Sun Enterprise™ xx00 Problem Solving Manual

Sun Enterprise™ xx00 Problem Solving Manual
Sun Enterprise™ xx00
Problem Solving Manual
Error Recognition, Actions, and Reporting Guide
Sun Microsystems, Inc.
901 San Antonio Road
Palo Alto, CA 94303-4900 U.S.A.
650-960-1300
Part No. 910-4188-11
February 2002, Revision A
Send comments about this document to: [email protected]
Sun Proprietary/Confidential: Internal Use Only
Copyright 2002 Sun Microsystems, Inc., 901 San Antonio Road, Palo Alto, CA 94303-4900 U.S.A. All rights reserved.
This product or document is distributed under licenses restricting its use, copying, distribution, and decompilation. No part of this product or
document may be reproduced in any form by any means without prior written authorization of Sun and its licensors, if any. Third-party
software, including font technology, is copyrighted and licensed from Sun suppliers.
Parts of the product may be derived from Berkeley BSD systems, licensed from the University of California. UNIX is a registered trademark in
the U.S. and other countries, exclusively licensed through X/Open Company, Ltd.
Sun, Sun Microsystems, Sun Enterprise, the Sun logo, AnswerBook2, docs.sun.com, sun.com, and Solaris are trademarks, registered
trademarks, or service marks of Sun Microsystems, Inc. in the U.S. and other countries. All SPARC trademarks are used under license and are
trademarks or registered trademarks of SPARC International, Inc. in the U.S. and other countries. Products bearing SPARC trademarks are
based upon an architecture developed by Sun Microsystems, Inc. The Energy Star logo is a registered trademark of EPA.
The OPEN LOOK and Sun™ Graphical User Interface was developed by Sun Microsystems, Inc. for its users and licensees. Sun acknowledges
the pioneering efforts of Xerox in researching and developing the concept of visual or graphical user interfaces for the computer industry. Sun
holds a non-exclusive license from Xerox to the Xerox Graphical User Interface, which license also covers Sun’s licensees who implement OPEN
LOOK GUIs and otherwise comply with Sun’s written license agreements.
Federal Acquisitions: Commercial Software—Government Users Subject to Standard License Terms and Conditions.
DOCUMENTATION IS PROVIDED “AS IS” AND ALL EXPRESS OR IMPLIED CONDITIONS, REPRESENTATIONS AND WARRANTIES,
INCLUDING ANY IMPLIED WARRANTY OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE OR NON-INFRINGEMENT,
ARE DISCLAIMED, EXCEPT TO THE EXTENT THAT SUCH DISCLAIMERS ARE HELD TO BE LEGALLY INVALID.
Copyright 2002 Sun Microsystems, Inc., 901 San Antonio Road, Palo Alto, CA 94303-4900 Etats-Unis. Tous droits réservés.
Ce produit ou document est distribué avec des licences qui en restreignent l’utilisation, la copie, la distribution, et la décompilation. Aucune
partie de ce produit ou document ne peut être reproduite sous aucune forme, par quelque moyen que ce soit, sans l’autorisation préalable et
écrite de Sun et de ses bailleurs de licence, s’il y en a. Le logiciel détenu par des tiers, et qui comprend la technologie relative aux polices de
caractères, est protégé par un copyright et licencié par des fournisseurs de Sun.
Des parties de ce produit pourront être dérivées des systèmes Berkeley BSD licenciés par l’Université de Californie. UNIX est une marque
déposée aux Etats-Unis et dans d’autres pays et licenciée exclusivement par X/Open Company, Ltd.
Sun, Sun Microsystems, Sun Enterprise, le logo Sun, AnswerBook2, docs.sun.com, sun.com, et Solaris sont des marques de fabrique ou des
marques déposées, ou marques de service, de Sun Microsystems, Inc. aux Etats-Unis et dans d’autres pays. Toutes les marques SPARC sont
utilisées sous licence et sont des marques de fabrique ou des marques déposées de SPARC International, Inc. aux Etats-Unis et dans d’autres
pays. Les produits portant les marques SPARC sont basés sur une architecture développée par Sun Microsystems, Inc.
L’interface d’utilisation graphique OPEN LOOK et Sun™ a été développée par Sun Microsystems, Inc. pour ses utilisateurs et licenciés. Sun
reconnaît les efforts de pionniers de Xerox pour la recherche et le développement du concept des interfaces d’utilisation visuelle ou graphique
pour l’industrie de l’informatique. Sun détient une licence non exclusive de Xerox sur l’interface d’utilisation graphique Xerox, cette licence
couvrant également les licenciés de Sun qui mettent en place l’interface d’utilisation graphique OPEN LOOK et qui en outre se conforment aux
licences écrites de Sun.
LA DOCUMENTATION EST FOURNIE “EN L’ETAT” ET TOUTES AUTRES CONDITIONS, DECLARATIONS ET GARANTIES EXPRESSES
OU TACITES SONT FORMELLEMENT EXCLUES, DANS LA MESURE AUTORISEE PAR LA LOI APPLICABLE, Y COMPRIS NOTAMMENT
TOUTE GARANTIE IMPLICITE RELATIVE A LA QUALITE MARCHANDE, A L’APTITUDE A UNE UTILISATION PARTICULIERE OU A
L’ABSENCE DE CONTREFAÇON.
Please
Recycle
Sun Proprietary/Confidential: Internal Use Only
Contents
1.
2.
3.
Sun Enterprise xx00
Problem Solving 1-1
1.1
Overview
1-1
1.2
System Problem Categories
1-2
Failure Detection and
Data Retrieval Tools 2-1
2.1
Power-on Self-Test (POST)
2-2
2.2
OpenBoot (OBP)
2.3
OBP Debug Command
2.4
kadb Command
2.5
Solaris System Commands and
Data Files 2-6
2-4
2-5
2-5
2.5.1
prtdiag Command
2-6
2.5.2
prtconf Command
2-7
2.5.3
adb Command
2-7
2.5.4
mdb Command
2-7
2.5.5
crash Command
2.5.6
Solaris Log Files
Failure Scenarios
2-8
2-8
3-1
iii
Sun Proprietary/Confidential: Internal Use Only
4.
5.
3.1
System Panic
3-1
3.2
System Soft Hang
3.3
System Hard Hang
3.4
Fatal Reset/Fatal Error
Fatal Reset Diagnosis
3-3
3-5
3-7
4-1
4.1
Console Information
4-1
4.2
Special Fatal Reset Considerations
4.3
Diagnosis #1 - UPA_A_ERR
(Etag Parity Error) 4-4
4.4
Diagnosis # 2 - UPA_A_ERR (Etag Parity Error) Multiple Errors
4.5
DT_PERR (DTAG Parity Error)
4.6
Diagnosis # 4 - DT_PERR (DTAG Parity Error) Multiple Errors
4.7
Diagnosis #5 - FTA_PERR
4-12
4.8
Diagnosis #6 - FTC_PERR
4-13
4.9
Diagnosis #7 - MTIMEOUT
4.10
Diagnosis #8 - FTUPAOV
4.11
Diagnosis #9 - Cacheable Write Error
4.12
Diagnosis #10 - Non-Cacheable Write
4.13
Diagnosis #11 - Interrupt Error
4-2
4-6
4-8
4-10
4-14
4-16
Enhanced Solaris Error Messaging
4-17
4-19
4-20
5-1
5.1
Improved Error Messages
5-1
5.2
Errors and Events
5.3
Details on Improved Error Handling
5-3
5.4
Details on Improved Error Messages
5-4
5.5
Error Message Categories
5-4
5.6
Error Messages Examples
5-6
5.7
EDP Event - Ecache Data Parity Event
5-2
iv
5-7
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
6.
5.8
Trap Level 1 Panic
5.9
WP Event - Writeback Data Parity Error
5.10
CP Event - Copyout Data Parity Error
5.11
UE Event - Uncorrectable Memory Error
5.12
BERR Event - Bus Error
5.13
CE Event - Correctable Memory Error
Configuration Steps
5-8
5-10
5-11
5-13
5-15
5-16
6-1
6.1
Step 1 - Initial System Configuration
6.2
Step 2 - Error Identification and
Error Response 6-3
6.3
Step 3 - Information Gathering and
Follow-up Tasks 6-6
A. Enabling Saving System Dump
6-2
A-1
A.1
Enabling savecore(1M) and Verifying Disk Space
A.2
Verify Core Dump Process
B. ISCDA Script
A-1
A-4
B-1
C. System Abort Sequences
C-1
C.1
L1/a Stop/a Keyboard Abort Sequences (Local Terminal Connection)
C.2
Abort Sequence via a Serial Terminal Server
C.3
Abort Sequence via a Direct Connect ASCII Terminal
C.4
Remote System XIR and Remote Power Control Sequences
D. OBP & kadb Commands
C-2
C-4
C-4
D-1
D.1
OBP Debug Commands
D-1
D.2
kadb Debug Commands
D-2
E. System Console Logging
C-1
E-1
Contents
Sun Proprietary/Confidential: Internal Use Only
v
E.1
Purpose for Console Logging
E.2
Configuring the Console on ttya
E.3
Console Logging Options - Data Logging Terminal Servers
E.4
Console Logging Options - Centralized Console Control
E.5
Console Logging Options - Tip line to ttya
F. LED Status Indicators
F.1
Overview
E-2
E-5
G-1
Device Driver Acronyms
G.2
Device Path Example
G.3
Device Path Decoding Tables
G-1
G-1
G-3
H-1
H.1
prtdiag Command
H.2
Location of the prtdiag Command (sun4u and sun4d)
H.3
Use of the prtdiag Command (sun4u)
H-1
I. Using the Explorer Utility
Installing Explorer
I-1
I.2
Executing Explorer
I-2
J. FTP Access to Sun
K. Appendix K - SunVTS
L.1
H-2
J-1
Instructions for USA
SunVTS Versions
L. Forth Debug
H-2
I-1
I.1
K.1
E-4
F-1
G.1
J.1
E-4
F-1
G. Decoding xx00 Device Paths
H. Use of prtdiag
E-1
J-1
K-1
K-1
L-1
Forth Debug - Explanation and Setting
vi
L-1
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
M. Sun Enterprise xx00 Firmware Levels
M.1
Purpose of Flash Prom Updates
M.2
Flashprom Version Matrix
N. ESD Handling & Tools
M-1
M-1
M-2
N-1
N.1
Electrostatic Discharge (ESD)
N-1
N.2
ESD Damage
N.3
ESD Control
N.4
Electrostatic Voltage At Workstations
N-1
N-2
N-3
Contents
Sun Proprietary/Confidential: Internal Use Only
vii
viii
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
CHAPTER
1
Sun Enterprise xx00
Problem Solving
1.1
Overview
Sun Microsystems offers the Sun Enterprise xx00 Server product line - a reliable,
adaptable, maintainable, and performance-oriented product.
Sun Microsystems continually strives to produce robust reliable products. It is also
Sun’s objective to move quickly on any issues and to apply the necessary product
expertise to resolve the issues in a timely manner. To this end, Sun Microsystems has
a world-class support organization, Sun Enterprise Service, which is designed to
facilitate the support of the entire Sun product line. If a problem with the Sun
Enterprise Server occurs, the first step is to identify that a problem has occurred and
then to notify Sun using the recommended service call reporting processes for your
specific situation.
1-1
Sun Proprietary/Confidential: Internal Use Only
1.2
System Problem Categories
The root cause of a problem can vary. Examples include unforeseen hardware
errors/failures, Solaris specific bugs, and third-party software problems.
The types of problems that one might encounter can be grouped into six main
categories:
■
System Panics: A system panic (crash) dumps specific areas (but not all) of active
memory so that a system dump (vmcore) file can be created. This vmcore file is
then analyzed to determine if the cause of the panic was software or hardware.
■
System Soft Hangs: This type of hang is characterized by a non-responsive userinterface. The affected system should still respond to network activity, such as the
ping command from a remote system.
The affected system should also respond to a system abort sequence. Refer to
Appendix C for detailed information on abort sequences.
The system LEDs should still be in their normal flashing/cycling state.
■
System Hard Hangs: This type of hang is also characterized by a non-responsive
user-interface.
Unlike the soft hang, all attempts to interrupt the system will fail. All network
activity is non-responsive. The system LEDs may also be in a frozen state.
■
Hardware Fatal Resets This is a result of an ‘‘illegal’’ hardware state being
detected.
A hardware fatal error can either be a transient error or a hard error. A transient
error is one that intermittently fails. A hard error is one that consistently fails in
the same way.
■
System Drop into OpenBoot (OBP): This type of interrupt is typically
characterized by an unresponsive system, and when console access is obtained,
the system is found at the OpenBoot or "ok" prompt.
Certain failure modes, such as Solaris stack overflow errors, can cause exception
conditions known as redstate conditions. This condition will cause Solaris to abort
in such a way that OpenBoot will not automatically react and reboot the system.
In some cases, a break signal may have been issued, by someone with console
access or by a spurious signal from a serial terminal server. In these cases the
system has not actually had an internal hardware or software problem, but was
simply halted.
Many times a system will be said to have experienced an unexpected system reboot.
Care must be taken to verify that this condition was not caused by a panic or Fatal
Reset.
1-2
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
Although the system appears to unexpectedly reboot, in many cases it is due to a
panic or Fatal Reset. Thus the need to be specific when determining why a system
interrupted.
E-Cache parity errors that occur in user space (not kernel space) may cause Solaris to
simply log the E-Cache failure data and reboot the system. The reason is that the
error information that is logged is sufficient to troubleshoot and correct the problem.
This reboot saves the time of taking a system panic dump. It also by-passes the need
to run fsck(1M) on file systems as they are un-mounted during the reboot process.
This new reboot feature for user level E-Cache parity errors is a function of changes
made in kernel update patches containing the kernel level scrubber for Solaris 2.6
and above. The KU patches are 105181-23 and higher for Solaris 2.6, 106541-13 and
higher for Solaris 7 and 108528-04 and higher for Solaris 8.
Chapter 1
Sun Enterprise xx00 Problem Solving
Sun Proprietary/Confidential: Internal Use Only
1-3
1-4
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
CHAPTER
2
Failure Detection and
Data Retrieval Tools
The Sun Enterprise xx00 Server product line ships with a number of software and
firmware tools used for generating, gathering, and analyzing data. These tools
include:
Firmware
■
■
■
■
■
■
■
■
■
■
Power On Self Test (POST)
OpenBoot (OBP)
NVRAM Variables
Standalone Solaris Debugger
kadb
Solaris Debuggers and Commands
prtdiag
prtconf
adb
crash
Solaris Messages
■
■
■
Console Access/Output
Solaris log files (/var/adm/messages)
XIR, Key Switch, Solaris system logs, Sun Management Center
2-1
Sun Proprietary/Confidential: Internal Use Only
2.1
Power-on Self-Test (POST)
POST has two main objectives: to test the system hardware (excluding peripheral
devices) to help diagnose a failing hardware component; and to configure the
available hardware for use by OpenBoot (OBP) and then Solaris.
POST resides in the OpenBoot (OBP) PROM on each CPU/Memory board, I/O
board, and disk board. POST controls the status LEDs on the system front panel and
all boards and displays diagnostic and error messages on a console terminal (ttya
serial port), if available.
