable is connected but the link has not come up.[2]
  • Each InfiniteReality pipe needs its own C-brick, through an I-, X- or N-brick.[2][11]
  • A board with its yellow fault LED lit, or its PWR LED dark, is the place to start inside the G-brick.[2]
  • Wrong colours on a monitor from another system: set its RGB inputs to 75 ohms.[3]
  • Blank screens when driving more than two monitors from a pipe: SGI gives about 300 Mpixel/s as the pipe's total output, so mix lower and higher resolutions to stay within it.[3]
  • A V-brick's V12 board needs a C-brick on one of its XIO ports, and the connector LEDs follow the same yellow and green scheme.[2]

Before calling it a hardware fault

[edit | edit source]
  • Chassis joined by NUMAlink or Xtown2 cables must not differ in ground potential by more than 0.5 V; SGI warns of severe damage.[3]
  • N-brick systems need L1 firmware 1.6.0 or later on every brick, and IRIX 6.5.13 or later.[11]
[edit | edit source]

References

[edit | edit source]
  1. ↑ 1.0 1.1 1.2 1.3 SGI, 007-4264-002, chapter 6, pp. 127–139. L1 and L2 controllers, L1 messages (Table 6-2).
  2. ↑ 2.0 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 Silicon Graphics, SGI Onyx 3000 Series Graphics System Hardware Owner's Guide, 007-4264-002, November 2001, chapter 2, pp. 35–60 (File:SGI Onyx 3000 Series Graphics System Hardware Owner's Guide (007-4264-002).pdf). G-brick, V-brick and N-brick front and rear panel indicators, board and connector LEDs.
  3. ↑ 3.0 3.1 3.2 3.3 3.4 3.5 3.6 SGI, 007-4264-002, chapter 4, pp. 92–119. Monitor, powering on and off, L2 touch display, ground potential caution.
  4. ↑ 4.0 4.1 4.2 4.3 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). env and log command sets.
  5. ↑ weblacky, replies of 6 May 2025 in "How to Incorporate an Origin 300 In Homelab?", IRIX Network forum, forums.irixnet.org.
  6. ↑ Silicon Graphics, SGI Onyx 300 Graphics System User's Guide, 007-4509-002, May 2002, chapter 4, pp. 118–119, Table 4-1 (File:SGI Onyx 300 Graphics System User's Guide (007-4509-002).pdf).
  7. ↑ SGI, 007-4264-002, chapter 1, pp. 4–5, Table 1-1. Power bay, PDU and G-brick power cables.
  8. ↑ jan-jaap, reply of 27 September 2023 in "Origin 350 - PCI 5V missing", IRIX Network forum, forums.irixnet.org.
  9. ↑ vvuk, jan-jaap, jwhat and fleedwood, "o350 env chip locations", IRIX Network forum, 28 February to 3 March 2022, forums.irixnet.org.
  10. ↑ 10.0 10.1 sgifan2010, Raion, jan-jaap and jpstewart, "CBRICK Error", IRIX Network forum, 11–13 November 2020, forums.irixnet.org.
  11. ↑ 11.0 11.1 Silicon Graphics, SGI Onyx 3000 N-Brick Addendum, 007-4424-001, p. 1 (File:SGI Onyx 3000 N-Brick Addendum (007-4424-001).pdf).


Retrieved from "https://retrotechwiki.com/index.php?title=SGI_Onyx_3000_Troubleshooting_Guide&oldid=29509"