Sun Fire[TM] V1280, E2900, or Netra 1280, 1290: a Solaris[TM] shutdown followed by a lom poweroff and poweron sequence can cause an unplanned SC reset.

Asset ID:	1-72-1006676.1
Update Date:	2012-01-20
Keywords:

Solution Type Problem Resolution Sure

Solution 1006676.1 : Sun Fire[TM] V1280, E2900, or Netra 1280, 1290: a Solaris[TM] shutdown followed by a lom poweroff and poweron sequence can cause an unplanned SC reset.

Applies to:

Sun Netra 1280 Server
Sun Fire V1280 Server
Sun Fire E2900 Server
Sun Netra 1290 Server - Version: Not Applicable and later [Release: N/A and later]
All Platforms

Symptoms

System Controller (SC) reset could occur unexpectedly.

A Sun Fire[TM] V1280, E2900, or Netra 1280, or 1290 can suffer an unplanned System Controller(SC) reset if Solaris[TM] on the domain is shutdown followed by an immediate poweroff and then poweron sequence at the lom prompt.

Cause

The behavior appears something like the following:

During the poweron, the fantray (FT) and PSUs power on without any issues.
An SC reset occurs while trying to poweron a Repeator (RP) or a System Board (SB).
A full hardware reset and POST execution follows.

The following example comes from the console log of a system that encountered this exact issue:

lom> poweron

/N0/PS0: powered on
Sat Apr 15 10:21:00 tkxhst21-sc lom: SysEvent 5 /N0/PS0, power state, On (5,1)
/N0/PS1: powered on
Sat Apr 15 10:21:01 tkxhst21-sc lom: SysEvent 5 /N0/PS1, power state, On (5,1)
/N0/PS2: powered on
Sat Apr 15 10:21:02 tkxhst21-sc lom: SysEvent 5 /N0/PS2, power state, On (5,1)
/N0/PS3: powered on
Sat Apr 15 10:21:03 tkxhst21-sc lom: SysEvent 5 /N0/PS3, power state, On (5,1)
/N0/FT0: powered on
Powering boards on ...

... <<--------------- The SC and ultimately, the hardware reset, occurs here and the remainder of the log shows SCPOST being executed.

@(#) SYSTEM CONTROLLER(SC) POST 43 2005/02/22 15:29
PSR = 0x044010e5
PCR = 0x04004000


Memory size = 32MB
SelfTest running at DiagLevel:0x40
[...]

Solution

The SC reset caused by this sequence of events is very rare (and harmless).
But, you can avoid being surprised by this behavior by running "resetsc" at the lom prompt before running the poweron command.

lom> poweroff
lom> resetsc   <<< Add this step!
lom> poweron

Executing resetsc will free up SC resources and place it in a "known" state. The poweron command will then power on all boards without any issues.

The poweroff command by itself does not remove standby (or housekeeping) power from the SC. Occasionally the SC Operating System (RTOS) can get out of sync with other boards after those boards are power cycled. A resetsc, reboots the RTOS and initializes various hardware registers to a "known" state. Powering off the SC by removing and re-inserting the four power leads from the V1280, E2900, or netra 1280, 1290 will also reboot the RTOS on the SC and have the same desired result.

The unplanned SC reset is harmless and the subsequent execution of POST usually results in all boards passing without any issues. The SC is capable of resetting itself when necessary, but to avoid the inconvenience of running POST on all system boards (and any delays as a result), run resetsc before powering on all components and save yourself some time and hassle of worrying if there is a problem.

Product
Sun Fire V1280 Server
Netra 1280 Server
Netra 1290 Server
Sun Fire E2900 Server

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.

Bug 5060577 describes the above problem and is unresolved.

Attachments

This solution has no attachment