Jump to content

SGI Origin 350 General Maintenance

From RetroTechCollection
SGI Origin 350 rear panel
Origin 350 compute module rear panel: power connector, console port, serial ports, L1 USB port, XIO and NUMAlink connectors, and the four PCI/PCI-X slots with the IO9 card in the lowest (SGI User's Guide, Figure 3-6)

Routine care for the SGI Origin 350, from SGI's SGI Origin 350 Server System User's Guide (007-4566-001) and owners' reports on the IRIX Network forum.[1] SGI lets owners change PCI and PCI-X cards, disk drives, power supplies, memory and the L1 controller display. Processors and the IO9 card are for SGI engineers, and only qualified service staff should replace the soldered lithium battery.[2][3]

Environment

[edit | edit source]
Origin 350 environmental specifications (SGI User's Guide, Tables A-1 and A-4)
Specification Value
Temperature, operating 5 to 35 °C up to 1,500 m; 5 to 30 °C from 1,500 to 3,000 m
Temperature, stored −40 to 60 °C
Humidity 10% to 95%, non-condensing
Altitude, operating Up to 3,000 m
Noise 6.0 bels sound power, up to 30 °C
Heat output 1315 Btu per hour maximum

[4]

In a rack, SGI asks for 20.96 cm of clear space in front of the module, 25.40 cm behind it and 15.24 cm on the right side; the left side needs none.[5] Keep the ventilation slots clear, and do not place the server on or near a radiator or in an enclosed space without ventilation.[3] The L1 controller raises an advisory when the intake air passes 30 °C, a critical warning at 35 °C and a fault at 40 °C; with the high-altitude setting the limits are 27, 31 and 35 °C.[6]

Cleaning

[edit | edit source]

SGI's instructions are to unplug the server before cleaning it and to use a damp cloth, not liquid or aerosol cleaners.[3]

Switching off

[edit | edit source]
  1. Shut down IRIX with init 0.[5]
  2. Press Ctrl+T to reach the L2 prompt, then power down one module with r 1 s 18 pwr d (rack 1, slot 18 in SGI's example) or every module with pwr d. The L1 display then shows Powered Down; the L1 controller itself stays on.[5]
  3. Without a console, press the power button on the front of each compute module.[5]
  4. The base compute module has no power switch. To turn off its L1 controller, unplug the module.[5]

Opening the module

[edit | edit source]
  1. Power the system off and disconnect all cables at the rear. SGI warns that components may be hot: wait about five minutes.[3]
  2. Remove the two screws holding the module to the front rack rails and pull it out until the safety latches stop it. For a card in either of the two lower PCI slots the module has to come right out of the rack.[3]
  3. Remove the ten Phillips screws and lift the hinged cover.[3]

Use a grounded wrist strap, and keep parts in their antistatic bags until you fit them.[3] To refit, close the cover with its ten screws, press the safety latches on both sides, slide the module in and refit the two rack screws.[3]

Power supplies

[edit | edit source]

The two supplies sit behind the bezel door on the right of the front panel, and either can be changed with the module running if the other is working. A module with one supply has to be powered off first.[3]

  1. Open the bezel door, remove the two Phillips screws on the screen cover and swing the cover open.[3]
  2. Push the interior release button to the right and pull up on the green handle to free the supply.[3]
  3. Slide the supply out on to an antistatic surface, keeping clear of the supply fan's ribbon cable.[3]
  4. Fit the new supply with its handle fully open, push it in, and press the handle down until it snaps home. Close the screen cover without pinching the ribbon cable, refit the two screws and close the bezel door.[3]

The LED at the top of the supply's front should then be steady green; the other states are on SGI Origin 350 Troubleshooting Guide.[3]

Disk drives

[edit | edit source]
  • The drives are behind the bezel door on the left of the front panel. With one drive, put it in the bottom bay.[3]
  • To fit a drive, open the bezel door fully, slide the sled into the bay guide rails with the locking handle swung open, push it in until the handle engages, then swing the handle closed.[3]
  • A data drive can be removed once it has spun down. Power the system off before removing the system drive. Do not pull the drive out by its handle.[3]

