Description

When collecting a snapshot on SPARC M5-32 or M6-32 Servers running certain Sun system firmware versions (listed below) with 'dataset=full', or when collecting a snapshot via the BUI with the Diagnostic Data option selected (in Custom Data Set), all the HOSTs will be powered off if you answer "yes" to the question: "Collecting a 'full' dataset may reset the host. Are you sure (y/n)?"

Occurrence

This issue can occur on the following platforms:

• SPARC M5-32 and M6-32 Servers with Sun system firmware version 9.2.0.b or later

Note 1: In the past it had been communicated that this warning, that hosts may be reset, only applied to T-series systems and that the M5-32 and M6-32 Servers could ignore this warning. However, with the changes introduced into Sun System Firmware version  9.2.0.b, this warning now applies to all systems.

Note 2: Systems with Sun system firmware versions prior to 9.2.0.b are not affected by this issue.

To determine the current firmware version on one of these servers, the following command can be used:

    -> show /System system_fw_version 

    /System
    Properties:
    system_fw_version = Sun System Firmware 9.2.0.b 2014/04/21 07:09

Symptoms

If the described issue occurs, logs will show that all powered on HOSTs are powered off at the same time with the following keywords:

by: Internal test application. Reason: Test

    -> show /SP/logs/event/list/

    36343  Tue May 20 11:24:03 2014  Power     Off       major   
       Power to /HOST3 has been turned off by: Internal test application, Reason: Test
    36342  Tue May 20 11:24:01 2014  Power     Off       major   
       Power to /HOST2 has been turned off by: Internal test application, Reason: Test
    36341  Tue May 20 11:23:59 2014  Power     Off       major   
       Power to /HOST1 has been turned off by: Internal test application, Reason: Test

Workaround

To work around this issue, collect snapshot using the fruid dataset rather than the full one:

    -> set /SP/diag/snapshot/ dataset=fruid
    Set 'dataset' to 'fruid'

    -> set /SP/diag/snapshot/ dump_uri=ftp://user[:password]@host//absolute-directory-path/

This issue is addressed in the following release:

• SPARC M5-32 and M6-32 systems Firmware version 9.5.3 or later (as delivered in patch 22270916)

See also: "SPARC M5-32 and M6-32 Servers: Firmware Image Software Version Matrix Information" <Document:1540816.1>

History

22-May-2014: Document released; State: Workaround
27-May-2014: Updated Note 1 in the Occurrence section
18-Feb-2016: Update Workaround/Resolution for FW release patch, issue is Resolved

Internal Section: Comments:

This regression was introduced with the implementation of  CR1733612.

Note: M6-32 SuperCluster is an engineered system and as such, is only certified with a certain level of fw, patches, etc.
9.2.0.b is NOT a supported version for M6-32 SuperCluster

Questions regarding this document should be addressed to
sunalertpublication_us_grp@oracle.com and copy the
responsible engineer/submitter listed below.

Internal Contributor/Submitter: benoit.baguette@oracle.com
Internal Eng Responsible Engineer: josline.nyagahima@oracle.com
Internal Services Knowledge Engineer: jeff.folla@oracle.com
Internal Eng Business Unit Group: Sup-Systems Sparc
Internal Escalation ID:
Internal Resolution Patches: 22270916

References

This solution has no attachment