POST is sometimes referred to as an “off-line” diagnostic tool which can be used to
verify the integrity of the hardware on the system. “Off-line” is defined under the
condition where Solaris is not running and POST is executing by itself.
Only POST can configure the system hardware, and only POST can enable hotpluggable boards. If a new unit (board or modular power supply) is added to the
card cage after the system has booted, the new unit will not work until the system is
rebooted, at which time POST reconfigures the system, using the units that are
found in the system at that time.
To capture the results of POST, an active console session is required. It is preferred to
have a console connection via the serial ttya port (serial port A) to the system. This
connection should be capable of logging the output via a large scroll buffer in the
terminal window being used or preferably via logging the console session to a file
using a tool such as "script".
POST can be invoked by one of the following methods:
■
Issuing the "reset-por" command at the OBP prompt:
Issuing the "reset-por" command at the OBP prompt causes POST to be entered
and executed. A physical power-cycle of the system will also cause full POST to
be run.
■
Key switch in Diagnostic position
There is a manual key switch interface to POST located on the front panel of the
Sun Enterprise xx00 Server. If the key switch is turned to the "Diagnostic"
position, each time the system is rebooted, POST will be executed at the
maximum level.
■
Setting the OBP value, diag-switch?, to TRUE
The diag-switch? OBP variable is FALSE by default. Setting diag-switch? To TRUE
will cause POST to be run at the level defined by the OBP variable diag-level
during the next system reboot.
2-2
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
There are basically three POST levels that can be defined by the OBP diag-level
variable. The levels are off, min, and max. The max level provides the most
comprehensive set of POST tests.
The key switch position in Diagnostic mode, as described above, takes precedence
over these OBP variables.
■
System encounters a Fatal Reset (hardware exception)
If the system detects an unrecoverable hardware exception, a Fatal Reset will
result. The system stops executing Solaris immediately and does not log any error
or failure information to the Solaris log files, i.e. /var/adm/messages. POST is
invoked as part of the system recovery (reboot) process.
Hardware components that fail POST will be marked as failed and will not be
used during this boot sequence. These components will remain disabled across
multiple system reboots. However, if the system is power cycled, the POST
failures are cleared. If these components fail POST again, they are again disabled.
If these components pass POST, they are then enabled and used in the system
configuration.
The list of failed components will persist over time and as such new failures will
accumulate after hardware errors. Failed component information will be cleared
after either of the following events:
■
The “reset-por” command from OBP (ok prompt) is issued.
■
A physical power-cycle of the system is performed. This power cycle normally
occurs during hardware replacement activities.
Note – IMPORTANT! Based on the information above, it is critical that POST output
(console output) be logged so that it can be analyzed further as necessary. It is
critical to use the prtdiag -v command PRIOR to any system power cycle in order
to capture detailed failure information that is stored in EPROM.
Chapter 2
Failure Detection and Data Retrieval Tools
Sun Proprietary/Confidential: Internal Use Only
2-3
2.2
OpenBoot (OBP)
The OpenBoot (OBP) firmware is stored in the boot PROM (programmable read-only
memory) of a system so that it is executed immediately after you turn on your
system. The primary task of the OpenBoot firmware is to boot the operating system
from either a mass storage device or from a network. The firmware also provides
extensive features for testing hardware and software interactively.
■
OBP Diagnostic and Information Commands on Table 2-1 show commands that
are available at the OBP (ok) prompt and allow the user the ability to perform
some low level tests of installed hardware.
These on-board tests allows the user to check devices such as the network
controller, the floppy disk system, memory, installed SBus cards and SCSI devices,
and the system clock. User-installed devices can be tested if these devices contain
firmware which includes a self-test feature.
TABLE 2-1
OBP Diagnostic/Information Commands
OBP Command
Description
probe-scsi
Identify devices attached to the built-in SCSI bus.
probe-scsi
Perform probe-scsi on all SCSI buses installed in the system
below the specified device tree node. If device-path is absent,
the root node is used.
test net
Test the network connection
test floppy
Test the floppy drive, if installed
test memory
Test number of megabytes specified in the selftest-#megs
NVRAM parameter; or test all of memory if diag-switch? is true
test-all
[devicespecifier]
2-4
watch-net
Monitor the network connection.
watch-clock
Test the clock function.
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
2.3
OBP Debug Command
OBP Debug Commands on TABLE 2-2 show commands that are available at the OBP
(ok) prompt. They allow the user the ability to perform some low level data
gathering.
These commands are normally executed when a system was running Solaris and
then hung and was forced (aborted) into OBP, or the system panic’d and the panic
dump process failed.
TABLE 2-2
2.4
OBP Debug Commands
OBP Command
Description
.registers
Display values in %g0 through %g7, plus %pc, %npc, %psr, %y,
%wim, %tbr.
.trap-registers
Display values in the trap related registers.
.locals
Display the values in the i, l and o registers.
.psr
Display the processor status register.
ctrace
Display the Solaris return stack showing C subroutines.
sync
Cause Solaris to attempt to save a system panic (vmcore).
kadb Command
kadb is a low-level kernel debug tool that is available from the OBP level. In normal
operation, kadb is not enabled because it is not normally required. To enable kadb,
Solaris must be booted with the kadb option. The following command is an
interactive method to enable kadb for a single boot.
ok> boot kadb
If the system is rebooted or interrupts (panic, Fatal Reset, etc.), the system will not
boot in kadb mode. You can set the boot-file OBP variable to enable kadb boots.
ok> setenv boot-file kadb
ok> boot kadb
Chapter 2
Failure Detection and Data Retrieval Tools
Sun Proprietary/Confidential: Internal Use Only
2-5
Note – IMPORTANT! Booting the system in kadb mode should ONLY be set based
on specific direction from Sun Microsystems support engineers. Booting kadb mode
should NOT be default on all systems.
Once the system is booted in kadb mode, the user can then elect to enter into
kadb at the time of the issue/problem by sending a break signal via the console
connection.
However, if the nature of the error is preventing the system from correctly saving
a system panic dump (vmcore file), booting kadb can be useful. Once the system
panics, the system will drop into kadb and once inside kadb, the user can then
attempt to analyze the systems behavior further.
An excellent reference book on panics and the use of kadb to analyze them is,
Panic, Unix System Crash Dump Analysis - by C. Drake and K. Brown.
For a detailed explanation on the command, refer to the man pages for kadb(1M).
2.5
Solaris System Commands and
Data Files
The operating system (Solaris 2.5.1 is the initial version for the Sun Enterprise xx00
Server product lines) contains several very useful data generating/gathering
commands. Solaris also maintains data files containing useful information.
2.5.1
prtdiag Command
/usr/platform/sun4u/sbin/prtdiag–This command displays both the system
configuration and hardware error information.
The prtdiag(1M) command provides information on POST failures as well as Fatal
Reset information. When a Sun Enterprise Server takes a fatal reset, relevant
information is copied to a region of non-volatile memory. The prtdiag command
accesses this region and translates the information into a readable format. This
command is only available on the Sun4u and Sun4d architectures. The
recommended command syntax is:
# /usr/platform/sun4u/sbin/prtdiag -v
For a detailed explanation of the command, refer to the man page for
prtdiag(1M).
2-6
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
2.5.2
prtconf Command
/usr/sbin/prtconf–The output of the prtconf command is a subset of the
output from prtdiag.
The prtconf command can be useful for other detailed hardware configuration
information. For a detailed explanation of the command, refer to the man page for
prtconf(1M).
The recommended command sequence takes the form:
# /usr/sbin/prtconf -pv
2.5.3
adb Command
/usr/bin/adb–The adb command is an on-line debugging utility that can be used
to analyze the running image of Solaris or a system panic dump (vmcore file).
The adb command to investigate a system panic dump (vmcore file) is:
# cd /var/crash/‘uname -n‘
# /usr/bin/adb -k unix.? vmcore.?
Where "?" is the integer number for the dump being analyzed.
Analyzing a system panic dump using adb must be done on the same system
architecture as the system that originally generated the panic. An excellent
reference book which describes panics and the use of adb to analyze crash dumps
is, Panic, Unix System Crash Dump Analysis - by C. Drake and K. Brown. For a
detailed explanation on the command, refer to the man pages for adb(1M)
The adb command used to investigate the running Solaris image (as root) is:
# /usr/bin/adb -k
2.5.4
mdb Command
usr/bin/mdb–The Solaris Modular Debugger (mdb) is another on-line debugging
utility that can be used to analyze the running image of Solaris or a system panic
dump (vmcore file). The mdb debugger is only available in Solaris 8.
The syntax for invoking mdb is basically the same as for adb described above.
Refer to the mdb man page and other mdb documentation for further
information.
Chapter 2
Failure Detection and Data Retrieval Tools
Sun Proprietary/Confidential: Internal Use Only
2-7
2.5.5
crash Command
usr/sbin/crash–The crash command is used to examine the system memory
image of a running or a crashed system by formatting and printing control
structures, tables, and other information.
The command to investigate a system panic dump (vmcore file) is:
# cd /var/crash/‘uname -n‘
# /usr/sbin/crash -n unix.? -d vmcore.?
Where "?" is the integer number for the dump being analyzed.
Analyzing a system panic dump using crash must be done on the same system
architecture as the system that originally generated the panic. For a detailed
explanation on the command, refer to the man pages for crash(1M)
The crash command used to investigate the running Solaris image (as root) is:
# /usr/sbin/crash -n /dev/ksyms -d /dev/mem
2.5.6
Solaris Log Files
/var/adm/messages–This is an ASCII text data file that is the standard repository for
all Solaris generated messages.
The messages file is archived over time and contains system warnings, errors, and
notifications. Messages and errors from system memory and peripheral controllers
and devices are what is primarily logged in these files.
The messages files also provides a time-stamped record of system events. It provides
a timeline against which system behavior can be compared for corroborating or
supporting evidence.
2-8
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
CHAPTER
3
Failure Scenarios
The following sections describes system failure scenarios and the information that is
required to perform the most complete and timely diagnosis of the system interrupt,
based on each type of failure scenario.
In many cases, the Sun Solution Center may ask that the system data gathering tool
known as "Explorer" be run. This will collect some of the information below, as well
as a number of other data components.
3.1
System Panic
For more information on system panics, refer to, Panic!: Unix System Crash Dump
Analysis by C. Drake and K. Brown.
Commands and/or different command combinations can and should be tailored to
the type of system exception being investigated. The output from the following list
of commands will provide an initial picture of what happened on the Sun Enterprise
Server xx00 in question.
IMPORTANT! It is also critical to inspect the /var/adm/messages file for relevant
information.
The system reaction to a Solaris panic varies based on the following:
■
If the system is booted normally, and the OBP variable auto-boot? is set to "true",
the system should attempt to save a system dump (vmcore file) and reboot to
multi-user mode automatically.
Solaris commands that should be issued after the system has rebooted from the
panic are:
# /usr/platform/sun4u/sbin/prtdiag -v
# /usr/sbin/prtconf -pv
3-1
Sun Proprietary/Confidential: Internal Use Only
# cd /var/crash/‘uname -n‘
# iscda unix.? vmcore.? > iscda.out
Where "?" is the integer number for the dump being analyzed.
Note – iscda is not part of the standard Solaris release, refer to Appendix B.
■
If the system is booted normally, and the OBP variable auto-boot? is set to "false",
the system should attempt to save a system dump (vmcore file). However, the
system will reboot and stop at the OBP (ok) prompt.
If the system saved a valid vmcore file, then the system should be booted
immediately so that the system dump (vmcore file) and other system information
can be saved and analyzed for root cause. The Solaris commands above should be
run once the system is booted.
■
If the system fails to save a system dump, and the system drops into OpenBoot
(OBP), the following OBP commands would be run and the output
logged/captured:
ok>
ok>
ok>
ok>
ok>
ok>
ok>
printenv
.registers
.locals
.psr
.trap-registers
ctrace
sync
Note – IMPORTANT! If the system saved a valid dump or attempted to and failed,
the sync command below may fail. The sync command is normally used to save a
vmcore dump after the system has dropped from Solaris to OBP.
■
If the system was last booted under kadb, a system panic should drop the system
into kadb. If booted under kadb, execute the following kadb commands and
record all the output.
<sp$<stacktrace
$<threadlist
$<thread.brief
$<cpus
$<msgbuf
3-2
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
3.2
System Soft Hang
A general definition of a system soft hang is when the usability of the system
gradually or suddenly ceases. In some cases, soft hangs have been known to only
affect new attempts to access the system. A gradual onset is usually characterized by
progressively slower response until the system stops responding. A sudden onset is
when all or some of system response suddenly ceases.
Some soft hangs will dissipate on their own while others will require that the system
be interrupted, that information be gathered at the OpenBoot level, and the system
be rebooted. A soft hang should respond to a break signal that is sent via the system
console.
The possible causes of a system hang include memory leaks, software (most likely
kernel, I/O related or device driver problems), network problems, and even possibly
defective hardware.
In the case of a soft hang, try to determine the extent of the problem by doing the
following:
1. Record the state of the LEDs on the front of the system, if possible.
Note any amber LEDs and verify that the cycle LED (green flashing LED) is flashing
on all System Boards, Clock Board and Main LED on the system front panel.
2. Determine if any network activity is working (i.e. via ping, etc.) and if any
existing logins from other users are active/responding.
If other active logins are responding, determine the state of the network interface
and review the contents of /var/adm/messages for any indications of problems.
Use commands such as ifconfig, netstat, ping, etc. to determine the state of the
network and to attempt to make network contact with other systems.
3. Determine if a console logging session to the system can be made through the
console (ttya) connection.
If a working console connection can be established, then the problem may not be a
true hang, but instead a network related problem. This network problem could be
system hardware (network controller) or possibly a hardware problem in the
network infrastructure (network hub/router/cable).
For suspected network problems, attempt to ping, rlogin or telnet to the affected
system. If it’s not possible, attempt to ping, rlogin or telnet to another system that is
on the same sub-network/hub/router that the affected system is on.
If NFS services are served by the affected system, determine if NFS activity is
sluggish/non-existent on other systems.
Chapter 3
Sun Proprietary/Confidential: Internal Use Only
Failure Scenarios
3-3
4. If there are no responding login sessions, and a console login session is unable to
be made, attempt to break the system so that it will fall into OpenBoot (OBP)
mode.
If you manage to drop into OBP or kadb (depending upon how the system is
booted) then the hang is classified as a soft hang. The output from the following list
of commands will provide an initial picture of what happened on the Sun
Enterprise xx00 server in question.
For systems that are booted normally (without kadb), the following are useful
commands:
ok> printenv
ok> .registers
ok> .locals
ok> .psr
ok> .trap-registers
ok> ctrace
ok> sync
For systems booted under kadb, the following are useful commands:
<sp$<stacktrace
$<threadlist
$<thread.brief
$<cpus
$<msgbuf
Some decisions can then be made on the contents of the stack trace. For example,
the length of the stack trace and certain key routine names can indicate a rather
normal stack or an abnormal stack.
