Clear PSH-Detected Faults

When the Oracle Solaris PSH facility detects faults, the faults are logged and displayed on the console. In most cases, after the fault is repaired, the corrected state is detected by the system and the fault condition is repaired automatically. However, this repair should be verified. In cases where the fault condition is not automatically cleared, the fault must be cleared manually.

1. After replacing a faulty FRU, power on the server.
2. At the host prompt, use the fmadm faulty command to determine whether the replaced FRU still shows a faulty state.

   # fmadm faulty
   TIME            EVENT-ID                              MSG-ID         SEVERITY
   Aug 13 11:48:33 ********-****-****-****-************  SUN4V-8002-6E  Major 
    
   
   Platform    : sun4v     Chassis_id  :
   Product_sn  :
    
   Fault class : fault.cpu.generic-sparc.strand
   Affects     : cpu:///cpuid=**/serial=*********************
                     faulted and taken out of service
   FRU         : "/SYS/MB"
   (hc://:product-id=*****:product-sn=**********:server-id=***-******-*****:
   chassis-id=********:**************-**********:serial=******:revision=05/
   chassis=0/motherboard=0)
                     faulty
    
   Description : The number of correctable errors associated with this strand has
                 exceeded acceptable levels.
                 Refer to http://sun.com/msg/SUN4V-8002-6E for more information.
    
   Response    : The fault manager will attempt to remove the affected strand
                 from service.
    
   Impact      : System performance may be affected.
    
   Action      : Schedule a repair procedure to replace the affected resource, the
                 identity of which can be determined using 'fmadm faulty'.

   • If no fault is reported, you do not need to do anything else. Do not perform the subsequent steps.
   • If a fault is reported, continue to Step 3.
3. Clear the fault from all persistent fault records.

   In some cases, even though the fault is cleared, persistent fault information remains and results in erroneous fault messages at boot time. To ensure that these messages are not displayed, run the following Oracle Solaris command:

   # fmadm replaced /SYS/MB

   For the component referenced in the example in Step 2 (the motherboard) enter the following command:

   # fmadm replaced ********-****-****-****-************

   Note that you should use one of the following fmadm commands released with the Oracle Solaris 10 05/09 OS based on whether you are replacing a component or whether you have repaired the component in some way:

   • Use fmadm repaired when a physical repair is performed to resolve a problem with a faulted component. For example, use this command if you have reinstalled a card after straightening a bent pin and reseating the card.

   • Use fmadm replaced when you install a new component to replace a faulted component and the new component is not automatically discovered by the system.

     ---

     Note - The system generally detects new components when a new serial number is introduced into the system, but in cases where fault information remains, the fmadm replaced command will clear any faults that persist in the system after discovery.

     ---

     Refer to the fmadm man page for more information about these commands.

4. Use the clear_fault_action property of the FRU to clear the fault.

   -> set /SYS/MB clear_fault_action=True
   Are you sure you want to clear /SYS/MB (y/n)? y
   set 'clear_fault_action' to 'true

Related Information

• Oracle Solaris Predictive Self-Healing Technology Overview

• PSH-Detected Fault Example