This chapter provides information about installing or removing components from your SGI system, as follows:
This section describes how to open the system for maintenance and upgrade, protect the components from static damage, and return the system to operation. The following topics are covered:
To prepare the system for maintenance, follow these steps:
If you are logged on to the system, log out. Follow standard procedures for gracefully halting the operating system.
At your system console, switch from console mode to system control mode by entering the following command:
$> Ctrl+T The system will respond with the system control prompt: L2> |
From the system control prompt (L2>), power off the system with the following command:
L2> power down |
Locate the power distribution unit(s) (PDUs) in the front of the rack and turn off the circuit breaker switches on each PDU.
When you finish installing or removing components, return the system to operation as follows:
Turn each of the PDU circuit breaker switches to the “on” position.
At your system console, enter the following command:
L2> power up |
Verify that the LEDs on the L1/L2 controller(s) turn on and illuminate green and that your controllers display that the system is powered on for each segment of the procedure, which indicates that the power-on procedure is proceeding properly.
If your system does not boot correctly, see “Troubleshooting Chart” in Chapter 7, for troubleshooting procedures.
| Caution: The components inside the system are extremely sensitive to static electricity. Always wear a wrist strap when you work with parts inside your system. |
To use the wrist strap, follow these steps:
Unroll the first two folds of the band.
Wrap the exposed adhesive side firmly around your wrist, unroll the rest of the band, and then peel the liner from the copper foil at the opposite end.
Attach the copper foil to an exposed electrical ground, such as a metal part of the chassis.
| Caution: Do not attempt to install or remove components that are not listed in Table 6-1. Components not listed must be installed or removed by a qualified SGI field engineer. |
Table 6-1 lists the customer-replaceable components and the page on which you can find the instructions for installing or removing the component.
Table 6-1. Customer-replaceable Components and Maintenance Procedures
Component | Procedure |
|---|---|
L1 controller panel | “Remove and Replace the System Control Display Panel”
|
IRU power supplies | “Removing or Adding IRU Power Supplies”
|
IRU fans (blowers) | “Removing and Replacing IRU Fans (Blowers)”
|
Replacing IRU blades | “Removing and Replacing a Compute/Memory Blade”
|
Add or replace DIMMs | “Memory DIMM Group Placement and Guidelines (12-DIMM Boards)”
|
Replace the DVD drive | “Remove and Replace the DVD Drive”
|
PCI, PCI-X, PCIe cards | “Adding or Replacing PCI/PCI–X Cards”
|
2D Gfx PCI card in base I/O | “Add or Remove a PCI Card in Base IA Blades”
|
Base I/O blade disk drives | “Installing or Replacing a Disk Drive in the IA/IA2 Blade”
|
While the individual rack units (IRUs) are not considered end-user replaceable, a number of their internal components can be removed and replaced. These include:
L1 controller panel
IRU power supplies
IRU cooling fans (also called blowers)
IRU blades and their components
To remove and replace either of the top two power supplies in an IRU, you must first remove the system control display panel. Use the following steps to remove the system control display:
Using a T10 torx driver, loosen the lower screw that holds the display panel to the IRU chassis, see Figure 6-1.
Lift the display assembly up about 1/3-inch (84 mm) and pull it outward from the IRU chassis.
Don't unplug the ribbon cable unless you are replacing the system control panel.
Let the control panel gently dangle and go to the section “Removing or Adding IRU Power Supplies”.
Replace the applicable power supply and then use the following set of steps to replace the system control panel.
Replace the system control panel using the following steps:
Position the controller panel assembly slightly above the retention screw and carefully slid it back into place, see Figure 6-2. Be careful not to snag the ribbon cable while replacing the panel.
Use the T10 torx driver to tighten the retention screw that holds the panel to the IRU chassis.
Restart the IRU or system as applicable.
Under most circumstances a single power supply in an IRU can be replaced without shutting down the IRU or the complete system. In the case of a fully configured (loaded) IRU, this may not be possible. In a fully loaded system, power supply failure could result in shutdown of the IRU due to the power loss. Note that when replacing either of the top two supplies, it is necessary to remove the system controller panel (see “Remove and Replace the System Control Display Panel” before proceeding).
Use the following steps to replace a power supply in the IRU:
Disconnect the power cord from the IRU power supply.
Press down on the retention latch at the top of the power supply, see Figure 6-3.
Using the power supply handle, pull the power supply straight out of the IRU chassis.
Install the new power supply with the retention latch at the top of the supply.
Slide the power supply into the chassis until the retention latch engages.
Reconnect the power cord to the power supply.