It is good practice to inspect the /var/adm/messages file for the following
information:
■
■
■
3-4
A large gap in the time stamp of Solaris/application messages
Indications of last root logins to determine if any administrators can add any
comments about the system state at the time of the hang
Warning messages about any hardware or software components
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
3.3
System Hard Hang
A system hard hang differs from a soft hang in one key area—a hard hang will not
respond to a system break sequence. For this reason, it can be difficult to determine
the root cause of a hard hang after a single occurrence.
Note – It is important to note the state of the system LEDs (i.e. are they cycling or
frozen/on solid). The state of the LEDs should be determined prior to initiating any
of the following system recovery operations.
In the case of a hard hang, try to determine the extent of the problem by doing the
following:
1. Record the state of the LEDs on the front of the system, if possible.
Note any amber LEDs and also verify that the cycle LED (green flashing LED) is
flashing on all System Boards, Clock Board and Main LED on the system front panel.
2. Determine if there are any active network connections that are still responding.
If no active connections respond, a new connection should be attempted using both
the console and the network. This includes attempting to ping the effected system.
If any login connection is successful, refer to the commands for the soft hang above
to attempt to recover the system and/or force a system panic.
3. If no active connections exist or can be made, attempt to break the system so that
it will drop into OBP or kadb, depending on how the system was last booted.
If the system is able to drop into OBP or kadb, refer to the appropriate command
sequences (in the soft hang situation) for each of these modes.
If the system is still not responding to the above access methods, the choices are
limited. Solaris may or may not be executing at this time. Due to the inability of
aborting to OBP/kadb, the user is left with using the XIR button at the rear of the
system. If that is unsuccessful in interrupting the system, attempt power-cycling the
system via the front panel key (all the way to the left is power off).
4. Attempt the XIR button. The XIR button should be tried first because it does not
power cycle the system, where a transient error may be reset/masked.
The XIR button is equivalent to issuing the reset command while in OBP. Note
that the XIR button may or may not cause the system to respond. This is
dependent upon the severity of the hard hang. If XIR is recognized, some system
information is saved.
To display the XIR information, execute the following command:
.xir-state-all
Chapter 3
Sun Proprietary/Confidential: Internal Use Only
Failure Scenarios
3-5
XIR failure data can then be sent to Sun and can be diagnosed at the following
internal Sun web site:
http://cte-www.uk/cgi-bin/xir-cgi.tcl
There is a key-sequence that can be used to try and initiate an XIR state. This key
sequence is two carriage returns, a tilde and the key sequence of the Control
Key/Shift Key/ and the letter x. For example:
<CR> <CR> <~> <Control-Shift-x>
To execute an XIR state using the key-sequence, the following criteria must be met:
■
The console must be connected to port A on the clock board.
■
The key switch must be in the On or Diagnostic setting. If it is in the Secure or Off
position, the remote key sequences and button resets are ignored.
■
Security features (such as OpenBoot security-mode) must be disabled.
■
The type speed must be no faster than 0.5 seconds and no slower than 5 seconds
between characters.
5. If the XIR button does not reset the system, power cycle the system to reset it and
begin the reboot sequence.
Note – Record the state of the system LEDs prior to powering the system off and
on.
Use the key switch to power off the system. It is recommended to wait at least 30
seconds before powering the system on with the key switch.
It is good practice to inspect the /var/adm/messages file for the following
information:
■
relevant data such as a large gap in the time stamp of Solaris/application
messages
■
indications of last root logins to determine if any administrators can add any
comments about the system state at the time of the hang
■
warning messages about any hardware or software components
If all attempts to gain control of the system or to gain additional information fails,
and the system must be power cycled to reset it, the Solaris deadman timer is one
method that will prevent all but a hardware failure from leading to a system hard
hang.
Note – The Solaris deadman timer should ONLY be enabled based on specific
instructions from Sun Microsystems. This mode should NOT be default on systems.
3-6
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
The deadman timer, based on the TOD chip, will hard reset the system if the timer
expires. To enable it, the appropriate /etc/system entry must be made and the
system rebooted for this to take effect.
set watchdog_enable=1
set snooping=1
By enabling the deadman timer, the system should drop into OBP, if the timer
expires. This is an indication of a system hang and the appropriate OBP commands
for a soft hang should be executed:
ok> printenv
ok> .registers
ok> .locals
ok> .trap-registers
ok> .psr
ok> ctrace
ok> sync
3.4
Fatal Reset/Fatal Error
A Fatal Reset/Fatal Error condition is most commonly a hardware fault that is
detected by the system. This condition is un-recoverable and continued operation of
Solaris would jeopardize the system because system integrity has been lost. Thus
Solaris is immediately terminated and as such no details of the Fatal Reset are
logged in common locations such as /var/adm/messages, etc.
There are a few cases of software (typically device driver problems) causing Fatal
Resets. These are rare and normally documented so that they can be identified
easily. Some of these driver bugs manifest themselves as timeout problems
(MTIMEOUT Fatal Resets). For past examples, refer to Sun Bug 4320047, 4306348
and 4230383. These bugs can be obtained via SunSolve Online (both internal or
external versions of SunSolve) or by contacting Sun Microsystems.
When a Fatal Reset is detected, the system will reset and enter into POST at
maximum diagnostic level (diag-level = max). Unfortunately, the most important
data that specifically defines what caused the Fatal Reset is displayed only to the
system console. If the system console is a frame buffer based monitor, the console
output is typically lost. This is also true if the console connection is via the ttya
serial port and the output is not being logged. The result is that important data from
the Fatal Reset, including the results of POST testing during the system recovery
process, is lost.
Chapter 3
Sun Proprietary/Confidential: Internal Use Only
Failure Scenarios
3-7
In the case of an intermittent error which caused the Fatal Reset, POST might not
find the offending component. In other cases of hard failure components, POST will
detect them, mark them as failed, and continue with the rest of the POST tests. The
service (yellow) LEDs both on the individual system board(s) that had the error and
the main service (yellow) LED on the front panel of the system should further define
the problem area. Any subsequent system reboots will show that the system has
‘‘off-lined’’ the offending component, where possible.
3-8
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
CHAPTER
4
Fatal Reset Diagnosis
This section defines a variety of possible Fatal Resets and the recommended
diagnosis for each type of error. Most of the following information is based on the
data obtained from the console at the time of the Fatal Reset. Where possible, output
from prtdiag -v is also shown.
4.1
Console Information
The single most valuable piece of information available from Fatal Resets is the
system console output at the time of the error. Due to the processing of some
Fatal Resets, the data in prtdiag may indicate components that have failed which
are not actually the root cause of the Fatal Reset. These innocent components
may be replaced, and the system may appear stable. This stability is not due to
the hardware replacement. The problem that is still in the system is transient and
there is only the illusion that the replacement hardware fixed the Fatal Reset.
IMPORTANT! Knowing the recent service history of systems which encounter
Fatal Resets can be valuable information to have when making the diagnosis.
IMPORTANT! Use of the prtdiag -v command BEFORE any hardware is
replaced and/or the system power cycled is a key data gathering process. Note
that the -v or verbose option must be used so that all the critical hardware failure
information is logged.
IMPORTANT! The FT_ARBERR and FT_SHERR error indicators are not always
valid in identifying a failing FRU. See Special Fatal Reset Considerations below.
4-1
Sun Proprietary/Confidential: Internal Use Only
4.2
Special Fatal Reset Considerations
Each Address Controller (AC) in the Sun Enterprise xx00 servers contains groups of
16 control wires. Two of these groups are the Arbitration wires and Shared wires.
An AC only drives one wire, the wire driven depends on the slot the board is in, for
example the board in slot 0 drives FT_ARB[0], board in slot 15 drives FT_ARB[15].
Parity cannot be calculated on these wires, so the AC that drives the wire samples
the value driven on the bus and compares it against the value that it drove.
During a Fatal Reset and a power cycle, the register in the AC that contains the slot
number is cleared, then the slot number is loaded back into the register.
The resets are supposed to be synchronized, so that every AC sees reset at the same
time. Unfortunately, a bug exists where the board that detects the fatal error gets
reset two cycles ahead of the rest of the boards in the system. If the board that
detects the fatal error is about to drive the arbitration wires, then it will drive
FT_ARB[0], even if it is Board 2. Board 0 detects the value on the arbitration wires is
not consistent with what it drove and sets the FT_ARBERR bit in the AC error
register. The same is true for the shared wires. In this case, the FT_SHERR error bit
is set in the AC error register.
4-2
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
The result of the bug mentioned above can be Fatal Resets that appear as follows:
Fatal Reset
0,0>FATAL ERROR
0,0>At time of error: System software was running.
0,0>Diagnosis: Board 0, centerplane pin, connector pin, AC
0,0>Diagnosis: Board 4, UPA PORT Device, AC
0,0>Log Date: Mar 23 2:58:10 GMT 2001
0,0>
0,0>RESET INFO for CPU/Memory board in slot 0
0,0>AC ESR 00000400.00000000 FT_ARBERR
0,0>DC[0] 00
0,0>DC[1] 00
0,0>DC[2] 00
0,0>DC[3] 00
0,0>DC[4] 00
0,0>DC[5] 00
0,0>DC[6] 00
0,0>DC[7] 00
0,0>FHC CSR 00050200 LOC_FATAL SYNC NOT_BRD_PRES
0,0>FHC RCSR 02000000 FATAL
0,0>
0,0>RESET INFO for CPU/Memory board in slot 4
0,0>AC ESR 00000000.00600001 IPREP FERR UPA_A_ERR
0,0>DC[0] 00
0,0>DC[1] 00
0,0>DC[2] 00
0,0>DC[3] 00
0,0>DC[4] 00
0,0>DC[5] 00
0,0>DC[6] 00
0,0>DC[7] 00
0,0>FHC CSR 00050030 LOC_FATAL SYNC BRD_LED_M BRD_LED_R
0,0>FHC RCSR 02000000 FATAL
0,0> Config policy change
0,0>
0,0>@(#) POST 3.9.28 2000/12/20 12:29
0,0>Copyright 2000 Sun Microsystems, Inc. All rights reserved.
0,0>
....
The problem is that the FT_ARBERR can be interpreted as being a failure and Board
0 would be replaced. This is not correct. The FT_ARBERR state is invalid. The root
cause of this error is an Etag Parity Error on Board 4, CPU 0 (CPU 8).
Chapter 4
Fatal Reset Diagnosis
Sun Proprietary/Confidential: Internal Use Only
4-3
As mentioned above, the FT_ARBERR can also be set to FT_SHERR. Either of these
error modes should be discounted as not being a failing FRU. There should be other
failure modes shown in the console output.
4.3
Diagnosis #1 - UPA_A_ERR
(Etag Parity Error)
Fatal Reset
0,0>
FATAL ERROR
0,0>
At time of error: System software was running.
0,0>
Diagnosis: Board 0, UPA PORT Device, AC
0,0>
Log Date: May 22 12:02:03 GMT 2000
0,0>
RESET INFO for CPU/Memory board in slot 0
0,0>
AC ESR 00000000.00600001 IPREP FERR UPA_A_ERR
0,0>
DC[0] 00
0,0>
DC[1] 00
0,0>
DC[2] 00
0,0>
DC[3] 00
0,0>
DC[4] 00
0,0>
DC[5] 00
0,0>
DC[6] 00
0,0>
DC[7] 00
0,0>
FHC CSR 00050200 LOC_FATAL SYNC NOT_BRD_PRES
0,0>
FHC RCSR 02000000 FATAL
Comments:
Fatal Reset Errors that have the IPREP and FERR error bits set in the Address
Controller Error Status Register indicate that an Ecache Tag parity Error may be the
cause of the Fatal Reset. In the typical error message shown above, the failure can be
diagnosed to be an Ecache Tag Parity Error on Board 0 CPU 0.
The Sun UltraSPARCSII Processor system CPUs will send a P_ERR P_REPLY to the
AC for the following two conditions;
4-4
■
Parity error on UPA address bus while AC is bus master & CPU is the slave.
■
E-Cache tag parity error. This is not reported as a trap like an E-Cache data error
because system coherence is lost for this condition and the system must be reset.
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
Improper torque may cause scenario (1) above but since the address bus is a bidirectional and the CPU is bus master, most of the time a mechanical connection
problem would probably also produce fatal reset errors with the UPA_PERR bit set,
i.e., the CPU is address bus master and the AC saw the parity error.
Absence of the UPA_PERR error bits (like the one shown above) indicates scenario
(2) or a CPU Etag parity error is the most likely cause.
prtdiag -v output:
The following is sample output from a prtdiag -v session after a UPA_A_ERR Fatal
Reset. This is prior to any hardware power cycling or replacement.
Analysis of most recent Fatal Hardware Watchdog:
======================================================
Log Date: Sat Nov 11 06:47:43 2000
Analysis for Board 0
-------------------AC: P_FERR error P_REPLY received from UPA Port
The error could be caused by:
CPU
Address Controller
AC: Illegal P_REPLY received from UPA Port
The error could be caused by:
CPU
Address Controller
Recommendation:
Note that UPA_A_ERR refers to CPU location 0 and UPA_B_ERR refers to CPU
location 1.
CPU 0 on CPU/Memory Board 0 would be the FRU implicated in causing this
failure and the CPU to replace.
Chapter 4
Fatal Reset Diagnosis
Sun Proprietary/Confidential: Internal Use Only
4-5
4.4
Diagnosis # 2 - UPA_A_ERR (Etag Parity
Error) Multiple Errors
Fatal Reset
7,0>FATAL ERROR
7,0> At time of error: System software was running.
7,0> Diagnosis: Board 9, UPA PORT Device, AC
7,0> Diagnosis: centerplane terminators
7,0>Log Date: Jan 19 13:54:45 GMT 2000
7,0>
7,0>RESET INFO for IO Type 4 board in slot 1
7,0> AC ESR 00002000.00000000 FTA_PERR
7,0> DC[0] 00
7,0> DC[1] 00
7,0> DC[2] 00
7,0> DC[3] 00
7,0> DC[4] 00
7,0> DC[5] 00
7,0> DC[6] 00
7,0> DC[7] 00
7,0> FHC CSR 00040000 LOC_FATAL
7,0> FHC RCSR 02000000 FATAL
7,0>RESET INFO for IO Type 4 board in slot 3
7,0> AC ESR 00002000.00000000 FTA_PERR
7,0> DC[0] 00
7,0> DC[1] 00
7,0> DC[2] 00
7,0> DC[3] 00
7,0> DC[4] 00
7,0> DC[5] 00
7,0> DC[6] 00
7,0> DC[7] 00
7,0> FHC CSR 00040000 LOC_FATAL
7,0> FHC RCSR 02000000 FATAL
Diagnosis # 2 - UPA_A_ERR (Etag Parity Error) Multiple Errors
(continued)
7,0>RESET INFO for CPU/Memory board in slot 7
7,0> AC ESR 00002000.00000000 FTA_PERR
4-6
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
7,0>
7,0>
7,0>
7,0>
7,0>
7,0>
7,0>
7,0>
DC[0]
DC[1]
DC[2]
DC[3]
DC[4]
DC[5]
DC[6]
DC[7]
00
00
00
00
00
00
00
00
7,0> FHC CSR 00050200 LOC_FATAL SYNC NOT_BRD_PRES
7,0> FHC RCSR 02000000 FATAL
7,0>RESET INFO for CPU/Memory board in slot 9
7,0> AC ESR 00000000.00600002 IPREP FERR UPA_B_ERR
7,0> DC[0] 00
7,0> DC[1] 00
7,0> DC[2] 00
7,0> DC[3] 00
7,0> DC[4] 00
7,0> DC[5] 00
7,0> DC[6] 00
7,0> DC[7] 00
7,0> FHC CSR 00050030 LOC_FATAL SYNC BRD_LED_M BRD_LED_R
7,0> FHC RCSR 02000000 FATAL
Comments:
The error on CPU/Memory Board 9 is either a CPU (location 1 / B port) Etag parity
error or an incoming system address parity error on the UPA. Engineering
experience says that it is most likely a CPU Etag (90%) as an ISAP would likely be
accompanied by the bit UPA_PERR.
