Big Data Appliance X3-2 Hardware
 
Big Data Appliance X4-2 Hardware
 
Big Data Appliance Hardware
 

PLA-Support>Eng Systems>BDA>Big Data Appliance>DB: BDA_EST
 

In this Document

Symptoms

Cause

Solution

References

Created from <SR 3-9966605076>

Applies to:

Symptoms

1. Running the BDA utility, bdacheckhw on a node, reports a memory failure:  fault.memory.intel.sb.dimm_ce.

The complete message looks like:

2. Running 'ipmitool sunoem cli "show faulty"' reports the same:

3. Rebooting the server clears the memory fault.

4. The ILOM snapshot after the reboot confirms the fault is cleared.  In other words, the ILOM snapshot confirms no DIMM faults are present after reboot.

a) From ./ilom/@usr@local@bin@spshexec_show_-script_@X@logs@event@list.out we see the fault and that it was cleared after reboot:

b) From -> show faulty after reboot, the fault is cleared:

c) Also from the ILOM snapshot the fault is "Repaired"/"Resolved" after reboot:

Cause

The bdacheckhw errors and ILOM snapshot prior to reboot confirm a memory fault i.e. fault.memory.intel.sb.dimm_ce"  due to excessive memory correctable errors.

Note: fault.memory.intel.sb.dimm_ce (from the ce) are memory correctable errors. 

There are a number of ways an excessive memory correctable / non correctable errors can occur:

•  Electrical or magnetic interference 
•  Dimm poorly seated in slot
•  High temperatures           
•  Data compare

Memory is tested with every post, so if a a server is rebooted and the error does not reoccur, memory is assumed to be ok.    MOS document: 1155200.1 - PSH Procedural Article for ILOM-Based Diagnosis, states that you should check FMA errors.

Solution

In the case of a memory fault due to excessive memory correctable errors the solution is to:

1. First reboot the server or clear the faults in the ILOM. See if these steps allow the Fault Manager to clear the memory faults.

You can enable x86 diagnostics (Enabled or Extended) to run at boot to further confirm that no memory faults remain.  To do so see: Running Enabled and Extended Diagnostics to Confirm Hardware Errors on the BDA (Doc ID 1956552.1)

2. The recommendation is to replace the faulted dimm only if there is re-occurrence of the fault on the same dimm.  See MOS document: 1155200.1 - PSH Procedural Article for ILOM-Based Diagnosis.

Also see the MOS document: 1438864.1 - SPX86-8004-CE - Fault due to excessive memory correctable errors (CE's)

References

This solution has no attachment