The IRU fans (blowers) are located at the rear of the unit (see Figure 6-5). You will need to access the rack from the back to remove and replace a fan. The IRU's system controller issues a warning message when a fan is not running properly. This means the fan RPM level is not within tolerance. When an IRU fan fails, the following things happen:
The L1 display on the IRU with the failed fan shows a warning message
ATTN: Fan (number) warning |
The console will show the same warning indicating the rack and IRU position
001c01 L2> Fan (number) warning limit reached @ 0 RPM |
A line will be added to the L1 system controller's log file indicating the fan warning.
If ESP is used, a warning message will be sent to it also, (see “SGI Electronic Support ” in Chapter 7).
The system controller monitors the temperature within each IRU. If the temperature increases due to the failed fan, the remaining five fans will run at a higher RPM to compensate for the missing fan. The system will continue running until a scheduled maintenance occurs. The fan numbers for the IRU (as viewed from the rear) are shown in Figure 6-5.
Note that a fan can be replaced while the IRU is operating. You will need a number 1 phillips-head screw driver to complete the procedure.
Use the following steps and illustrations to replace an IRU fan:
Undo the (captive) screw (located in the middle of the blower assembly handle). The handle has a notch for the screw access, see Figure 6-6.
Grasp the blower assembly handle and pull the assembly straight out.
Slide a new blower assembly completely into the open slot, see Figure 6-7.
Tighten the blower assembly screw to secure the new fan.
The blades installed in your SGI Altix 4700 can generally be removed from or installed in an IRU without powering down the entire system. However, the operating system should be halted and the power to the IRU should be shut down in most cases. Note that the IA blade holding your system disk(s) should never be removed without backing up and gracefully powering down the unit.
Use the following information and illustrations to remove and replace a compute/memory blade from the host IRU:
Open the front door of the rack and access the IRU blade slated for replacement.
Move the blade's retention tab gently to the left to disengage the locking tab. Be careful not to use too much force as the tab is breakable.
Pull the blade's retention tab outward until it is fully extended, see Figure 6-8.
Slide the blade outward by pulling the bottom ring gently out.
Remove the blade completely from the IRU and place it on an anti-static work surface or in anti-static packaging.
Slide the new blade carefully into the IRU slot until fully seated.
Carefully push the retention tab back into place so that the blade is locked down, see Figure 6-9 for an example.
Restart the IRU or system.
The compute/memory boards installed in your system IRU(s) have space for a total of either eight or twelve dual-inline memory modules (DIMMs). The guidelines for use of these DIMMs are similar for the two styles of board.
Verify the number of DIMM slots on your board and review the guidelines and descriptions in the applicable section before upgrading or replacing memory on a blade.
Memory is contained on cards that are referred to as DIMMs (dual in-line memory modules). Each individual memory or compute/memory blade can contain four, eight, or twelve DIMMs installed in DIMM slots located on the blade's node board. These twelve DIMM slots are organized into three groups of four DIMMs each, see Figure 6-10.
| Important: DIMMs are installed one per DIMM slot, and must be installed in groups of four in the 12-DIMM boards. |
Figure 6-11 shows an example of the physical DIMM locations within the processor blade assembly. Follow these guidelines when installing DIMM groups in the 12-DIMM boards:
Memory is increased or decreased in four-DIMM group increments only. Each of the four DIMMs that make up a DIMM group must be the same memory size and speed; however, different blades can differ in memory size.
The first four DIMMs (group 0) must be in place for the blade's memory to operate properly.
Mixing blades with different capacity DIMM groups within an IRU is permitted.
When installing 4 GB DIMMs, check with your SGI service representative. Space limitations may require that a maximum of eight 4 GB DIMMs be installed on each blade.
Memory is contained on cards that are referred to as DIMMs (dual in-line memory modules). Each individual memory or compute/memory blade can contain four or eight DIMMs installed in DIMM slots located on the blade's node board. These eight DIMM slots are organized into two groups of four DIMMs each, see Figure 6-12.
| Important: DIMMs are installed one per DIMM slot, and must be installed in groups of four in the 8-DIMM boards. |
Figure 6-13 shows an example of the physical DIMM locations within the processor blade assembly.
Follow these guidelines when removing/installing DIMMs in the 8-DIMM boards:
Memory is increased or decreased in four-DIMM group increments only. Each set of four DIMMs that make up a DIMM group must be the same memory size and speed.
Group 0 can be a different memory capacity than group 1.
The first four DIMMs (group 0) must be in place for the blade's memory to operate properly.