The FTA_PERRs reported by the other slots are likely artifacts.
Recommendation:
See Fatal Reset #1 above (UPA_A_ERR) for further information. The implicated FRU
in this example is CPU 19 or CPU 1 on CPU/Memory Board 9.
Chapter 4
Fatal Reset Diagnosis
Sun Proprietary/Confidential: Internal Use Only
4-7
4.5
DT_PERR (DTAG Parity Error)
Fatal Reset
0,0>FATAL ERROR
0,0>
At time of error: System software was running.
0,0>
Diagnosis: Board 6, Dtag A (UPA Port 0), AC
0,0>Log Date: Dec 17 22:20:15 GMT 2000
0,0>
0,0>RESET INFO for CPU/Memory board in slot 6
0,0>
AC ESR 00000020.00000000 DT_PERRA
0,0>
DC[0] 00
0,0>
DC[1] 00
0,0>
DC[2] 00
0,0>
DC[3] 00
0,0>
DC[4] 00
0,0>
DC[5] 00
0,0>
DC[6] 00
0,0>
DC[7] 00
0,0>
FHC CSR 00050030 LOC_FATAL SYNC BRD_LED_M BRD_LED_R
0,0>
FHC RCSR 02000000 FATAL
0,0> Config policy change
Comments:
The DT_PERRA in slot 6 indicates a Duplicate Tag SRAM (DTAG) parity error.
These DTAG SRAMs reside on the CPU/Memory boards.
prtdiag -v output:
The following is sample output from a prtdiag -v session after a DT_PERRA or
DT_PERRB Fatal Reset. This is prior to any hardware power cycling or replacement.
Analysis of most recent Fatal Hardware Watchdog:
======================================================
Log Date: Sun Dec 17 14:20:15 2000
Analysis for Board 6
-------------------AC: UPA Port A Dtag Parity Error
The error could be caused by:
Data Tags for UPA Port A
Address Controller
4-8
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
Recommendation:
Note that DT_PERRA refers to DTAG SRAMs that refer to CPU location 0 and
DT_PERRB refers to CPU location 1. Again, these DTAG SRAMs reside on the
CPU/Memory Board, not on the CPU Module themselves.
In this example, the reporting CPU/Memory Board 6 should be replaced. The CPUs
and memory on this CPU/Memory Board are good.
Chapter 4
Fatal Reset Diagnosis
Sun Proprietary/Confidential: Internal Use Only
4-9
4.6
Diagnosis # 4 - DT_PERR (DTAG Parity
Error) Multiple Errors
Fatal Reset
0,0>FATAL ERROR
0,0> At time of error: System software was running.
0,0> Diagnosis: Board 0, centerplane pin, connector pin, AC
0,0> Diagnosis: Board 14, Dtag A (UPA Port 0), AC
0,0> Diagnosis: centerplane terminators
0,0>Log Date: Jun 22 13:22:00 GMT 1999
0,0>RESET INFO for CPU/Memory board in slot 0
0,0> AC ESR 00000400.00000000 FT_ARBERR
0,0> DC[0] 00
0,0> DC[1] 00
0,0> DC[2] 00
0,0> DC[3] 00
0,0> DC[4] 00
0,0> DC[5] 00
0,0> DC[6] 00
0,0> DC[7] 00
0,0> FHC CSR 00050200 LOC_FATAL SYNC NOT_BRD_PRES
0,0> FHC RCSR 02000000 FATAL
0,0>RESET INFO for IO Type 4 board in slot 1
0,0> AC ESR 00002000.00000000 FTA_PERR
0,0> DC[0] 00
0,0> DC[1] 00
0,0> DC[2] 00
0,0> DC[3] 00
0,0> DC[4] 00
0,0> DC[5] 00
0,0> DC[6] 00
0,0> DC[7] 00
0,0> FHC CSR 00040000 LOC_FATAL
0,0> FHC RCSR 02000000 FATAL
0,0>RESET INFO for IO Type 4 board in slot 3
0,0> AC ESR 00002000.00000000 FTA_PERR
4-10
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
0,0> DC[0] 00
0,0> DC[1] 00
0,0> DC[2] 00
0,0> DC[3] 00
0,0> DC[4] 00
0,0> DC[5] 00
0,0> DC[6] 00
0,0> DC[7] 00
0,0> FHC CSR 00040000 LOC_FATAL
0,0> FHC RCSR 02000000 FATAL
0,0>RESET INFO for CPU/Memory board in slot 14
0,0> AC ESR 00000020.00000000 DT_PERRA
0,0> DC[0] 00
0,0> DC[1] 00
0,0> DC[2] 00
0,0> DC[3] 00
0,0> DC[4] 00
0,0> DC[5] 00
0,0> DC[6] 00
0,0> DC[7] 00
0,0> FHC CSR 00050030 LOC_FATAL SYNC BRD_LED_M BRD_LED_R
0,0> FHC RCSR 02000000 FATA
Comments:
Although this Fatal Reset output diagnoses several possible components, the real
problem is a Dtag parity error on system board in slot 14.
The FT_ARBERR in slot 0 and the FTA_PERRs reported by the other slots are likely
fatal reset artifacts.
Recommendation:
In this example, the reporting CPU/Memory Board 14 should be replaced. The
CPUs and memory on this CPU/Memory Board are good.
Chapter 4
Fatal Reset Diagnosis
Sun Proprietary/Confidential: Internal Use Only
4-11
4.7
Diagnosis #5 - FTA_PERR
0,0>FATAL ERROR
0,0> At time of error: POST was testing Board 0 Centerplane
0,0> Diagnosis: Board 0, backplane pins, board connector pins, AC
0,0>Log Date: Jun 19 7:10:00 GMT 1999
0,0>
0,0>RESET INFO for CPU/Memory board in slot 0
0,0> AC ESR 00002000.00000000 FTA_PERR
0,0> DC[0] 00
0,0> DC[1] 00
0,0> DC[2] 00
0,0> DC[3] 00
0,0> DC[4] 00
0,0> DC[5] 00
0,0> DC[6] 00
0,0> DC[7] 00
0,0> FHC CSR 00050200 LOC_FATAL SYNC NOT_BRD_PRES
0,0> FHC RCSR 02000000 FATAL
0,1>
Comments:
POST was testing "board 0 centerplane". However, there are cases where this type of
Fatal Reset is not really caused by CPU/Memory Board 0, or the Centerplane. There
may be other components in the system that is causing this Fatal Reset to occur.
Recommendation:
Although for this example, the reporting CPU/Memory board 0 can be replaced
(existing CPUs and Memory on that board should be good), this error may return. If
this is the case, it is recommended to try and test the system rigorously to determine
if any components do fail. Replacement of the centerplane may not always correct
this problem. This also points to some other FRU in the system not being called out
in the Fatal Reset data.
4-12
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
4.8
Diagnosis #6 - FTC_PERR
# Fatal Reset
0,0>FATAL ERROR
0,0> At time of error: System software was running.
0,0> Diagnosis: Board 7, backplane pins, board connector pins, AC
0,0>Log Date: Aug 31 7:16:02 GMT 1997
0,0>
0,0>RESET INFO for IO Type 3 board in slot 7
0,0> AC ESR 00001000.00000000 FTC_PERR
0,0> DC[0] 00
0,0> DC[1] 00
0,0> DC[2] 00
0,0> DC[3] 00
0,0> DC[4] 00
0,0> DC[5] 00
0,0> DC[6] 00
0,0> DC[7] 00
0,0> FHC CSR 00040000 LOC_FATAL
0,0> FHC RCSR 02000000 FATAL
Comments:
Basically the same analysis as Fatal Reset case #5, except that we know system
software was running and there are at least two boards in the system.
Recommendation:
In this example, the reporting I/O Board in slot 7 is the FRU to replace. Existing
SBus controllers should be good.
Chapter 4
Fatal Reset Diagnosis
Sun Proprietary/Confidential: Internal Use Only
4-13
4.9
Diagnosis #7 - MTIMEOUT
# Fatal Reset
0,0>FATAL ERROR
0,0> At time of error: System software was running.
0,0> Diagnosis: Board 7, any system board MTIMEOUT (target of
operation)
0,0>Log Date: Feb 12 6:35:31 GMT 199e
0,0>
0,0>RESET INFO for CPU/Memory board in slot 7
0,0> AC ESR 00000000.01000001 MTIMEOUT UPA_A_ERR
0,0> DC[0] 00
0,0> DC[1] 00
0,0> DC[2] 00
0,0> DC[3] 00
0,0> DC[4] 00
0,0> DC[5] 00
0,0> DC[6] 00
0,0> DC[7] 00
0,0> FHC CSR 00040000 LOC_FATAL
0,0> FHC RCSR 02000000 FATAL
Comments:
This error means that the CPU (location 0 / Port A) on CPU/Memory Board in slot
7, issued a request via the AC to a board which it knows is present, yet no response
was received for 16M clock cycles (193ms @ 83Mhz). Note that this error cannot be
caused by reading a non-existent location (that would cause a panic if it was issued
by a CPU).
An example of this type of problem is where Psycho+ appears to not complete a
register read to a PCI device in the allotted time.
In general this problem could be the fault of the reporting board, or the target of the
board’s operation. Unfortunately, we don’t know the target. However, the target is
many times an I/O board. As such, if there is only one or a few I/O boards in the
system, you can narrow down your search.
4-14
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
NOTE: There are reported cases where a similar looking Fatal Reset will occur and
the cause had been found to be a CPU Module. This has been seen during POST and
the error signature appears as:
0,0>
TESTCASE 00000000.01000002 AC ESR
prtdiag -v output:
The following is sample output from a prtdiag -v session after an MTIMEOUT
Fatal Reset. This is prior to any hardware power cycling or replacement.
Analysis of most recent Fatal Hardware Watchdog:
======================================================
Log Date: Mon May 8 14:33:19 2000
Analysis for Board 7
-------------------AC: Timeout on a UPA Master Port
The error could be caused by:
Undetermined Address Controller in system
Undetermined Board in system
Recommendation:
Unfortunately, the MTIMEOUT Fatal Reset gives little information regarding the
true root cause of the error. Given that, it is recommended that the following actions
be taken.
1. Review the system history to determine if any recent components have been
recently replaced. Determine if any of these components could be likely suspects
in the error.
2. If possible, shutdown the system and key-reset (power cycle) the system so that
all components are active and run extended POST to determine if the failure is
solid (not intermittent).
3. If the the failure is solid, use standard troubleshooting techniques to isolate the
failing component.
4. If the failure is intermittent (meaning extended POST did not fail), try to run
SunVTS. If SunVTS does not indicate a failing component, monitor the system for
future failures. Especially on the CPU/Memory Board that was called out by the
last MTIMEOUT error. This monitoring has proven beneficial in finding the
correct FRU causing this type of Fatal Reset.
Chapter 4
Fatal Reset Diagnosis
Sun Proprietary/Confidential: Internal Use Only
4-15
4.10
Diagnosis #8 - FTUPAOV
Fatal Reset
0,0>FATAL ERROR
0,0> At time of error: System software was running.
0,0> Diagnosis: Board 3, centerplane, board connector, AC, any
other AC
0,0>Log Date: Jul 29 20:14:41 GMT 1998
0,0>
0,0>RESET INFO for CPU/Memory board in slot 3
0,0> AC ESR 00000002.00000000 FTUPAOV
0,0> DC[0] 00
0,0> DC[1] 00
0,0> DC[2] 00
0,0> DC[3] 00
0,0> DC[4] 00
0,0> DC[5] 00
0,0> DC[6] 00
0,0> DC[7] 00
0,0> FHC CSR 00050020 LOC_FATAL SYNC BRD_LED_M
0,0> FHC RCSR 02000000 FATAL
Comments:
The Firetruck to UPA queue overflowed on CPU/Memory Board in slot 3. Most
likely this is caused by a broken AC. Since CPU/Memory Board 3 is the only board
reporting a problem, it is likely the reporting board. Otherwise, it could be some
other board.
This is not likely a centerplane problem because you would usually see some sort of
parity error in addition to the queue overflow.
Note – For some firetruck parity errors, all CPU/Memory and I/O boards will
report errors. There may be one board that does not report an error. Based on this
error mode, the board that did not report the error would be the suspect in sending
of incorrect/bad data. Replacement of the non-reporting board would be a good
first action.
4-16
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
Recommendation:
For this example, replace the reporting CPU/Memory Board in slot 3. Existing
CPUs and Memory should be good.
4.11
Diagnosis #9 - Cacheable Write Error
Fatal Reset
0,0>FATAL ERROR
0,0> At time of error: System software was running.
0,0> Diagnosis: Board 2, software, any system board
0,0>Log Date: Mar 19 3:34:39 GMT 1998
0,0>
0,0>RESET INFO for CPU/Memory board in slot 2
0,0> AC ESR 00000000.00000041 ICWS UPA_A_ERR
0,0> DC[0] 00
0,0> DC[1] 00
0,0> DC[2] 00
0,0> DC[3] 00
0,0> DC[4] 00
0,0> DC[5] 00
0,0> DC[6] 00
0,0> DC[7] 00
0,0> FHC CSR 00050830 LOC_FATAL SYNC EPDB_OFF BRD_LED_M BRD_LED_R
0,0> FHC RCSR 02000000 FATAL
Comments:
This error is caused by a cacheable write being sent by the CPU (first CPU / Port A)
on CPU/Memory Board 2 to an unmapped or non-accepting destination. Cacheable
writes should only go to memory boards.
While this could be caused by software (e.g. system software). An example of a
system software problem causing this error is during a Dynamic Reconfiguration
operation where memory is being un-configured.
Thus the most likely problem in this example is either the reporting CPU/Memory
Board 2 (or CPU), or the destination memory board. Unfortunately, the destination
memory board is unknown.
Chapter 4
Fatal Reset Diagnosis
Sun Proprietary/Confidential: Internal Use Only
4-17
prtdiag -v output:
The following is sample output from a prtdiag -v session after an ICWS Fatal
Reset. This is prior to any hardware power cycling or replacement.
Analysis of most recent Fatal Hardware Watchdog:
======================================================
Log Date: Fri Mar 17 01:08:48 2000
Analysis for Board 2
-------------------AC: UPA Cacheable write to unmapped destination
The error could be caused by:
This Board
Recommendation:
Determine if any Dynamic Reconfiguration events were in process at the time of the
failure. If not, for this example, the reporting CPU/Memory Board in slot 2 should
be replaced. If available, the CPU Module in location 0 on CPU/Memory Board 2
should be replaced as well.
