IBM eX5 Implementation Guide

Chapter 6. IBM System x3850 X5 and x3950 X5 299

For more information and to download Electronic Service Agent, go to the following website:

https://www-304.ibm.com/support/electronic/portal/navpage.wss?category=5&locale=en

_US

6.11.4 Recovery process

This section provides an overview of several recovery procedures. These procedures do not

replace the Problem Determination and Service Guide. You can solve many problems without

outside assistance by following the troubleshooting procedures in the Problem Determination

and Service Guide, which is available at the following website:

http://ibm.com/support/entry/portal/docdisplay?lndocid=MIGR-5084848

The Problem Determination and Service Guide describes the diagnostic tests that you can

perform, troubleshooting procedures, and explanations of error messages and error codes.

If you have completed the diagnostic procedure and the problem remains, and you have

verified that all code is at the latest level, and all hardware and software configurations are

valid, contact IBM or an approved warranty service provider for assistance.

Two-node configuration does not power on after a failure

The following list provides useful tips for bringing a 2-node configuration back online after a

failure:

򐂰 Check the IMM Event Log.

򐂰 Check the LEDs on the light path diagnostic panel.

򐂰 Make sure that the power cord is connected to a functioning power source.

򐂰 Make sure that the power cord is fully seated in the power supply.

򐂰 Check the power-supply LEDs.

򐂰 Check the QPI link LEDs. When the QPI link LEDs are lit, they indicate that the QPI links

are fully established.

򐂰 Remove all power cords from both nodes. Wait a few seconds and replug the cords in the

following order: Top PS 2, Bottom PS 1, Top PS 1, and Bottom PS 2. You will need to have

power to all nodes at relatively the same time to ensure that the IMMs can communicate

across the QPI cables.

򐂰 Remove the QPI cables to separately debug the servers. Before you remove the QPI

cables, the power

must be removed.

One node fails in a 2-node configuration

In certain cases, resolving an issue in a 2-node configuration takes longer because it is a

difficult task. When you need the system back online as quickly as possible, you can operate

the system as a single-node configuration. Use the following steps to configure the node that

is still working:

1. Remove all power cords from the working node.

2. Remove all QPI cables from the working node.

3. If available, install the QPI wrap cards.

4. If the working node is not fully populated with memory, add the memory of the second

node. Remember the installation sequence for the memory.

300 IBM eX5 Implementation Guide

5. If the working node does not have hard drives installed, add the hard drives with the

operating system from the second node. At the boot sequence, you will get a prompt to

import the RAID configuration.

6. Reconnect the power cords.

7. Power on the system.

Power failure on the MAX5

The MAX5 is turned off only if the connected server issues a power-off request and you have

disconnected the MAX5 power cord from the power source. You cannot turn off the MAX5

expansion module manually. Use the following steps to boot your system without the MAX5:

1. Remove all power cords from the MAX5 and x3850 X5.

2. Remove all QPI cables.

3. If available, install the QPI wrap cards for the x3850 X5.

4. Reconnect the power cords to the x3850 X5.

5. Power on the system.