False CPU Thermal Trip errors are being seen on Sun Blade X6250.

Asset ID:	1-73-1019279.1
Update Date:	2010-08-17
Keywords:

Solution Type FAB (standard) Sure

Solution 1019279.1 : False CPU Thermal Trip errors are being seen on Sun Blade X6250.

Impact

Systems are exhibiting false Thermal Trip errors on CPU modules. On some systems there will just be false errors within the ELOM SEL log, with others the CPU will also falsely show disabled within the ELOM cli/gui. Please make note that if experiencing this issue all of these signs noted above are false, and the host OS will still show all CPUs and cores online and functioning normally.

To help determine a real event from a false event the below paragraph is a description of how a system will behave if a real Thermal Event took place.

As far as the ELOM is concerned the error messages will be the same as a false Thermal Trip except you should notice that the host system is rebooting right after the time of the Thermal Trip error, and not before. Where in these false events the Thermal Trip errors take place just after a normal reboot of the host, or randomly while the host system is up and running with no reboot associated. The host in the presence of a real event will reboot unexpectedly, and the OS will not show the CPUs as available if the CPU was truly disabled by the BIOS.

This has become a service issue because engineers have been replacing Blades incorrectly due to these false failure signs. This has in turn caused a shortage of X6250 Blades in some Regions.

Contributing Factors

All Sun Blade X6250 Server Modules running with earlier than firmware/BIOS SW1.3 are impacted by this issue.

Symptoms

Example SEL log error listed below;

   Nonrecoverable ,2008/03/06 15:04:01 ,Processor 0 thermal trip detected

Example ipmitool sel elist output below;

   # ipmitool -H x.x.x.x -U root sel elist
     53 | 12/05/2007 | 14:15:43 | Processor Processor 0 | Thermal Trip | Asserted
     8b | 01/28/2008 | 08:46:14 | Processor Processor 0 | Thermal Trip | Asserted
     c6 | 02/26/2008 | 11:47:15 | Processor Processor 0 | Thermal Trip | Asserted

On systems that show CPUs disabled within the ELOM you will see;

   -> show CPU0

   /SYS/CPU/CPU0
       Targets:

       Properties:
           Designation = CPU 0
           Manufacturer = Intel
           Name = Clovertown
           Speed = 2333MHz
           Status = disabled

Root Cause

There are two root causes to this issue. The first which includes Thermal Trips on Clovertown CPUs is due to an ELOM firmware issue. During Power ON/OFF of the system there is some signal noise on LN93 which ELOM is reading as a false thermal trip. To resolve ELOM has added temperature judgement to fix the false thermal trip readings.

The second which includes the Thermal Trips on Harpertown CPUs is due to a BIOS issue. Recent changes within the ELOM command structure have caused BIOS to be out of sync with the structure changes which causes the CPUs to show disabled. This will be resolved by syncing up BIOS and ELOM command structure.

Corrective Action

Workaround:

If it is determined the Thermal Trip errors being experienced are false then the errors can be ignored and nothing should be done until the SW release resolving this issue is available.

Resolution:

This issue will be resolved with the next release of firmware/BIOS SW1.3.

References:

Escalation ID: 65875681, 1-23543493

For information about FAB documents, its release processes, implementation strategies and billing information, go to the following URL:

http://tns.central/fab

For Sun Authorized Service Providers go to:

http://tns.central/SPE/fab/

In addition to the above you may email:

[email protected]

Internal Contributor/submitter
[email protected]

Internal Eng Responsible Engineer
[email protected] Responsible Manager: [email protected]

Internal Services Knowledge Engineer
[email protected]

Internal Eng Business Unit Group
NSG (Network Systems Group)

Internal Sun Alert & FAB Admin Info
21-May-2008: Completed draft and sent to Extended Review.
23-May-2008: Incorporated feedback from Ext Rvw and sending to Publish.
17-Dec-2009: Replaced Product with Swordfish Nomenclature

Attachments

This solution has no attachment