4-18
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
4.12
Diagnosis #10 - Non-Cacheable Write
Fatal Reset
0,0>FATAL ERROR
0,0> At time of error: System software was running
0,0> Diagnosis: Board 0, software, any system board
0,0>Log Date: Mar 20 2:05:33 GMT 1996
0,0>
0,0>RESET INFO for CPU/Memory board in slot 0
0,0> AC ESR 00000000.00000021 INCWS UPA_A_ERR
0,0> DC[0] 00
0,0> DC[1] 00
0,0> DC[2] 00
0,0> DC[3] 00
0,0> DC[4] 00
0,0> DC[5] 00
0,0> DC[6] 00
0,0> DC[7] 00
0,0> FHC CSR 00040000 LOC_FATAL
0,0> FHC RCSR 02000000 FATAL
Comments:
Same as previous, only a non-cacheable write. In this case the destination is likely to
be an I/O board. Source (Master) was the CPU (location 0 / port A) on
CPU/Memory Board 0.
prtdiag -v output:
Analysis of most recent Fatal Hardware Watchdog:
======================================================
Log Date: Sat May 13 03:07:59 2000
Analysis for Board 0
-------------------AC: UPA Non-cacheable write to unmapped destination
The error could be caused by:
This Board
Chapter 4
Fatal Reset Diagnosis
Sun Proprietary/Confidential: Internal Use Only
4-19
Recommendation:
Determine if any Dynamic Reconfiguration events were in process at the time of the
failure. If not, for this example, the reporting CPU/Memory Board in slot 0 should
be replaced. If available the CPU Module in location 0 on CPU/Memory Board 0
should be replaced as well.
4.13
Diagnosis #11 - Interrupt Error
0,0> At time of error: System software was running.
0,0> Diagnosis: Board 4, software, any system board
0,0>Log Date: Dec 4 19:07:57 GMT 1997
0,0>
0,0>RESET INFO for CPU/Memory board in slot 4
0,0> AC ESR 00000000.00000011 IIS UPA_A_ERR
0,0> DC[0] 00
0,0> DC[1] 00
0,0> DC[2] 00
0,0> DC[3] 00
0,0> DC[4] 00
0,0> DC[5] 00
0,0> DC[6] 00
0,0> DC[7] 00
0,0> FHC CSR 00040030 LOC_FATAL BRD_LED_M BRD_LED_R
0,0> FHC RCSR 02000000 FATAL
Comments:
Similar to the previous Fatal Reset, only an interrupt was sent instead of a write.
The source was the CPU (location 0 / port A) on CPU/Memory Board in slot 4. The
target of any interrupt would have been another CPU board (not I/O).
4-20
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
prtdiag-v output:
Analysis of most recent Fatal Hardware Watchdog:
======================================================
Log Date: Wed Dec 1 10:10:00 1999
Analysis for Board 4
AC: UPA Interrupt to unmapped destination
The error could be caused by:
This Board
--------------------
Recommendation:
Determine if any Dynamic Reconfiguration events were in process at the time of the
failure. If not, for this example, the reporting CPU/Memory Board in Slot 4 should
be replaced. If available the CPU Module in location 0 on CPU/Memory Board 4
should be replaced as well.
Chapter 4
Fatal Reset Diagnosis
Sun Proprietary/Confidential: Internal Use Only
4-21
4-22
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
CHAPTER
5
Enhanced Solaris Error Messaging
5.1
Improved Error Messages
With the introduction of the kernel level scrubber patches (Kernel Update or KU
patches) for Solaris 2.6 and above, the CPU, Ecache, and memory error messages
have been improved to be more accurate and complete. Text descriptions have been
rewritten to emphasize the important parameters associated with each event. Also,
the logic for reporting hardware errors has changed to ensure that error events are
reported accurately, completely, and in the order they occurred. These new error
messages will make it easier to determine the CPU that has encountered an error.
There are related patches to Sun Management Center so that it will recognize the
improved error messages; without them, the management console will under-report
the occurrence of corrected main memory errors. Refer to the Solaris KU patch
readme files for further information.
Note – Solaris 2.5.1 and prior releases do not support improved error messaging..
5-1
Sun Proprietary/Confidential: Internal Use Only
5.2
Errors and Events
SunSPARC processors can detect errors that are reported in the following types of
events (as detailed in the SunSPARC-I/II User’s Manual, 802-7220-02):
5-2
■
ETP – A parity error was detected by the CPU when reading from the Ecache Tag
SRAM. This is a fatal error because system coherency has been lost. The system
will reset (POR) and Starfire domains will arbstop (UPA Fatal error). No Solaris
error message will be generated.
■
EDP – A parity error was detected by the CPU when reading from the Ecache
Data SRAM on a cache hit.
■
LDP – A parity error was detected by the CPU while reading main memory
through its Ultra Data Buffer (UDB) chip on an Ecache miss. Note that the Ecache
itself is not involved. This can occur when the CPU is reading non-cacheable data
(for example, a frame buffer or I/O device), or when filling a line of cache from
main memory.
■
WP – A parity error was detected by one of the UDB chips while data was being
written back from the Ecache into main memory. The UDB chips convert the data
with bad parity into data with bad ECC, so that a subsequent access to the same
physical address will result in a UE. (See UE below.) (The conversion of a parity
error to a latent UE does not occur on either UltraSPARC-IIi or -IIe, which is one
of the reasons why improved error handling is not available on those processors.)
■
CP – A parity error was detected during a copyout transaction; that is, a data
transfer from one CPU’s Ecache to another CPU. This error is detected by the
UDB chips of the providing CPU, resulting in the CP event. The providing CPU’s
UDB chips convert the data with bad parity to data with bad ECC, so that the
UDBs of the receiving CPU will report a UE event. (See UE below.)
■
UE – An un-correctable memory error has occurred. This event refers to an error
in the main system memory, reported by the system databus on a read access.
The underlying source of this error could be main memory, another CPU module
(see CP above), or another UPA device (for example, the I/O controller). The
UDB chips detect this error.
■
CE – A correctable error was detected when reading from main memory, or when
reading from another CPU’s UDB chips. The data read has been corrected and
valid data is given to the CPU and the CPU’s Ecache. This error is detected by
the UDB chips.
■
BERR – A bus error has occurred during an attempt to read from a memory
address. Either there is no device at that address, or the device at that address
has returned a bus error. Therefore, bus errors are caused by a programming
error or by a corrupted or defective device.
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
■
5.3
TO – A bus timeout was encountered during an attempt to read from a memory
address. Too much time has elapsed waiting for a device at that address to
respond.
Details on Improved Error Handling
Any of the above mentioned errors can occur in kernel instruction space, kernel data
space, user instruction space, user data space, or when the kernel reads or writes
user data (as in copyin). Depending on these different states, the operating system
will react differently so as to maximize system availability.
On EDP, LDP, CP, UE, BERR, and TO events, the system will panic if the affected
data is in kernel space or if the error occurs while the CPU is at a trap level greater
than zero. Otherwise, the process that caused the error will be killed immediately
(sent SIGKILL) and the system will be rebooted (as if a privileged user had entered
"init 6").
Note – An active SC2.X cluster node will panic with a "Failfast timeout" (usually
with "Device closed while Armed") when rebooted. It is therefore useful to check
the system messages for EDP, LDP, CP, UE, BERR, and TO events while encountering
"Failfast timeout" panics.
On WP events, an error is reported, and the memory scrubber is notified to scan all
of system memory for the latent UE the hardware has written to memory (see below
for the behavior of the memory scrubber on encountering UE events). If some CPU
later attempts to read this location (other than on behalf of the memory scrubber), a
UE event will occur. Hence, when a UE event is encountered, it is recommended
that the log be checked for an earlier WP event that may have in fact caused the UE
event.
If the memory scrubber detects a UE event the system will neither panic nor reboot
but trigger a recovery mechanism instead. If the page containing the corrupted data
is not in use, it will be retired and the error will be cleared. If it is in use, it will be
marked for retirement and clearing if and when it is no longer in use.
Note – Due to hardware limitations there is no improved error handling for
UltraSPARC-IIi and UltraSPARC-IIe based systems.
Chapter 5
Enhanced Solaris Error Messaging
Sun Proprietary/Confidential: Internal Use Only
5-3
5.4
Details on Improved Error Messages
For each error that is detected, the kernel generates an individual report. This is a
major change; previously, some errors would hide other errors, and some errors
were combined into a single message. The report typically consists of several error
messages. Each message [3] contains an AFT ("Asynchronous Fault Trap") tag that
eases filtering, and an errID code that associates all of the messages emitted for the
same event. The errID is a 64-bit code that corresponds to a specific set of error bits
in the Asynchronous Fault Status Register (AFSR) at a specific instance in time; the
value has no intrinsic meaning.
Each message may be longer than one physical line; long messages are folded using
embedded newlines. Each folded line begins with four space characters.
Note – Because of the introduction of improved error messages, any tool using the
affected error messages may have to be modified. Neither the format nor the
content of kernel error messages are committed interfaces, and both may change
without notice. Users (both internal and external) who rely on the exact format
and/or content do so at their own risk.
5.5
Error Message Categories
The error messages can be grouped into four categories.
Category 1: Messages that identify the type and source of an error:
WARNING: [AFT1] EDP event on CPU1 Instruction access at TL=0, errID
0x0000ad88.6cd9989f
AFSR 0x00000000.80408000<PRIV,EDP> AFAR 0x00000000.0f0c8080
AFSR.PSYND 0x8000 (Score 95) AFSR.ETS 0x00 FAULT_PC 0x780b481c
UDBH 0x0000 UDBH.ESYND 0x00 UDBL 0x0000 UDBL.ESYND 0x00
Either the [AFT0] tag (for correctable errors) or the [AFT1] tag (for un-correctable
errors) is present in the message. An "errID" field appears at the end of the first line
of the message. Messages from this category are displayed on the console and
collected in the log file.
5-4
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
Note – This is the default behavior. The /etc/system setting report_ce_console is
no longer referenced and should therefore be removed.
To aid diagnosis of an Ecache-related error, especially if multiple components are
involved, a heuristic algorithm has been included that automates analysis of the
P_SYND bytes. Every component reporting a failure has its AFSR decoded and a
score ranging from 5 to 95 is assigned ("Score 95" in the above example).
The Score indicates the likelihood that this component was the original source of the
bad parity. The higher the value, the higher the likelihood that this component was
the original source.
Category 2: Messages that supply a cache line or memory dump:
[AFT2] errID 0x0000ad88.6cd9989f PA 0x00000000.0f0c8080 E$tag
0x00000000.0bc001e1 E$State: Modified E$parity 0x05
[AFT2] E$Data (0x00): 0xffffffff.beefface *Bad* PSYND=0x8000
[AFT2] E$Data (0x08): 0x00000000.00000000
[AFT2] E$Data (0x10): 0x6d656d6d.6f727920
[AFT2] E$Data (0x18): 0x6572726f.7220696e
[AFT2] E$Data (0x20): 0x6a656374.6f720000
[AFT2] E$Data (0x28): 0x6d656d74.65737420
[AFT2] E$Data (0x30): 0x6d757465.780059f8
[AFT2] E$Data (0x38): 0x00000300.00c11000
[AFT2] Event PA displayed in AFAR was derived from E$Tag
Messages from this category are targeted for Sun Microsystems support staff to be
used in backline diagnosis and for statistics.
The [AFT2] tag is always present in these messages. The "errID" field appears at the
beginning of the first line of the message. Messages from this category are by
default only collected in the log file.
Category 3: Messages from the kernel error recovery code:
[AFT3] errID 0x00000058.0d0dc830 Above Error detected by protected
Kernel code that will try to clear error from system
Messages from this category supply analysis information from the kernel error
recovery code, thereby indicating the actions the kernel took to contain the error.
The [AFT3] tag is always present in these messages. An "errID" field appears at the
beginning of the first line of the message.
Chapter 5
Enhanced Solaris Error Messaging
Sun Proprietary/Confidential: Internal Use Only
5-5
Messages from this category are by default only collected in the log file.
Category 4: Messages that indicate the disposition of an error:
panic[CPU1]/thread=30000670800: [AFT1] errID
0x00000392.89cbfefc EDP Error(s)
See previous message(s) for details
Messages from this category state the final handling (like panic or reboot) of a
previously encountered error.
Either the [AFT0] tag (for correctable errors) or the [AFT1] tag (for un-correctable
errors) is present in the message. The "errID" field appears at the beginning of the
first line of the message. Messages from this category are displayed on the console
and collected in the log file.
5.6
Error Messages Examples
The following compares previous messages with the new, improved error messages.
Note that this is not an exhaustive list, but a sampling of possible messages for each
event type. This also just shows what appears on the console; the log-only messages
are not shown.
Error messages shown below do not necessarily appear exactly as they appear on the
console. Due to message lengths, etc. the message lines shown below may wrap
around, however the entire content of the message lines are shown below.
5-6
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
5.7
EDP Event - Ecache Data Parity Event
Previous Style Message - Kernel Mode (Panic):
panic[CPU1]/thread=3000225bcc0: CPU1 Ecache SRAM Data Parity
Error:
AFSR 0x00000000.80408000 AFAR 0x00000000.0bd83bd0
Improved Message - Kernel Mode (Panic):
WARNING: [AFT1] EDP event on CPU1 Data access at TL=0, errID
0x00000093.6323e6f8AFSR 0x00000000.80408000<PRIV,EDP> AFAR
0x00000000.06901980
AFSR.PSYND 0x8000 (Score 95) AFSR.ETS 0x00 Fault_PC 0x78128a84
UDBH 0x0000 UDBH.ESYND 0x00 UDBL 0x0000 UDBL.ESYND 0x00
panic[cpu1]/thread=30000ae5000: [AFT1] errID 0x00000093.6323e6f8
EDP Error(s)
See previous message(s) for details
Previous Style Message - User Mode (Panic):
panic[CPU3]/thread=30001f4fa00: CPU3 Ecache SRAM Data Parity
Error:
AFSR 0x00000000.00400080 AFAR 0x00000000.01820000
Chapter 5
Enhanced Solaris Error Messaging
Sun Proprietary/Confidential: Internal Use Only
5-7
Improved Message - User Mode (Reboot):
Aug 16 16:47:20 thishost SUNW,UltraSPARC-II: WARNING: [AFT1]
EDP event on CPU3 Data access at TL=0, errID 0x00000057.d35eff81
Aug 16 16:47:20 thishost
AFSR 0x00000000.00400080<EDP> AFAR
0x00000000.05e24418 AFSR.PSYND 0x0080 (Score 95) AFSR.ETS 0x00
Fault_PC 0x11ce8 UDBH 0x0000 UDBH.ESYND 0x00 UDBL 0x0000
UDBL.ESYND 0x00
Aug 16 16:47:20 thishost unix: NOTICE: Scheduling clearing of
error on page 0x00000000.05e24000
Aug 16 16:47:20 thishost unix: WARNING: [AFT1] initiating reboot
due to above error in pid 309 (mtst)
Aug 16 16:47:23 thishost unix: NOTICE: Previously reported
error on page 0x00000000.05e24000 cleared
Improved Message - User Data (Reboot):
INIT: New run level: 6
The system is coming down. Please wait.