Memory

[edit | edit source]
  • Fit DIMMs two at a time, the first pair in slots 0 and 1, then slots 2 and 3, 4 and 5, or 6 and 7. Both DIMMs of a pair must be the same size; different pairs may differ.[3]
  • Open the module, take out the plastic air baffle over the DIMMs, open the ejector latches, align the three notches with the keyed socket and press down until the DIMM seats. SGI warns that too much pressure can damage the socket. Refit the baffle before closing the cover.[3]
  • Origin 350 DIMMs do not fit the Origin 200, the SGI 2000 series, the Onyx2 or the Octane.[3]

PCI and PCI-X cards

[edit | edit source]
  • A PCI card and a PCI-X card on the same bus both run in PCI mode, and cards of different speeds on one bus both run at the slower speed.[3]
  • The slot above the IO9 card shares its bus, so it runs only in PCI mode at up to 66 MHz.[3]
  • Cards in the two top slots can be changed with the module still in the rack. For the two lower slots, take the module out and remove the chassis rail (five Phillips screws). Leave the IO9 card to an SGI engineer.[3]

Clock battery

[edit | edit source]

The NVRAM and time-of-day clock are on the IO9 card.[2] One owner who pulled the IO9 cards from his Origin 350 modules to replace their SNAPHAT battery and crystal packs found three of the five dead and the other two low.[7] SGI's guide reserves the soldered lithium battery for qualified service staff.[3]

Monitoring

[edit | edit source]

At the L1 prompt, env lists every monitored rail, fan and temperature with its limits. On one owner's working module it reported rails including 1.8V, 12V, 3.3V, 12V IO, 5V AUX, PCI 5V, PCI 3.3V and the node board's processor core, fans for the two exhaust positions, the supply and the PCI area, and temperature sensors on the interface board, PCI riser, node and Bedrock ASIC.[8] Run it now and then to catch a drifting rail or a slowing fan early.

[edit | edit source]

References

[edit | edit source]
  1. ↑ Silicon Graphics, SGI Origin 350 Server System User's Guide, 007-4566-001, June 2003 (File:SGI Origin 350 Server System User's Guide (007-4566-001).pdf).
  2. ↑ 2.0 2.1 SGI, 007-4566-001, chapter 3, pp. 119-126. IO9 card functions, customer-configurable items.
  3. ↑ 3.00 3.01 3.02 3.03 3.04 3.05 3.06 3.07 3.08 3.09 3.10 3.11 3.12 3.13 3.14 3.15 3.16 3.17 3.18 3.19 3.20 3.21 3.22 3.23 SGI, 007-4566-001, chapter 6, pp. 155-201. Safety instructions, PCI and PCI-X cards, disk drives, power supplies (Table 6-1), memory, L1 controller display.
  4. ↑ SGI, 007-4566-001, appendix A, pp. 212-214, Tables A-1 and A-4. Environmental and compute module specifications.
  5. ↑ 5.0 5.1 5.2 5.3 5.4 SGI, 007-4566-001, chapter 1, p. 26, Table 1-2, and pp. 53-61. Space and clearance requirements, powering on and off, power switches.
  6. ↑ SGI, 007-4566-001, chapter 7, pp. 206-207, Table 7-2. L1 temperature messages.
  7. ↑ "O350 Voltage Fault:ATTN: 1.8V low fault limit reached @ 1.396V.", IRIX Network forum, post by jwhat, 16 November 2020, forums.irixnet.org/thread-2563.html. Replacing SNAPHAT battery modules on IO9 cards.
  8. ↑ "o350 env chip locations", IRIX Network forum, post by jan-jaap, 28 February 2022, forums.irixnet.org/thread-3424.html. env output from a working Origin 350.


Retrieved from "https://retrotechwiki.com/index.php?title=SGI_Origin_350_General_Maintenance&oldid=29466"