Mixing blades with different capacity DIMM groups within an IRU is permitted.
| Note: Use of 4GB “stacked” DIMMs will necessitate using only group 0 slots on the blade. A stacked DIMM uses two memory chips “piggybacked” on the same memory part location. |
Use the following steps to remove the DVD drive from an IA blade:
Power off the IRU, use instructions in “Powering the System On and Off” in Chapter 1.
Extract the IA blade you are going to install the new DVD drive in. See the information in “Replacing IRU Components”.
After removing the IA blade from the IRU chassis, place it on a stable flat surface with the left side facing upward (see Figure 6-14).
Use a Phillips type screwdriver to remove the DVD access panel screw, (see Figure 6-14).
Lift the DVD access panel clear of the IA blade and set it aside.
Unscrew and remove the DVD drive assembly retention screw.
Gently but firmly disengage the DVD power and data cable connectors from the old DVD drive and slide the DVD unit out of the IA blade housing.
Slide the new DVD drive assembly into the IA blade housing.
Connect the DVD power and data cable connectors to the replacement drive.
Screw in the DVD drive assembly retention screw.
Replace the DVD access panel and tighten the retention screw.
Return the IA blade to its IRU slot.
Power-up the IRU and/or system.
This section provides a brief overview of the PCI Express (PCIe) technology that will be available as an option with your system. PCI Express has both compatibility and differences with older PCI/PCI-X technology. Check with your SGI sales or service representative for more detail on PCI Express board options available with the SGI Altix 4700.
PCI Express is compatible with PCI/PCI-X in the following ways:
Compatible software layers
Compatible device driver models
Same basic board form factors
PCIe controlled devices appear the same as PCI/PCI-X devices to most software
PCI Express technology is different from PCI/PCI-X in the following ways:
PCI Express uses a point-to-point serial interface vs. a shared parallel bus interface used in older PCI/PCI-X technology
PCIe hardware connectors are not compatible with PCI/PCI-X (see Figure 6-16)
Potential sustained throughput of x16 PCI Express is approximately four times that of the fastest PCI-X throughputs
PCI Express technology uses two pairs of wires for each transmit and receive connection (4 wires total). These four wires are generally referred to as a lane or x1 connection. This is also called “by 1”. SGI Altix 4700 PCIe technology uses a x16 connector (64 wires) or “by 16” in PCI Express card slots. This technology will support PCIe boards that use connectors up to x16 in size. Table 6-2 shows this concept.
Table 6-2. SGI PCIe Support Levels
SGI x16 PCIe Connector |
|
|---|---|
x1 PCIe cards | Supported |
x2 PCIe cards | Supported |
x4 PCIe cards | Supported |
x8 PCIe cards | Supported |
x16 PCIe cards | Supported |
x32 PCIe cards | Not supported |
| Warning: Before installing, operating, or servicing any part of this product, read the “Safety Information” in Appendix B. |
This section provides instructions for adding or replacing a PCI/PCI–X card in the I/O blade or PCI expansion blades installed in your system's Individual Rack Units (IRUs). To maximize the operating efficiency of your cards, be sure to read all the introductory matter before beginning the installation.
There are different types of blades used in the IRU that hold PCI/PCI-X and PCIe boards. Note the following guidelines:
The IA/IA2 blade (base I/O blade) containing the system disk(s) and DVD will only accommodate half-height PCI/PCI-X cards.
The three-slot double-wide PCI blade supports only PCI/PCI-X cards that mount and install with an SGI sliding mounting bracket.
| Caution: To protect the PCI cards from ESD damage, SGI recommends that you use a grounding wrist strap while installing a PCI card. |
Use the following instructions and illustrations to install or replace a PCI/PCI-X card in the three-slot blade:
Follow the instructions in “Replacing IRU Components” and extract the three-slot blade from the IRU.
To extract the card carrier from a slot in the blade, pull the handle on the selected card carrier straight out, as shown in Figure 6-17.
If you are replacing a card, gently remove the existing card from the carrier.
If you are adding a card, extract the carrier metal filler plate by pushing down on it, as shown in Figure 6-18. This filler plate covers the area where your card connectors will protrude out to the rear panel of the blade.
| Note: If you are adding or replacing a half-height PCI card, skip ahead to step 6. |
If the full-height card that you want to install does not fit in the carrier, follow these steps:
Loosen the three screws on the guide bar with a Phillips screwdriver (if your carrier uses T8 Torx screws, you will need a Torx screw driver).
Adjust the carrier guide bar (also known as the alignment rail) up or down as necessary (see Figure 6-19).
Mount the card in the carrier, (see Figure 6-20) and ensure the following occurs:
– The card connectors protrude through the bulkhead where the metal filler plate was located.