System services are now being stopped.
Print services stopped.
Aug 16 16:47:27 thishost syslogd: going down on signal 15
The system is down.
syncing file systems... done
rebooting...
Resetting ...
5.8
Trap Level 1 Panic
Previous Solaris Message - Kernel Data at TL=1 (Panic):
panic[CPU3]/thread=30001cfabe0: Async data error at tl1:
0x00000000.0ab8f760 AFSR 0x00000000.80400080
5-8
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
AFAR
Improved Message - Kernel Data at TL=1 (Panic):
WARNING: [AFT1] EDP event on CPU3 Data access at TL>0, errID
0x00000111.53a7b8dd AFSR 0x00000000.80408000<PRIV,EDP> AFAR
0x00000000.01f47dc0 AFSR.PSYND 0x8000 (Score 95) AFSR.ETS 0x00
Fault_PC 0x1002fe20 UDBH 0x0000 UDBH.ESYND 0x00 UDBL 0x0000
UDBL.ESYND 0x00
panic[cpu3]/thread=30000a4e040: [AFT1] errID 0x00000111.53a7b8dd
EDP Error(s)
See previous message(s) for details
Previous Solaris Message - Kernel Instruction at TL=1 (Panic):
panic[CPU3]/thread=3000226a140: Async instruction error at tl1:
AFAR 0x00000000.0dd55f70 AFSR 0x00000000.80408000
Improved Message - Kernel Instruction at TL=1 (Panic):
WARNING: [AFT1] EDP event on CPU3 Instruction access at TL>0, errID
0x00000043.24bfd349 AFSR 0x00000000.80400800<PRIV,EDP> AFAR
0x00000000.0605c790 AFSR.PSYND 0x0800 (Score 95) AFSR.ETS 0x00
Fault_PC 0x1002fe20 UDBH 0x0000 UDBH.ESYND 0x00 UDBL 0x0000
UDBL.ESYND 0x00
panic[cpu3]/thread=30000ad05c0: [AFT1] errID 0x00000043.24bfd349
EDP Error(s)
See previous message(s) for details.
Chapter 5
Enhanced Solaris Error Messaging
Sun Proprietary/Confidential: Internal Use Only
5-9
5.9
WP Event - Writeback Data Parity Error
Previous Solaris Message (Panic):
panic[CPU1]/thread=30001b26640: CPU1 Ecache Writeback Data Parity
Error:
AFSR 0x00000000.00800080 AFAR 0x00000000.0d5010f0
Improved Message (Panic Deferred):
Aug 16 16:50:56 thishost SUNW,UltraSPARC-II: WARNING: [AFT1] WP
event on CPU1, errID 0x0000002b.3c7cd6d9
Aug 16 16:50:56 thishost
0x000001c8.01802800
AFSR 0x00000000.00800080<WP> AFAR
Aug 16 16:50:56 thishost
AFSR.PSYND 0x0080 (Score 95) AFSR.ETS
0x00 Fault_PC 0x11d7c UDBH 0x0000 UDBH.ESYND 0x00 UDBL 0x0000
UDBL.ESYND 0x00
Aug 16 16:50:56 thishost SUNW,UltraSPARC-II: WARNING: [AFT1]
Uncorrectable Memory Error on CPU3 Data access at TL=0, errID
0x0000002b.45daae92
Aug 16 16:50:56 thishost
0x00000000.03824418
AFSR 0x00000000.80200000<PRIV,UE> AFAR
Aug 16 16:50:56 thishost AFSR.PSYND 0x0000(Score 05) AFSR.ETS 0x00
Fault_PC 0x10023414 UDBH 0x0000 UDBH.ESYND 0x00 UDBL 0x0203<UE>
UDBL.ESYND 0x03 UDBL Syndrome 0x3 Memory Module 190x
Aug 16 16:50:56 thishost SUNW,UltraSPARC-II: WARNING: [AFT1] errID
0x0000002b.45daae92 Syndrome 0x3 indicates that this may not be a
memory module problem
Aug 16 16:50:56 thishost unix: NOTICE: Scheduling clearing of error
on page 0x00000000.03824000
Aug 16 16:50:58 thishost unix: NOTICE: Previously reported error
on page 0x00000000.03824000 cleared
5-10
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
Note – The last message (reporting clearing of the error) may appear much later, or
may never appear, as the page may never drop out of use. Also, the message
reporting scheduling of clearing may occur more than once, as the memory scrubber
may encounter the particular UE more than once before it can be cleared.
5.10
CP Event - Copyout Data Parity Error
Previous Solaris Message (Panic):
panic[CPU3]/thread=2a100105d40: CPU3 UE Error: Ecache Copyout on
CPU1:
AFSR 0x00000000.01000080 AFAR 0x00000000.06c53090
Improved Message Kernel Mode (Panic):
WARNING: [AFT1] Uncorrectable Memory Error on CPU3 Data access at
TL=0, errID 0x0000003a.30aafcba AFSR 0x00000000.80200000<PRIV,UE>
AFAR 0x00000000.00347dc0 AFSR.PSYND 0x0000 (Score 05) AFSR.ETS
0x00 Fault_PC 0x78067b54 UDBH 0x0203<UE> UDBH.ESYND 0x03 UDBL
0x0000 UDBL.ESYND 0x00 UDBH Syndrome 0x3 Memory Module 190x
WARNING: [AFT1] errID 0x0000003a.30aafcba Syndrome 0x3 indicates
that this may not be a memory module problem
WARNING: [AFT1] CP event on CPU1 (caused Data access error on
CPU3), errID 0x0000003a.30aafcba AFSR 0x00000000.01008000<CP> AFAR
0x00000000.00347dc0 AFSR.PSYND 0x8000(Score 95) AFSR.ETS 0x00 UDBH
0x0000 UDBH.ESYND 0x00 UDBL 0x0000 UDBL.ESYND 0x00
panic[cpu3]/thread=2a100157d40: [AFT1] errID 0x0000003a.30aafcba
UE Error(s)
See previous message(s) for details.
Chapter 5
Enhanced Solaris Error Messaging
Sun Proprietary/Confidential: Internal Use Only
5-11
Improved Message User Mode (Reboot):
Aug 16 17:06:44 thishost SUNW,UltraSPARC-II: WARNING: [AFT1]
Uncorrectable Memory Error on CPU3 Data access at TL=0, errID
0x0000002b.963a3d3c
Aug 16 17:06:44 thishost
AFSR 0x00000000.00200000<UE> AFAR
0x00000000.00224418 AFSR.PSYND 0x0000 (Score 05) AFSR.ETS 0x00
Fault_PC 0x12380 UDBH 0x0000 UDBH.ESYND 0x00 UDBL 0x0203<UE>
UDBL.ESYND 0x03
Aug 16 17:06:44 thishost
UDBL Syndrome 0x3 Memory Module 190x
Aug 16 17:06:44 thishost SUNW,UltraSPARC-II: WARNING: [AFT1] errID
0x0000002b.963a3d3c Syndrome 0x3 indicates that this may not be a
memory module problem
Aug 16 17:06:44 thishost SUNW,UltraSPARC-II: WARNING: [AFT1] CP
event on CPU1 (caused Data access error on CPU3), errID
0x0000002b.963a3d3c
AFSR
0x00000000.01000080<CP> AFAR 0x00000000.00224418 AFSR.PSYND
0x0080 (Score 95) AFSR.ETS 0x00 UDBH 0x0000 UDBH.ESYND 0x00 UDBL
0x0000 UDBL.ESYND 0x00
Aug 16 17:06:44 thishost unix: NOTICE: Scheduling clearing of error
on page 0x00000000.00224000 WARNING: [AFT1] initiating reboot due
to above error in pid 304
Aug 16 17:06:46 thishost unix: NOTICE: Previously reported error
on page 0x00000000.00224000 cleared
INIT: New run level: 6
The system is coming down. Please wait.
CP Event - Copyout Data Parity Error (continued)
Print services stopped.
syslogd: going down on signal 15
The system is down.
syncing file systems... done
rebooting...
Resetting ...
5-12
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
Note – Due to a coding error, early versions of some of the patches produce the
string "CP Error" instead of "CP event"; programs that parse the messages must be
prepared to deal with both
5.11
UE Event - Uncorrectable Memory Error
Previous Solaris Message - CPU Reference to Memory:
panic[CPU1]/thread=2a1000R7dd40: UE Error: AFSR
0x00000000.80200000 AFAR 0x00000000.089cd740 Id 0 Inst 0 MemMod
U0501 U0401
Improved Message - CPU Reference to Memory Kernel Mode (Panic):
WARNING: [AFT1] Uncorrectable Memory Error on CPU1 Instruction
access at TL=0, errID 0x0000004f.818d9280 AFSR
0x00000000.80200000<PRIV,UE> AFAR 0x00000000.0685c7a0 AFSR.PSYND
0x0000 (Score 05) AFSR.ETS 0x00 Fault_PC 0x7815c7a0 UDBH
0x0203<UE> UDBH.ESYND 0x03 UDBL 0x0000 UDBL.ESYND 0x00 UDBH
Syndrome 0x3 Memory Module 190x
WARNING: [AFT1] errID 0x0000004f.818d9280 Syndrome 0x3 indicates
that this may not be a memory module problem
panic[cpu1]/thread=30000ad6320: [AFT1] errID 0x0000004f.818d9280
UE Error(s)
See previous message(s) for details.
Chapter 5
Enhanced Solaris Error Messaging
Sun Proprietary/Confidential: Internal Use Only
5-13
Improved Message - CPU Reference to Memory User Mode (Reboot):
Aug 16 17:03:04 thishost SUNW,UltraSPARC-II: WARNING: [AFT1]
Uncorrectable Memory Error on CPU1 Instruction access at TL=0,
errID 0x00000032.593d8229
Aug 16 17:03:04 thishost
AFSR 0x00000000.00200000<UE> AFAR
0x00000000.04921bf0 AFSR.PSYND 0x0000 (Score 05) AFSR.ETS 0x00
Fault_PC 0x11bf0
Aug 16 17:03:04 thishost
UDBH 0x0203<UE> UDBH.ESYND 0x03 UDBL
0x0000 UDBL.ESYND 0x00 UDBH Syndrome 0x3 Memory Module 190x
Aug 16 17:03:04 thishost SUNW,UltraSPARC-II: WARNING: [AFT1] errID
0x00000032.593d8229 Syndrome 0x3 indicates that this may not be a
memory module problem
Aug 16 17:03:04 thishost unix: NOTICE: Scheduling clearing of error
on page 0x00000000.04920000
Aug 16 17:03:07 thishost unix: NOTICE: Previously reported error
on page 0x00000000.04920000 cleared
Aug 16 17:03:07 thishost unix: WARNING: [AFT1] initiating reboot
due to above error in pid 304 (mtst)
INIT: New run level: 6
The system is coming down.
Please wait.
System services are now being stopped.
Print services stopped.
Aug 16 17:03:13 thishost syslogd: going down on signal 15
The system is down.
syncing file systems... done
rebooting...
Resetting ...
5-14
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
Previous Solaris Message - SBus I/O Reference to Memory:
panic[CPU1]/thread=2a10007dd40: SBus0 UE Primary Error DMA read:
AFSR 0x40001be0.00000000 AFAR 0x00000000.02818000 MemMod U0501
U0401 Id 31
Improved Message - SBus I/O Reference to Memory:
WARNING: SBus0 UE Primary Error DMA read: AFSR 0x40001be0.00000000
AFAR 0x00000000.0d25c000 MemMod U0501 U0401 Id 31
panic[cpu0]/thread=2a10007dd40: Fatal Sbus0 UE Error
5.12
BERR Event - Bus Error
Previous Solaris Message:
panic[CPU1]/thread=30000d2c300: CPU1 Privileged Bus Error:
0x00000000.84000000 AFAR 0x00000000.03422000
AFSR
Improved Message - Kernel Mode (Panic):
WARNING: [AFT1] Bus Error on System Bus in privileged mode from
CPU1 Data access at TL=0, errID 0x0000002c.52b3d2c8 AFSR
0x00000000.84000000<PRIV,BERR> AFAR 0x00000000.05224410
AFSR.PSYND 0x0000 (Score 05) AFSR.ETS 0x00 Fault_PC 0x780671a4
UDBH 0x0000 UDBH.ESYND 0x00 UDBL 0x0000 UDBL.ESYND 0x00
panic[cpu1]/thread=30000b06080: [AFT1] errID 0x0000002c.52b3d2c8
BERR Error(s)
See previous message(s) for details.
Chapter 5
Enhanced Solaris Error Messaging
Sun Proprietary/Confidential: Internal Use Only
5-15
5.13
CE Event - Correctable Memory Error
Previous Solaris Messages:
May 8 14:35:30 thishost SUNW,UltraSPARC-II: CPU1 CE Error: AFSR
0x00000000.00100000 AFAR 0x00000000.8abb5a00 UDBH Syndrome 0x85
MemMod U0904
May 8 14:35:30 thishost SUNW,UltraSPARC-II:
ECC Data Bit 63
was corrected
May 8 14:35:30 thishost unix: Softerror: Intermittent ECC Memory
Error, U0904
Improved Message:
Aug 16 16:34:48 thishost SUNW,UltraSPARC-II: [AFT0] Corrected
Memory Error on CPU1, errID 0x00000036.629edc25 AFSR
0x00000000.00100000<CE> AFAR 0x00000000.00347dc0 AFSR.PSYND
0x0000 (Score 05) AFSR.ETS 0x00 Fault_PC 0x1002fe20
Aug 16 16:34:48 thishost
UDBH Syndrome 0x85 Memory Module 1904
Aug 16 16:34:48 thishost SUNW,UltraSPARC-II: [AFT0] errID
0x00000036.629edc25 Corrected Memory Error on 1904 is Intermittent
Aug 16 16:34:48 thishost SUNW,UltraSPARC-II: [AFT0] errID
0x00000036.629edc25 ECC Data Bit 63 was in error and corrected
5-16
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
CHAPTER
6
Configuration Steps
The following section describes the actions to take to properly configure a system to
capture failure information, and the specific problem solving steps to take after one
of the system problems scenarios previously discussed has been identified.
As mentioned previously, the Sun Solution Center engineers may ask for additional
information or recommend additional steps. The following is a guide to configure
systems and capture the most appropriate information for the majority of error
situations.
Step 1 Initial System Configuration (before errors occur)
Step 2 Error Identification and Error Response
Step 3 Information Gathering and Follow-up Tasks
6-1
Sun Proprietary/Confidential: Internal Use Only
6.1
Step 1 - Initial System Configuration
Table 6-1 shows Sun Enterprise xx00 server preparation steps to properly recover
and save necessary error information for further problem solving.
TABLE 6-1
System Configuration Task List
Configuration Item
Reason
Instructions or Process
Frequency
General Information
Gathering
Some detailed failure
information is only logged
to the system console.
Refer to Appendix E for
additional information.
Every Server
Update the Kernel Update
(KU) patches on Solaris
2.5.1 and above.
Kernel Level E-Cache
Scrubber and enhanced
error messaging for 2.6 &
above.
