Sun SPARC Enterprise T5220 Server
 
Sun SPARC Enterprise T5120 Server
 

GCS>Sun Microsystems>Servers>CMT Servers
 
GCS>Support>KM>Content>Video
 

Applies to:

Goal

Solution

1) Procedure for determining an FBDIMM has faulted,  determining which FBDIMM has faulted, locating the faulted FBDIMM on the motherboard, replacing the faulted FBDIMM, if necessary clearing the FBDIMM's faulted state in the Service Processor fault log and Solaris FMA fault log, and confirming that the replacement FBDIMM installed in place of the faulted one is itself non-faulted.

To identify the FBDIMM faulted by POST or PSH (Preventive Self Healing) technology:

In Solaris run:

# fmadm faulty

# fmadump -V -u

Using the from the fma erreport  will give the full path (and location) of the Faulted DIMM.

Using the Service Processor's built-in (in Service Processor firmware) commands:

sc>showfaults -v

(ILOM CLI equivalent ->show /SP/faultmgmt )

Once you have powered of, removed the system's power cord(s) and top cover and have fitted yourself with an antistatic strap, you can confirm the location of the faulted FBDIMM by the following procedure:

Open (lift up) the air baffle to access the FBDIMM fault locator button (also known as the "reminder" button).

Press the FBDIMM fault locator button on the motherboard.

The button is located on the left edge of the motherboard, next to /SYS/MB/CMP0/BR3/CH1/D1 (J1701).

On the motherboard, next to each FBDIMM location, there is an amber fault LED. With the push of the FBDIMM fault locator button, the LED next to the faulted FBDIMM location will light.

Important to mention is that since still at this time, Dec 2007, each of the Sun FBDIMM sizes has one Sun Part FRU number. One number for the 1GB FBDIMM, another for the 2GB FBDIMM and another for the 4GB FBDIMM.

From the Sun System HandBook:

The FBDIMMs are CRU's and they are RoHS:Y

1GB FBDIMM  501-7952

2GB FBDIMM  501-7953

4GB FBDIMM  501-7954

Certainly in the T5120/5220 FBDIMMs work together in pairs. However, when you order an FBDIMM by FRU number you will reeive ONE FBDIMM. If an FBDIMM goes faulty in the system you will replace just that one FBDIMM, not it and its Pair partner.

When you remove the Faulted FBDIMM, you can determine its FRU # from the Sun Label on that FBDIMM. Even before you open up the system, knowing the location of the faulted FBDIMM, you get the FBDIMM's Sun FRU part number by logging into the Service Processor as  admin  (alom CLI) and run

sc> showfru

And furthermore, if you have a recent explorer in the Tx000 output directory you will find the "showfru" file which will give you the FBDIMM FRU number. And in that same directory the "showfault_-v" and "showcomponent" files will present a state of Faults captured by Service Processor and whether any Host components were disabled manually or as a result of Host init fw or POST having noted a fatal fault, respectively.

If you don't replace the FBDIMM or the Service Processor did not automatically detect an FBDIMM replacement and automatically clear the mention of the faulted FBDIMM from the log - as a result of an

sc>resetsc

or

->reset /SP

then manually clear the FBDIMM fault from the log by running:

sc>clearfault

ILOM CLI equivalent

->set /SYS/ clear_fault_action=true

If you do replace the faulted FBDIMM, and the Service Procssor and the Service Processor did not detect the replacement and automatically clear the log upon Service Processor reset another way to do so would be to enable the FBDIMM and that will clear the the mention of the prior FBDIMM fault from the log:

sc>enablecomponent

Example of an asr-key: /SYS/MB/CMP0/BR3/CH1/D1

ILOM CLI equivalent

->set /SYS/ component_state=ok

To operationally, on the HOST side, to confirm replacement of the faulted FBDIMM was effective:

Run MAX POST

If

# fmadm faulty 

still indicates the FBDIMM location as faulted, then to return that FBDIMM location to a non-faulted state in the FM log, run:

# fmadm repair

UUID used would be the same as used in the clearfault  command above and as would have appeared in the output of the # fmadm faulty command above.

FBDIMM Configuration Big Rules

See Section 5.8 of the Service Manual for diagrams of the 4, 8 & 16 FBDIMM Configs

Sun Supported Number of FBDIMMs in the T5120/5220 are:

4 FBDIMMs, 8 FBDIMMs and 16 FBDIMMs.

The Sun T5120/5220 Serice Manual, as well as a label on the inside of the T5120/5220 Top Cover show in which motherboard FBDIMM slots to install FBDIMMs for each memory configuration - 4, 8 & 16 FBDIMMs.

Also, taking advantage that the T5220's Air Baffle in the T5220 covers the FBDIMM area of the motherboard, Sun has labeled the FBDIMM locations on that Baffle.