– The edge connectors fit between the bottom guides of the carrier.
– The top of the card fits under the clips on the guide bar.
Tighten the three screws on the guide bar. Skip step 6 and proceed to step 7.
To install a half-height PCI card in a carrier, follow these steps:
Remove the two screws from the carrier guide bar, as shown in Figure 6-21.
| Note: Take care not to lose the threaded inserts in the carrier guide bar. |
Loosen the third screw of the carrier guide bar.
Place the half-height PCI card in the carrier so that the connector edge of the card rests against the lower guide tab of the carrier (see Figure 6-21).
Move the carrier guide bar (adjustment rail) so that it holds the card firmly in place (see Figure 6-22).
Place the PCI card bracket so that the screw holes of the bracket align with the empty screw holes of the carrier guide bar, as shown in Figure 6-23.
Secure the bracket to the carrier guide bar with two T8 Torx screws.
Tighten the screw that you loosened in step 6b.
Insert the carrier-mounted PCI card into the vacant slot, using the slot guide, as shown in Figure 6-24.
Push in horizontally on the carrier seating bar, as shown in Figure 6-25, to seat the card securely in the slot.
Install the blade into the IRU.
Power on the IRU or restart the system.
Run the ls -pci PCI hardware inventory command to verify the installation. This command lists PCI hardware that the operating system discovered during the boot operation.
To add or replace a PCI card in the IA/IA2 blade (base I/O), follow these steps:
Power off the IRU by following the power-off instructions in the “Powering the System On and Off” in Chapter 1 if you are removing the IA blade.
Extract the blade you are going to install the new PCI card in. See the information in “Replacing IRU Components” if you have not already removed the blade. Note that in the case of extracting an IA blade that holds the system disk, you must shut down the operating system.
After removing the IA blade from the IRU chassis, place it on a stable flat surface with left side facing upward (see Figure 6-26).
Using a Tx20 Torx driver, remove the front screw securing the PCI card access door.
Using a Tx20 Torx driver, remove the top PCI card access door screw.
Remove the PCI card access door.
Add/replace/remove a PCI card(s) following standard ESD rules. Make sure any newly installed PCI card(s) are firmly seated in the slot and secure.
Replace the PCI card access door by installing it into the slotted opening and sliding it forward, as shown in Figure 6-27.
Reinstall the top PCI card access door screw.
Reinstall the front PCI card access door screw.
Reinstall the IA blade in the IRU chassis.
Return the system to service.
Boot your operating system software. (See your software operation guide if you need instructions to boot your operating system.)
Run the ls -pci PCI hardware inventory command to verify the installation. This command lists PCI hardware that the operating system discovered during the boot operation.
The IA/IA2 blade (base I/O) in an IRU has two system disk drive bays as shown in Figure 6-28. This section describes how to install or remove the drives. The IA2 base I/O blade is an enhanced version of the IA that supports RAID 1, DVD-R/W, and DVI/VGA graphics. IA2 blades are available on most systems ordered in January 2007 or later.
Before you remove a disk drive, you must power off the IRU or system according to the instructions in the “Powering the System On and Off” in Chapter 1.
To remove a disk drive, follow these steps:
Open the front of the rack and access the two disk drive bays, as shown in panel 1 of Figure 6-29.
Remove the disk drive by depressing its handle lock with your thumb and pulling the handle away from the chassis until the handle disengages the disk drive connector from the blade's backplane connector.
Carefully slide the disk drive out of the blade and place it on an ESD-safe surface. Do not use the handle to pull the disk drive out of the bay.
After you remove the disk drive, go to the next section to install a new one, or, power on the IRU according to the instructions in the “Powering the System On and Off” in Chapter 1.
Before you install a disk drive, you must power off the IRU that houses the IA blade (base I/O) where the disk(s) are installed, see the instructions in the “Powering the System On and Off” in Chapter 1.
Note that you should always back up all files before replacement of the disk. Be sure to replace any Altix 4700 hard disk with an SGI approved drive.
| Caution: To protect the system from ESD damage, SGI recommends that you use a grounding wrist strap while installing the disk drive. |
To install a disk drive, see Figure 6-30 and follow these steps:
Access the two disk drive bays (see panel 1 in Figure 6-30).
| Note: If you are installing only one disk drive, install it in the left-side housing nearest the DVD drive. |
Position the drive assembly so that it engages the bay guide rails, and then gently push the drive into the blade. Do not use the locking handle to push the drive into the bay.
Swing the locking handle towards the chassis until the locking handle engages the latch.
After you finish installing the disk drive, power on the IRU or system according to the instructions in the “Powering the System On and Off” in Chapter 1.