Refer to the Solaris KU patch
README file for install
instructions.
Every Server
Update system
flashprom/EEPROM
Newer versions (27 &
above) capture additional
failure data.
Refer to firmware patch
(103346) README file for
instructions.
Every xx00 Server
Enable saving of a system
panic dumps.
To save vmcore dumps for
further analysis for panics.
Refer to Appendix A for
information.
Every Server
Install iscda script
To produce initial vmcore
dump analysis data.
Refer to Appendix B for
information.
Every Server
Enable Console Logging
6-2
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
6.2
Step 2 - Error Identification and
Error Response
Table 6-2 describes the symptom, the possible reason for the problem, and the
specific steps to take, once the problem has been identified.
TABLE 6-2
Identifying a System Error
Symptom
Identification Tasks
Data Collection
System Rebooted
Determine if any of the
following can be associated with
the system reboot.
1. Was system reboot a user initiated
event? f so, understand why this is
considered unexpected reboot.
PossibleAdministrationissue.
Consult with user who shutdown
system unexpectedly to determine
the reason.
2. Did the system reboot due to ECache Parity Errors? Solaris 2.6
and higher KU patches have this
capability.
Collect /var/adm/messages file
for further review for indications
of
E-Cache parity errors.
3. Does a new vmcore/unix system
dump exist in /var/crash/‘uname
-n‘ If yes, this was a system panic.
New vmcore/unix files indicates a
system panic occurred. See
System Panic in Step 3 below.
4. Determine if a Fatal Reset/Fatal
Error or other hardware problem
was the cause of the system reboot.
Console output can indicate this.
Collect console and prtdiag -v
output. Determine if any new
hardware problems were detected.
See Fatal Reset in Step 3 below.
5. Review console messages for
indications of why the system may
have rebooted.
Look for indications of Fatal
Resets or Fatal Errors, system
panics which failed to save a
dump, etc.
1. Does a new vmcore/unix dump
exist in /var/crash/‘uname -n‘
If yes, then this was a system panic.
Execute the iscda script to obtain
an initial analysis of the system
panic.
See System Panic in Step 3 below.
2. Review console messages for
indications of why the system may
have or have not saved a vmcore
file.
Look for indications as to why the
system failed to save a dump or
why the dump aborted. Dump
device too small, second panic
during the panic process, etc.
System Panic
Determine if any of the
following can be associated with
a system panic.
Chapter 6
Sun Proprietary/Confidential: Internal Use Only
Configuration Steps
6-3
TABLE 6-2
Identifying a System Error
Symptom
Identification Tasks
Data Collection
Fatal Reset / Error
1. Determine if a Fatal Reset/Fatal
Error or other hardware problem
was the cause of the reboot.
Console output can indicate this.
Use console output and prtdiag -v
to determine if any new hardware
problems have been detected.
See Fatal Reset in Step 3 below.
1. If possible, determine the state of
the system LEDs to verify if they
are in a normal operating
condition.
Observe the system to determine
the state of the LEDs. Prtdiag -v
will provide some info if it can be
used.
2. Determine if there is any existing
login sessions that are active. If so,
use these sessions to collect
additional information and if
necessary force a system panic.
If an active login can be accessed,
commands (ifconfig, netstat, etc.)
should be issued to verify the
state of the network. This active
login can also be used to force a
panic.
3. Determine if a console login can be
initiated using the serial console
port.
Perform the same steps as shown
above for an active network login.
4. If no access is possible, attempt to
send a system break signal via the
system console.
If OBP can be entered, refer to
Step 3 below regarding further
OBP data collection commands.
5. Locate the XIR button and prepare
to push it. Document the system
LEDs and console for any
response.
The XIR button may or may not
cause the system to respond, since
this is a hard hang condition.
6. Prepare to Power Cycle the system.
This will reset the system and
reboot.
1. If possible, determine the state of
the system LEDs to verify if they
are in a normal operating
condition.
Observe the system to determine
the state of the LEDs. Prtdiag -v
will provide some info if it can be
used.
Determine if any of the
following can be associated
with a Fatal Reset / Fatal Error.
System Soft Hangs
Determine if any of the
following reasons can be
associated with a system soft
hang.
System Hard Hangs
Determine if any of the
following reasons can be
associated with a system hard
hang.
6-4
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
TABLE 6-2
Identifying a System Error
Symptom
Identification Tasks
Data Collection
System Hard Hangs
(continued)
2. Determine if there is any existing
login sessions that are active. If so,
use these sessions to collect
additional information and if
necessary force a system panic.
If an active login can be accessed,
commands (ifconfig, netstat, etc.)
should be issued to verify the
state of the network. This active
login can also be used to force a
panic.
3. Determine if a console login can be
initiated using the serial console
port.
Perform the same steps as shown
above for an active network login.
4. If no access is possible, attempt to
send a system break signal via the
system console.
If OBP can be entered, refer to
Step 3 below regarding further
OBP data collection commands.
5. Locate the XIR button and be
prepared to push it. Note the
system LEDs and console for any
response.
The XIR button may or may not
cause the system to respond, since
this is a hard hang condition.
6. Prepare to Power Cycle the system.
This will reset the system and
reboot.
Chapter 6
Sun Proprietary/Confidential: Internal Use Only
Configuration Steps
6-5
6.3
Step 3 - Information Gathering and
Follow-up Tasks
Problem
Error Response
Follow-up Tasks / Reporting
System Rebooted
1. If reboot was caused by an ECache parity error, collect
messages file.
Collect /var/adm/messages file to
be given to Sun Support for
analysis.
2. Verify that all hardware is present
and functional.
Use prtdiag -v to determine
hardware status.
3. Verify that all I/O (disks,
filesystems, etc.) are present, still
mirrored, etc.
Use appropriate Solstice DiskSuite
and Veritas commands.
4. Verify that all user and system
processes are functional.
Use appropriate commands to get
state of user and system processes.
5. Report problem to Sun as an
Unexplained System Reboot
Place service call to Sun Service
using documented call process.
6. Schedule maintenance window as
necessary for any service actions.
As recommended by the Sun
support Engineer.
1. If valid vmcore/unix dump exists,
execute iscda script for initial
analysis.
Collect output from iscda. Key on
panic string information.
2. If no vmcore file, review console
output for additional information.
Review console output for panic
string and reason dump failed.
3. Collect messages file as well and
review for errors prior to the panic.
Collect /var/adm/messages file to
be given to Sun Support for
analysis.
4. Verify that all hardware is present
and functional.
Collect prtdiag -v output to
determine hardware status.
5. Verify that all I/O (disks,
filesystems, etc.) are present, still
mirrored, etc.
Use appropriate Solstice DiskSuite
and/or Veritas commands.
6. Verify that all user and system
processes are functional.
Use appropriate commands to get
state of user and system processes.
7. Report problem to Sun as a
System Panic
Place service call to Sun Service
using documented call process.
System Panic
6-6
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
Problem
Fatal Reset / Error
System Soft Hang
Error Response
Follow-up Tasks / Reporting
8. Schedule maintenance window as
necessary for any service actions.
As recommended by the Sun
support Engineer.
1. If Fatal Reset/Fatal Error occurred,
review console messages.
Collect console messages to
determine root cause of Fatal Reset.
2. PRIOR to power cycling the
system, capture prtdiag -v output.
Collect prtdiag -v for additional
Fatal Reset data & hardware status.
3. Verify that all I/O (disks,
filesystems, etc.) are present, still
mirrored, etc.
Use appropriate Solstice DiskSuite
and Veritas commands.
4. Verify that all user and system
processes are functional.
Use appropriate commands to get
state of user and system processes.
5. Report problem to Sun as an
Fatal Reset/Fatal Error
Place service call to Sun Service
using documented call process.
6. Schedule maintenance window as
necessary for any service actions.
As recommended by the Sun
support Engineer.
7. .If necessary, enable console
logging if possible.
Request that customer enable
console connection/logging via
serial ttya port.
1. If possible, determine the state of
the system LEDs to verify if they
are in a normal operating
condition.
Document LED state on all
individual CPU/Memory Boards,
Clock Boards and front panel LEDs.
2. Determine if there is any existing
login sessions that are active. Use
these logins before proceeding any
further.
Note if login sessions are available,
and if so, use Solaris commands to
get system status.
3. Determine if a console login can be
initiated using the serial console
port (ttya port). Look for errors in
/var/adm/messages, etc.
Note if login sessions are available,
and if so, use Solaris commands to
get system status.
4. If no access is possible, attempt to
send a system break signal via the
system console.
Execute OBP commands to gain
additional failure information.
Refer to Appendix C & D.
5. Press the XIR button. This button
is located at the rear of the xx00
server.
Document the system LEDs and
monitor the console for any output.
Chapter 6
Sun Proprietary/Confidential: Internal Use Only
Configuration Steps
6-7
Problem
Error Response
Follow-up Tasks / Reporting
System Soft Hang (continued)
6. Power Cycle the system.
Power off the system using the key
switch and turn key to the
Diagnostic position to run
maximum level POST. Document
any errors.
7. Verify the hardware state of the
system.
ICollect prtdiag -v for hardware
status & hardware configuration.
8. Verify that all I/O (disks,
filesystems, etc.) are present, still
mirrored, etc.
Use appropriate Solstice DiskSuite
and Veritas commands.
9. Verify that all user and system
processes are functional.
Use appropriate commands to get
state of user and system processes.
10.Report problem to Sun as an
Soft System Hang.
Place service call to Sun Service
using documented call process.
11.Schedule maintenance window as
necessary for any service actions.
As recommended by the Sun
support Engineer.
1. If possible, determine the state of
the system LEDs to verify if they
are in a normal operating
condition.
Document LED state on all
individual CPU/Memory Boards,
Clock Boards and front panel LEDs.
2. Determine if there is any existing
login sessions that are active. Use
these logins before proceeding any
further.
Note if login sessions are available,
and if so, use Solaris commands to
get system status.
3. Determine if a console login can be
initiated using the serial console
port (ttya port). Look for errors in
/var/adm/messages, etc.
Note that console login via the
serial port is working, but network
connections are not. Suspect a
network problem.
4. If no access is possible, attempt to
send a system break signal via the
system console.
Execute OBP commands to gain
additional failure information.
Refer to Appendix C & D.
5. Press the XIR button. This button
is located at the rear of the xx00
server.
Document the system LEDs and
monitor the console for any output.
6. Power Cycle the system
Power off the system using the key
switch and turn key to the
Diagnostic position to run
maximum level POST. Document
any errors.
System Hard Hang
6-8
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
Problem
Error Response
Follow-up Tasks / Reporting
System Hard Hang (continued)
7. Verify the hardware state of the
system.
Collect prtdiag -v for hardware
status & hardware configuration.
8. Verify that all I/O (disks,
filesystems, etc.) are present, still
mirrored, etc.
Use appropriate Solstice DiskSuite
and Veritas commands.
9. Verify that all user and system
processes are functional.
Use appropriate commands to get
state of user and system processes.
10.Report problem to Sun as a
Hard System Hang.
Place service call to Sun Service
using documented call process.
11.Schedule maintenance window as
necessary for any service actions.
As recommended by the Sun
support Engineer.
Chapter 6
Sun Proprietary/Confidential: Internal Use Only
Configuration Steps
6-9
6-10
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
CHAPTER
A
Enabling Saving System Dump
The following section defines the four basic steps to enable the process of saving a
system dump, known as a vmcore file.
For a complete description reference InfoDoc 12031.
A.1
Enabling savecore(1M) and Verifying
Disk Space
Note – Frequency Recommendation: This process should be performed on every
Sun server.
Step 1: Enable savecore(1M).
The savecore program must be enabled (turned on) in the Solaris scripts:
■
Solaris 2.5.1 and 2.6: /etc/rc2.d/S20sysetup
(Disabled by default)
■
Solaris 7 and above:
(Enabled by default)
/etc/rc2.d/S75savecore
▼ Enable savecore in Solaris 2.5.1 or 2.6.
● Manually un-comment the six lines in /etc/rc2.s/S20sysetup
if [ ! -d /var/crash/‘uname -n‘]
then mkdir -m 0700 -p /var/crash/‘uname -n‘
if
echo ’checking for crash dump...\c ’
A-1
Sun Proprietary/Confidential: Internal Use Only
savecore /var/crash/‘uname -n‘
echo ’’
▼ Enabling savecore in Solaris 7 and 8
By default, savecore is ENABLED in Solaris 7 and 8.
● Run the dumpadm command as root to verify this.
The following is example output.
# dumpadm
Dump content: kernel pages
Dump device: /dev/dsk/c0t0d0s1 (swap)
Savecore directory: /var/crash/machinename
Savecore enabled: yes
Step 2. Verify that there is sufficient swap space to dump memory.
Swap space is used to initially save the dump of system memory. By default Solaris
uses the first swap device that is defined. This first swap device is known as the
dump device.
For servers, this swap/dump device size should be at least 1Gbyte and preferably
2Gbyte.
To determine the initial swap device that is defined, use the swap -l command.
The following is an example output.
# swap -l
swapfile
/dev/dsk/c0t3d0s0
/dev/dsk/c0t1d0s0
/dev/dsk/c0t1d0s1
32,24
32,8
32,9
dev
16
16
16
swaplo
blocks
free
4097312
4062048
4097312
4060576
4097312
4065808
The amount of space available is taken from the "blocks" column and then
multiplied by 512. Taking the blocks from the first entry, c0t3d0s0, we see a device
size of ~2Gbyte.
4097312 * 512 = 2,097,823,744 or approximately 2Gbyte.
A-2
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
Step 3. Verify that there is sufficient file system space for vmcore files.
The scripts mentioned above (/etc/rc2.d/S20sysetup or
/etc/rc2.d/S75savecore) define the directory where savecore will save the
resulting vmcore files.
Note – It is recommended that there be a minimum of 1Gbyte of free space on the
file system used to store vmcore files.
By default the file system and directory where savecore files will be saved is:
/var/crash/‘uname -n‘
i.e. for the "mysystem" server, the default directory is:
/var/crash/mysystem
The directory/file system specified must have space for the resulting vmcore.
This can be checked by looking at the free space of the file system. Use the df -k
command to validate this.
# df -k /var/crash/‘uname -n‘
If there is no space in /var/crash (the default) then any other locally mounted (not
NFS) file system can be used.
To change the default directory, the following commands can be used to remove the
current savecore dump location, create a new directory in a larger file system, then
create a symbolic link to point from the old location to the new location:
#
#
#
#
cd /var/crash
rmdir -f mysystem
mkdir /some_large_filesystem/mysystem
ln -s /some_large_filesystem/mysystem mysystemhost
Step 4. Verify the following savecore patches are applied.
If the swap/dump device shown above is configured over 2Gbyte, the following
patches must be applied. Note that the actual system dump (vmcore file) may be
much smaller than 2Gbyte, and the issue these patches correct is simply a problem
with swap/dump device sizes greater than 2Gbyte.
For simplicity, it is recommended that these patches be applied no matter the size of
the swap/dump device.
