Sun SPARC Enterprise M8000, M9000 - Solaris 11 Dynamic Reconfiguration (DR) failes due to "Unable To Probe PCI Leaf"

Asset ID:	1-72-1988103.1
Update Date:	2018-03-05
Keywords:

Solution Type Problem Resolution Sure

Solution 1988103.1 : Sun SPARC Enterprise M8000, M9000 - Solaris 11 Dynamic Reconfiguration (DR) failes due to "Unable To Probe PCI Leaf"

Applies to:

Sun SPARC Enterprise M9000-64 Server - Version All Versions and later
Sun SPARC Enterprise M8000 Server - Version All Versions and later
Oracle Solaris on SPARC (64-bit)

Symptoms

Solaris 11 domains with quad board XSB configuration in M8000 or M9000 are suffered by missing IO base card after Dynamic Reconfiguration operation ( addboard -c configure <XSB> ), but DR will finish and won't abort.

In /var/adm/messaages the following error can be observed:

Nov 24 10:42:43 CET 2014 Nov 24 10:42:43 mxxxx interpreter[25511]: ABORT: Segmentation Fault: Fault Addr: 0x00000020
Nov 24 10:42:43 CET 2014 Nov 24 10:42:43 mxxxx efdaemon[879]: Wait pid: 25511 status: 100
Nov 24 10:42:43 CET 2014 Nov 24 10:42:43 mxxxx opl_cfg: WARNING: IKP: Unable to probe PCI leaf (0-0-0)
Nov 24 10:42:45 CET 2014 Nov 24 10:42:45 mxxxx dr: OS configure dr@0:SB0::cpu0

>>> efdaemon with PID 879 waiting for PID 25511
>>> efdaemon - embedded FCode interpreter daemon

The FCode interpreter is aborting with a segmentation fault. Even the interpreter is failing with SIGSEGV, there is no application core written. The "efdaemon" is waiting for the that process, which aborted already, hence the operation is not successful in terms of the PCI probe.

As a consequence of the above behavior, internal disks, network devices ( bge ) supported by the missing IO base card ( IOUA) are missing, e.g. in cfgadm -alv, format, ifconfig.

It has been observed so far only with quad mode XSB configuration, but it seems reasonable it applies also to UNI mode configurations, because the probe mechanism is the same in Solaris and the difference of the XSB is just the amount of resources on the XSB.

Changes

Before this issue occurred, Dynamic reconfiguration was used with Solaris 10 without issues.

After upgrade to Solaris 11 the mentioned symptoms are observed, independently which Solaris 11 version was applied.

Cause

Bug:15629477 - SUNBT6934717 DISK IS MISSING AFTER DR OPERATION IN NEVADA 13

The cause is software only, therefor hardware replacements of IO base cards are not appropriate actions.

Solution

A solution was implemented via Bug 21093175 for Solaris 11.3 and Solaris 11.2 Support Repository Update.

Solaris 11.3 : FCS
Solaris 11.2 : SRU 11.3

Workaround to get the missing base IO card back into the configuration is to reboot the affected Solaris instance.

No hardware is suspect for the described issue. In fact the IO base card is a victim, and only not visible rather than
defective, after DR operation finished.

To discuss this information further with Oracle experts and industry peers, we encourage you to review, join or start a discussion in an appropriate
My Oracle Support Community - Oracle Sun Technologies Community.

Link to the Bug: 15629477 - SUNBT6934717 DISK IS MISSING AFTER DR OPERATION IN NEVADA 13

References

<NOTE:1452392.1> - How to Install & Remove IDRs on Solaris 11 and Later Releases
<NOTE:1537514.1> - Managing Interim Diagnostic Relief (IDR) Packages with Solaris11 IPS
<BUG:15629477> - SUNBT6934717 DISK IS MISSING AFTER DR OPERATION IN NEVADA 130
<BUG:21093175> - IMPLEMENT WORKAROUND FOR 15629477

Attachments

This solution has no attachment