X4100 and X4200 May Encounter Unscheduled System Reboots Due to Double-Bit Uncorrectable Memory Errors

Asset ID:	1-73-1000095.1
Update Date:	2010-09-01
Keywords:

Solution Type FAB (standard) Sure

Solution 1000095.1 : X4100 and X4200 May Encounter Unscheduled System Reboots Due to Double-Bit Uncorrectable Memory Errors

Related Items


Sun Fire X4100 Server
 Sun Fire X4200 Server

Related Categories


GCS>Sun Microsystems>Sun FAB>Standard>Reactive

PreviouslyPublishedAs
200113

Product
Sun Fire X4100 Server
Sun Fire X4200 Server

Bug Id
<SUNBUG: 6364001>

Part

Part No: 540-6497
Part Description: 2GB ECC Registered DIMM Module

Impact

A small proportion of X4100 and X4200 systems have been experiencing unscheduled reboots.

Contributing Factors

The reboot could happen anytime there is heavy traffic between the CPU and DIMMs.

Symptoms

The BIOS Event Log (DMI) will show "Sync flood error" just prior to the reboot. The System event log (SEL) of the ilom if interrogated with ipmitool (available on Resource cd) will show messages similar to these:

e00 | 03/21/2006 | 04:58:39 | OEM #0xfb | f00 | 03/21/2006 | 04:58:49 | Memory | Memory Device Disabled | CPU 0 DIMM 0
1000 | 03/21/2006 | 04:58:55 | System Firmware Progress | Motherboard initialization

Root Cause

DDR1 memory on these platforms may have an issue dealing with going in or out of the PowerDown mode and trigger uncorrectable ECC errors that cause system reboots. BIOS 034 and earlier enables the PowerDown mode (self-refresh/low-power mode) on the DIMMs with the wrong topology setting for these systems.

Workaround

Resolution

Upgrade to BIOS 036 or later. Statistically, BIOS 036 reduces the probability of an unscheduled reboot with certain registered DIMMs and increases stability. BIOS 036 will disable the PowerDown mode per AMD's recommendation. BIOS 036 can be obtained via the following website:

http://www.sun.com/servers/entry/x4100/downloads.jsp

Note: Some corner-case registered DIMMs with poor noise immunity coupled with corner-case noisy motherboards may not be fixed with BIOS 036. If the issue continues, the CFE or field representative should have their case escalated and the engineer assigned should refer to the TSC VSP - X4100/X4200 website (listed below) for further remediation actions.

http://systems-tsc.uk/twiki/bin/view/Products/ProdIssuesSunFireX4100

Previously Published As
102619
Internal Comments

It is recommended that if a field engineer is doing a motherboard replacement or other FRU replacement, BIOS 036 or later should be loaded.

Upgrading to BIOS 036 or later should be the first step in resolving memory related issues.

Customers should be advised to upgrade their LSI firmware/MPT BIOS firmware if moving to BIOS 036.

Sun supplied vendor DIMMs meet all of the JEDEC specs and are not faulty in their own right.

Refer to Product Notes 1.2.1 (819-1162-21) and Release Note Supplement 1.2.1 (819-4344-10) for further information.

Hardware Remediation Details

Related Information

Manual: Product Notes 1.2.1 - PN 819-1162-21
URL: http://systems-tsc.uk/twiki/bin/view/Products/ProdIssuesSunFireX4100
http://www.sun.com/servers/entry/x4100/downloads.jsp

Internal Contributor/submitter
brian.jones@sun.com

Internal Eng Business Unit Group
NSG (Network Systems Group)

Internal Eng Responsible Engineer
derek.tsai@sun.com

Internal Services Knowledge Engineer
sean.hassall@sun.com

Internal Escalation ID
1-14109922, 1-13950402, 1-15145059, 1-15344836, 1-15612351, 1-15844911, 1-16558422, 1-17641524

Internal Kasp FAB Legacy ID
102619

Internal Sun Alert & FAB Admin Info
Critical Category:
Significant Change Date:
Avoidance: Upgrade
Responsible Manager: beth.s.beasley@sun.com
Original Admin Info: null

Product_uuid
54e2ac49-df71-11d9-89e6-080020a9ed93|Sun Fire X4100 Server
c6e795ef-df6f-11d9-89e6-080020a9ed93|Sun Fire X4200 Server

Attachments

This solution has no attachment