Description

Sun SPARC Enterprise T5440 systems running firmware prior to 7.4.4.f may erroneously report a 'MB_DC_POK' power fault, which prevents the system from powering down from either Integrated Lights Out Manager (ILOM) or Solaris.

Occurrence

This issue can occur on the following platform:

SPARC Platform:

• Sun SPARC Enterprise T5440 servers without firmware version 7.4.4.f (as delivered in patch 147311-05) or later

Note: This issue only occurs on T5440 systems running system firmware prior to 7.4.4.f. No other systems or platforms are affected by this issue.

To determine the firmware version installed on the system, use the following ILOM command:

    -> show /HOST sysfw_version

       /HOST
       Properties:
       sysfw_version = Sun System Firmware 7.4.4.f 2012/09/07 16:28

Symptoms

If the described issue occurs, the system will briefly enter a powered down state and then transition back to a powered on state when trying to power down the host. This may happen consistently or only sporadically.

It is likely that this issue has been experienced if, after running the following on the Service Processor, the system does not power down:

    stop /SYS

This can be determined by doing the following:

    show /SYS

with the system showing "power_state = On".

The cause of the failure can then be determined by doing the following on the Service Processor:

   show /SP/logs/event/list

If entries similar to:

    "/SYS/MB/CPU0/DVRM_15/OOV_+1V5: asserted failure"

and

    SP detected fault at time Wed May  9 20:15:18 2012. Host Power Failure:
    MB_DC_POK Fault

are present in the log, then the issue described has occurred.

Workaround

There is no workaround for this issue. Please see the Resolution to this issue below.

This issue is addressed on the following platform:

SPARC Platform:

• Sun SPARC Enterprise T5440 servers with firmware version 7.4.4.f (as delivered in patch 147311-05) or later

Notes: Since loading new system firmware requires the host to be powered down, updating the system firmware may require a modified process.

If a power fault problem occurs inconsistently, repeatedly trying stop "/SYS" may finally power down the host. However, if attempting to stop /SYS fails to power down the system, then perform the following using ILOM:

    -> show /SP/policy HOST_AUTO_POWER_ON

and make a note of the value. If the value is currently set to "enabled" then do the following:

    -> set /SP/policy HOST_AUTO_POWER_ON=disabled

then pull all the power cables to the system. When power is restored to the system and the Service Processor comes up, the host will remain powered down. At this point, follow the recommended procedures for updating the system firmware.

When the system firmware has been successfully upgraded, if the value of "HOST_AUTO_POWER_ON" was disabled (during previous steps), use the above 'set' command to change the value back to 'enabled'.

Patches

 <SUNPATCH:147311-05>

History

01-Nov-2012: Date of Resolved release
30-Jan-2013: Maintenance update to migrate BugID; no change in content

Internal Section: Comments:

This is a hardware issue that will not be fixed by hardware. The fact that system firmware would turn the power back on without warning could be considered a bug in system firmware; however this was considered correct behavior. With this fix, firmware will always power down the host if this error occurs. In addition, it was agreed that firmware would not report the existence of the error since this could be considered a call generator.  The fix in this Sun Alert is considered as the resolution to the problem.

Internal Contributor/Submitter: david.arneson@oracle.com
Internal Eng Responsible Engineer: david.arneson@oracle.com
Oracle Knowledge Analyst: Jeff.folla@oracle.com
Internal Eng Business Unit Group: Systems Group - SYS
Internal Escalation ID:3-2876176281, 3-3819889211, 3-4511644851, 3-5157762801, 3-5941978643
Internal Pending Patches: None
Internal Resolution Patches: 147311-05

References

This solution has no attachment