A P P E N D I X A |
Troubleshooting |
This chapter consists of tables of the most common issues you might experience with error messages you see and troubleshooting suggestions. It contains the following sections:
If you are having problems connecting to the network, check your cabling to ensure that the switch is properly connected to the management port as well as the InfiniBand ports and connectors. Then refer to Diagnosing Switch Indicators to verify that the corresponding port on the switch is functioning properly.
If you have connected a device to a port on the switch, but the Link LED is off, then check the following items:
Verify that all system components have been properly installed. If any network cabling appears to be malfunctioning, test it in an alternate environment where you are sure that all the other components are functioning properly.
If a port does not work, check the following:
The Sun IB switch logs information about important system events to both internal non-volatile event logs and Solaris host-based system logs (using syslog mechanism). The internal non-volatile event log is used only by ALOM. You can view its contents by using the ALOM showlogs command.
The Solaris host based system log is used by both InfiniBand management software and ALOM. System logs can be accessed through usual Solaris syslog means.
The contents of the switch's non-volatile event log can be seen using the ALOM showlogs command.
If a fault occurs, ensure that the problem you encountered is actually caused by the switch. If the problem appears to be caused by the switch, then follow these next steps.
1. Use the showlogs command with the appropriate option.
2. Repeat the sequence of commands or other actions that led to the error.
3. Make a list of the commands or circumstances that led to the fault.
4. Make a list of any error messages displayed.
If the SUNWsibs9p package is installed and set up for the switch, event messages from the switch will occur on the host (See System Log Proxy). Some messages are informational, others represent errors. The messages can be divided into two categories: messages from the switch platform itself (TABLE A-1 and TABLE A-2) and messages from the InfiniBand management software (TABLE A-3).
TABLE A-2 shows informational messages from the platform.
The switch was rebooted by the watchdog due to a hang of the system. |
The InfiniBand (IB) management software is responsible for setting up and controlling all the IB devices that are connected together. The events reported by this software can therefore be for some other device than the switch where the software is running. In InfiniBand a port on a IB device (switch or channel adapter card) is uniquely identified by a value called PortGUID. This value is often displayed in the IB related syslog messages. This value is also used by the showib command when listing the connections in the IB topology.
Many of the messages also contain IB-specific parameters. Refer to the InfiniBand Specification, Volume 1 for an exact explanation. All the IB related syslog messages contain a prefix of the type IBSRM event event_number. This part of the message is omitted from TABLE A-3 to make it more readable.
For more information about the InfiniBand Specification go to:
You must register before you can download specifications.
TABLE A-4 provides a list of common ALOM difficulties and their solutions.
This section contains information about certain types of error messages you might see when using the ALOM command shell:
These messages appear in response to a command you typed at the sc> prompt.
TABLE A-5 describes usage error messages that are displayed when you typed the command using improper command syntax. Refer to the description of the command for the correct syntax.
TABLE A-6 describes the possible CLI messages.
TABLE A-7 describes additional important messages or prompts for confirmation.
TABLE A-8 lists general errors that ALOM reports.
TABLE A-9 lists the error messages that appear when ALOM detects problems with field-replaceable units (FRUs) or customer-replaceable units (CRUs).
Note - The software refers to both FRUs and CRUs as FRUs, as in removefru or showfru. |
Copyright © 2004, Sun Microsystems, Inc. All Rights Reserved.