(Note the order of FBDIMM install when going from the 4 to 8 to 16 FBDIMM configuration, that is as mentioned above, on the label on the inside of the top cover)

All FBDIMMs in the system must be of the same size.

At minimum, Channel 0, FB-DIMM Slot 0 in all branches

must be populated with FB-DIMMS of the same density (same type).

The vendor of the AMB (Asynchronous Memory Buffer) chip on both FBDIMMs on a Memory Controller channel must be the same.

The AMB vendor and the version of the AMB on both FBDIMMs of an FBDIMM pair must be the same (A pair is comprised of an FBDIMM on one Memory Controller Channel and another FBDIMM kn the same electrical position on the like Channel of the paired Memory Controller.

All FBDIMMs in a Branch must have the same Sun part number.

At this time, the manufacturers of the FBDIMMs supported by Sun in the T5120/5220 hardware platform use the same AMB vendor and revision of that AMB. This means that for the 1GB FBDIMM there is but ONE FRU number, for the 2GB FBDIMM ONE FRU number, for the 4GB FBDIMM ONE FRU number - regardless of which Sun Supported FBDIMM vendor manufactured the FBDIMM.

2) Procedure for Adding FBDIMM X-OPTIONS:

Supported Number of FBDIMMs for the T5120/5220 are..

  4   FBDIMMs (Group 1)

  8   FBDIMMs (Groups 1 and 2)

  16  FBDIMMs (Groups 1, 2, 3 & 4 - fully populated configuration)

At minimum, Channel 0, FBDIMM Slot 0 in all branches must be populated with FBDIMMS of the same density (same type).

In branches populated with more than one FB-DIMM  (i.e., in 8 and 16 FB-DIMM configurations), FBDIMMs are addressed in pairs. Each pair must be identical  (same Sun part number).

Upgrading from a 4 FBDIMM configuration requires  moving two of the FBDIMMs to ensure matching pairs in all occupied slots

At this time, Dec 2007, there is only 1 FRU Part  Number the 1GB FBDIMM, another for the 2GB FBDIMM and a third for the 4GB DIMM.

Since all FBDIMMs in the system must be the same size, when adding FBDIMM X-Option - going from 4 FBDIMMs to 8 FBDIMMs and to 16 FBDIMMs -  all  FBDIMMs in the system will have the SAME Sun FRU Part Number.

Power down the host system (using the front panel power button) or STOP or poweroff command from the SP's root/ILOM or SP's admin/ALOM command line interfaces, respectively.

Unplug the system's power cord(s)

Remove the system's top cover.

Implement all the usual anti-static-discharge methods...

Prepare an antistatic surface to set parts on during  the removal, installation, or replacement process.

Place ESD-sensitive components such as the printed  circuit boards on an antistatic mat. The following  items can be used as an antistatic mat:

          o Antistatic bag used to wrap a Sun replacement part

          o Sun ESD mat, part number 250-1088

          o A disposable ESD mat (shipped with some replacement parts or optional system components)

*  Attach an antistatic wrist strap.

When servicing or removing server components,  attach an antistatic strap to your wrist and then to a metal area on the chassis.

Specific to the T5120 --- Disconnect and stow the hard drive data cable.

a. Unplug the hard drive data cable from J6401 on the  motherboard.

b. Remove the hard drive data cable from its seat on the CPU air baffle.

c. Place the hard drive cable end out of the way of the air baffle.

* Rotate the air baffle up and toward the front of the  system.

* Push down on the ejector tabs on each side of the FB-DIMM to be removed, until the FB-DIMM is  released.

* Grasp the top corners of the faulty FB-DIMM and remove it from the server.

* Place the FB-DIMM on an antistatic mat.

Procedure for Installing an FBDIMM

* Unpack the replacement FB-DIMM and place it on an antistatic mat.

* Ensure that the ejector tabs are in the open position.

* Line up the replacement FB-DIMM with the connector.

* Align the FB-DIMM notch with the key in the connector.  This ensures that the FB-DIMM is oriented correctly.

* Push the FB-DIMM into the connector until the ejector tabs lock the FB-DIMM in place. If the FB-DIMM does not easily seat into the connector, verify that the orientation of the FB-DIMM is correct. If the orientation is reversed, damage to the FB-DIMM might occur.

* Rotate the air baffle back into its operating position. The air baffle snaps into position above the CPU and memory modules.

* Specific to the T5120 - Reconnect the hard drive data cable.

a. Route the hard drive data cable over the fan module and along the air baffle.

b. Plug the data cable into J6401 on the motherboard.

Close Up the System and Test

* Replace the system's top cover.

* Plug in the system's power cord(s).

* If the system administrator would like the Host powered-up, that can be done using the front panel power button or via a login to the SP's   root  or   admin  accounts  and the appropriate   START  or  poweron, respectively, may be used.

* Run MAX POST to test the replacement or added FBDIMMs.

This solution has no attachment