Fujitsu M10-4S: How to Replace A CMUL with a Broken XSCFU on a Multiple BB M10-4S

Asset ID:	1-71-2007005.1
Update Date:	2018-04-05
Keywords:

Solution Type Technical Instruction Sure

Solution 2007005.1 : Fujitsu M10-4S: How to Replace A CMUL with a Broken XSCFU on a Multiple BB M10-4S

Applies to:

Fujitsu M10-4S - Version All Versions and later
Information in this document applies to any platform.

Goal

Provide steps to service a faulty XSCFU on a M10-4S multiple building block system.

System has multiple M10-4S BB’s ( building blocks ).
An XSCFU on a BB has failed, therefore the master XSCFU has no way to manage the BB with the failed XSCFU.
The system has multiple PPAR’s

Solution

The XSCFU is part of the CMUL Field Replaceable Unit ( FRU ) and should not be swapped to a different CMUL.

There are two options to repair the CMUL:

Option1: Stop the entire system (all the PPAR’s) to repair it.
Option2: Stop the affected PPAR to repair it.

In either case, the entire system or the affected PPAR will need to be powered off. Turning off the power to the affected BB may partially fail.

You can power off the system using either the poweroff command on master XSCFU or the shutdown(1M) / init(1M) command on Solaris.

When the poweroff command is used, the command will be issued from master XSCFU thru the other building block XSCFU’s to Solaris.

When the PPAR consists of multiple BB’s, the master XSCFU still has a path, via the other BB, to initiate shutdown. Therefore, shutdown will be initiated and should complete properly. BUT, it will fail to turn off the power of the failed BB

When the PPAR consists of single BB, the master XSCFU does not have a path to initiate shutdown. Therefore, shutdown itself won’t be started. In this case Solaris will need to be shutdown from the domain using either shutdown(1M), init(1M), or ldm stop-domain (if it’s guest domain). The shutdown will complete successfully but powering off will not be possible on the failed BB.

To manage such situation:

Shutdown Solaris instances on the target PPAR, or shutdown all the Solaris instances as noted above and confirm instances are properly shutdown.
Disconnect power cable of the failed BB to power off the BB. Even if the BB has not started (Powered) at the moment of failure, it is recommended to power off the failed BB. (Because the communication between the master XSCF and broken XSCF is at strange state).
Start normal repair procedure:

To start maintenance on a powered off system, disconnect power, and start maintenance.
To start maintenance work on single powered off PPAR (with other PPARs running), use the replacefru command on the master XSCF.

Attachments

This solution has no attachment