■
Solaris 2.5.1: 108083-01
Chapter A
Enabling Saving System Dump
Sun Proprietary/Confidential: Internal Use Only
A-3
■
Solaris 2.6:
107490-01
For additional detailed swap device configuration and sizing hints, start with these
SunWorld Online articles:
■
Swap Space part 1
http://www.sunworld.com/swol-12-1997/swol-12insidesolaris.html
■
Swap Space part 2
http://www.sunworld.com/swol-01-1998/swol-01insidesolaris.html
■
Clearing up questions
http://www.sunworld.com/sunworldonline/swol-07-1998/swol-07perf.html
A.2
■
Adrian Cockcroft and Richard Pettit’s new book.
■
Sun Performance and Tuning, 2nd Edition.
Verify Core Dump Process
The following section describes how to validate that a system is capable of saving a
valid system dump (vmcore).
For a complete description, reference Sun InfoDoc 12031 or contact Sun
Microsystems for assistance.
Note – Frequency Recommendation: Test once on every machine where a system
dump has not already been taken successfully due to an error.
This is normally a task that would be done just prior to placing a system into
production.
If you are the system administrator or system owner, you must force your
system to crash in order to test your savecore setup. The following
describes this process for Solaris releases up to and including Solaris 7.
1. Back up all of your data.
System crashes can result in non-recoverable and catastrophic loss of data.
2. Gracefully halt your system using ’halt’ or ’init 0’.
A-4
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
3. At the ok> boot prom prompt enter: sync
Your system should start to panic at this time. You should see "dumping" messages.
Next, the system will attempt to reboot. During this process you should see some
savecore messages.
4. Once the system is rebooted, look in your savecore directory and see if you have
system crash dump files there.
They will be named "unix.?" and "vmcore.?", where ? is the integer dump number.
There should also be a "bounds" file. This contains the next crash number for
savecore to use.
For Solaris 8, there is an option (-d) to the reboot(1M) command. This
option to reboot causes a system dump to be taken, just prior to the
system being rebooted.
1. Back up all of your data.
System crashes can result in non-recoverable and catastrophic loss of data.
2. Reboot the system using the -d option to reboot: reboot -d
The system will panic and then attempt to reboot. During this process you should
see some savecore messages.
3. Once the system is rebooted, look in your savecore directory and see if you have
system crash dump files there.
They will be named "unix.?" and "vmcore.?", where ? is the integer dump number.
There should also be a "bounds" file. This contains the next crash number for
savecore to use.
Chapter A
Enabling Saving System Dump
Sun Proprietary/Confidential: Internal Use Only
A-5
A-6
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
CHAPTER
B
ISCDA Script
The Initial System Crash Dump Analysis script (iscda script) is used to assist in the
analysis of Solaris 2.X system crashes. It may also be used on live systems, but this
is not normally the mode of execution for iscda
For a complete description, reference Sun InfoDoc 10214 or contact Sun
Microsystems for assistance.
Note – Frequency Recommendation: Make script available to every machine. Run
the iscda script only when a system panic dump (vmcore file) is produced.
The iscda script has been tested and runs on Solaris 2.3 through 2.6. Iscda also
works on Solaris 7, with some "symbol not found" messages and the crash time will
be incorrect if you are booted in 64 bit mode.
To prepare a system to run iscda, place the iscda script in the directory which
savecore(1M) uses to save the vmcore and unix files. Execute chmod so that it is
executable as root (you will need to run it as root).
To produce a text file summary analysis of a system dump (vmcore file), execute
iscda as follows:
# iscda unix.? vmcore.? > iscda.out
Where ? is the integer number extension on the vmcore and unix files.
A few examples of when this script may be helpful:
■
Due to the size of the vmcore files, transferring them can take time.
When placing a service call due to a system panic, inform the Sun engineer that
you have iscda output. This iscda output can be easily e-mailed, etc. for an initial
look into the reason for the system panic.
■
Text items in the Solaris panic string and/or stack can be used to search SunSolve
for related problems.
B-1
Sun Proprietary/Confidential: Internal Use Only
Note that some panics have similar appearances with regards to the panic string
and stack trace, so any bugs believed to be related should be confirmed by the
Sun engineer handling your call.
■
If you wish to save a ’summary’ of the core file for your own history rather than
saving the whole kernel core file which is considerably larger.
Note – Iscda output doesn’t take the place of analyzing the entire system
dump/vmcore file, however, if the panic is a known issue, this output can
sometimes help to reduce the overall troubleshooting time.
B-2
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
CHAPTER
C
System Abort Sequences
This section outlines keyboard sequences and procedures. It will give exact syntax as
well as excerpts from resultant information.
IMPORTANT! The abort sequence does NOT cause an automatic dump of system
memory. If a vmcore file (dump) is the goal of the abort sequence, then make sure a
sync command is issued while at the OBP ok prompt. If the sync command is not
issued, and the system is rebooted, the state of the system is lost.
Note – Sometimes the local "Stop/a" key sequence or the remote send break
sequences below do not work. If there is a local keyboard plugged into the
keyboard connection at the rear of the system, attempt to disconnect/re-connect it.
This sometimes forces the system to the OBP prompt.
C.1
L1/a Stop/a Keyboard Abort Sequences
(Local Terminal Connection)
IMPORTANT! The L1/a or Stop/a keyboard sequence only works if you have a
graphical based console and a terminal connected directly to the server.
For cases where the serial ttya port is being used as the console interface, refer to the
next section regarding how to send a break signal via a terminal server or other
device.
The abort sequence is the method whereby a user can immediately drop from
running Solaris into kadb or OBP. This method takes the machine to a state where
additional information can be gathered. The system can also be recovered or
rebooted from this state.
C-1
Sun Proprietary/Confidential: Internal Use Only
The sequence is generated by holding down (depressing) the "Stop" key and then
pressing the lower case "a" key. The "Stop" key is located on the upper left side of the
Sun keyboard. It has the word Stop printed on the top and on some Sun keyboards,
have L1 printed on the front face of the key as well.
On the console, the following output will be seen.
Type ‘go’ to resume
ok go
The system should resume executing Solaris.
Or to cause the system to save a dump, issue a sync command.
Type ’go’ to resume
ok sync
C.2
Abort Sequence via a Serial Terminal
Server
This method is to be used when access to the Sun Enterprise xx00 console is via a
network terminal server, using the telnet protocol.
mysystem% telnet annex_box
Trying 129.xxx.xxx.xxx...
Connected to annex_box
Escape character is ‘^]’
Enter Annex port name or number 52
Annex username: user_name
Annex password:
Permission granted
Attached to port 52
C-2
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
At this point, you should try to login to the system. If the system is hung, etc. you
may not be able to login. If so, skip down below to where you enter the telnet
command mode.
mysystem console login
Unix(r) System V Release 4.0 (mysystem)
login: root
passwd:
Sun Microsystems Inc. SunOS 5.5.1 :02/28/96 May 1996
#
Issue the appropriate command sequence to enter the command mode of telnet.
Normally this is done by pressing the "Control Key" and the "]" key simultaneously.
The "]" key is often referred to as the right bracket key.
From the telnet command prompt, issue the send brk command to send a break
signal from the terminal server.
telnet> send brk
Type ‘go’ to resume
{1b} ok
{1b} ok go
The system should resume executing Solaris.
Or to cause the system to save a dump, issue a sync command.
Type ’go’ to resume
{1b} ok sync
Chapter C
System Abort Sequences
Sun Proprietary/Confidential: Internal Use Only
C-3
C.3
Abort Sequence via a Direct Connect
ASCII Terminal
This method is used when access to the Sun Enterprise xx00 console is via an ASCII
terminal directly connected to the system console (ttya) serial port.
Note – The key sequence to generate a break can vary, based on the type of ASCII
terminal used. Some ASCII terminals have a break key that will generate the
necessary sequence. Some ASCII terminals will send a break if the "Control Key" is
held down and then a lower case "x" is pressed. For the appropriate key sequence to
generate a break, consult the manual for the ASCII terminal being used.
You may or may not see a login prompt at the console screen.
On the console, the following output will be seen.
Type ‘go’ to resume
ok go
The system should resume executing Solaris.
Or to cause the system to save a dump, issue a sync command.
Type ’go’ to resume
ok sync
C.4
Remote System XIR and Remote Power
Control Sequences
The following terminal key sequences can be executed ONLY with a console that is
connected to ttya serial port of the Sun Enterprise xx00 server. In addition, the
following criteria must be met:
■
C-4
The key switch must be in either the On or Diagnostic setting. If it is in the Secure
or Off position, the remote key sequences and button resets are ignored.
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
■
Security features (such as OpenBoot security-mode) must be disabled.
■
The type speed must be no faster than 0.5 seconds and no slower than 5 seconds
between characters.
To force an XIR event, execute the following command. This key sequence is two
carriage returns, a tilde, and the key sequence of the Control Key/Shift Key/ and the
letter x. For example:
<CR> <CR> <~> <Control-Shift-x>
Once the system enters the XIR state, the following command will display XIR data.
.xir-state-all
The XIR failure data from the above command can be diagnosed at:
http://cte-www.uk/cgi-bin/xir-cgi.tcl
To power cycle (toggle power off then back on) on a Sun Enterprise xx00 server, the
following command sequence is used:
<CR> <CR> <~> <Control-Shift-p>
Chapter C
System Abort Sequences
Sun Proprietary/Confidential: Internal Use Only
C-5
C-6
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
CHAPTER
D
OBP & kadb Commands
D.1
OBP Debug Commands
TABLE D-1 show commands that are available at the OBP (ok) prompt. They allow the
user the ability to perform some low level data gathering.
These commands are normally executed when a system was running Solaris and
either hung and was forced (aborted) into OBP, or the system panic dump process
failed and the system is left at the OBP prompt.
TABLE D-1
OBP Debug Commands
registers
Display values in %g0 through %g7, plus %pc, %npc, %psr, %y, %wim, %tbr.
.trap-registers
Display values in the trap related registers.
locals
Display the values in the i, l and o registers.
psr
Display the processor status register.
ctrace
Display Solaris return stack showing C subroutines.
sync
Cause Solaris to attempt to save a system panic (vmcore).
D-1
Sun Proprietary/Confidential: Internal Use Only
D.2
kadb Debug Commands
If booted under kadb, the system will enter kadb if the system panic’d or if the
system was sent an abort sequence. TABLE D-2show commands that are available at
the kadb prompt. They allow the user the ability to perform some low level kernel
debugging.
Some of this output may be lengthy, so be prepared to log more than just a few lines
of information.
TABLE D-2
D-2
kadb Debug Commands
<sp$<stacktrace
Display Solaris kernel stack trace information.
$<threadlist
Display a list of threads that were executing at the time.
$<thread.brief
Display more thread information.
$<cpus
Display CPU structures.
$<msgbuf
Display the Solaris message buffer.
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
CHAPTER
E
System Console Logging
System console logging is the ability to collect and log Sun Enterprise server console
output. This output is delivered through the serial port (ttya port) on each of the
Sun Enterprise 6x00, 5x00, 4x00, 3x00 servers.
E.1
Purpose for Console Logging
The purpose of console logging is so that the console messages can be captured and
used to improve the quality and timeliness of problem diagnosis. Fatal Reset details
and POST output after a Fatal Reset is directed to the console.
This console data can result in fewer cases where a system interrupts and no data
appears to be recorded. In many of these interrupts, this data is the only output to
the console because in some failure modes, Solaris has already terminated and there
is no software still running in the system capable of logging messages to traditional
file system locations.
For this reason, console logging provides additional diagnostic information and
reduces the number of "unexplained system reboots", as the important
diagnostic/failure data is captured. This can also ensure that only the defective FRU
is replaced.
The following sections outline the possible console logging options. Note that there
may be other software and hardware vendors with equivalent products, however,
the functionality of these other products should be similar to what is discussed
below.
E-1
Sun Proprietary/Confidential: Internal Use Only
E.2
Configuring the Console on ttya
This functionality takes care of both normal system administration activities and the
analysis of system resets. A standard null modem cable is used to connect a serial
device (i.e. Network terminal server or ASCII terminal) to the ttya serial port. There
are OBP variables that should be set so as to handle this configuration.
There are two ways to enable the console on ttya.
▼
■
using the eeprom command
■
setting the variable directly while in OpenBoot (OBP)
The eeprom command.
■
The eeprom command displays/sets OBP variables.
Output has been deleted to save space.
mysystem# /usr/bin/eeprom
output-device=screen
input-device= keyboard
■
The following eeprom commands set the output-device and input-device OBP
variables to be ttya (serial console port A).
mysystem# eeprom output-device=ttya
mysystem# eeprom input-device=ttya
■
The eeprom command displays OBP variables and their values.
Output has been deleted to save space.
mysystem# eeprom
output-device=ttya
input-device=ttya
■
To revert back to a direct connect monitor and keyboard, execute the following
commands.
mysystem# eeprom output-device=screen
mysystem# eeprom input-device=keyboard
E-2
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
You can verify that the values have been changed back via the eeprom command
▼
Setting the variable directly while in OpenBoot
(OBP).
■
The printenv OBP command will show OBP variables and their current setting.
Output was deleted to save space.
ok> printenv
output-device screen
input-device keyboard
■
The setenv OBP command will set OBP variables.
ok> setenv output-device ttya
ok> setenv input-device ttya
ok> printenv
output-device ttya
input-device keyboard
■
To revert back to a direct connect monitor and keyboard, execute the following
commands.
ok> setenv output-device=screen
ok> setenv input-device=keyboard
You can verify that the values have been changed back via the eeprom command
There are other OBP variables that may require setting, depending upon the type of
connection that is being used. The default ttya port characteristics are denoted by
the variables listed below. The speed of the ttya port is hard-coded to be 9600
BAUD.
ttya-mode=9600,8,n,1,ttya-ignore-cd=true
ttya-rts-dtr-off=false
Chapter E
System Console Logging
Sun Proprietary/Confidential: Internal Use Only
E-3
E.3
Console Logging Options - Data Logging
Terminal Servers
A replacement for traditional terminal servers which do not have console logging
capability is a console server device from Lightwave, Inc. Lightwave console server
is the equivalent to a traditional network based terminal server, however, the
Lightwave device has memory added which is used as a "wrap around" message
buffer.
As console messages are output from the Sun Enterprise 6x00, 5x00, 4x00, 3x00
servers and Sun SparcServer 1000, 1000E SparcCenter 2000, 2000E servers, they are
stored in this memory. As the memory fills up, the oldest messages are overwritten.
One can connect to this console server via the network, and then display the
contents of the memory buffer for a specific system, thus retrieving the stored
console messages.
More information on the Lightwave console sever can be found at:
http://www.lightwavecom.com/products/conserver.htm
http://www.lightwavecom.com/products/ConsoleServer800.htm
E.4
Console Logging Options - Centralized
Console Control
A centralized console control solution is available from Aurora Technologies.
This is a solution that allows a single Sun workstation to serve as a console access
and logging point. Hardware is installed in the Sun workstation which supports
multiple serial ports and system consoles that are being controlled and monitored
via these ports. The workstation can both grant console access as well as log all
console activity on its local disk for review at anytime.
More information can be found at:
http://www.auratek.com/controltwr/controltwr.html
E-4
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
E.5
Console Logging Options - Tip line to
ttya
This may be one of the least expensive console logging options, but can create
challenges when attempting to monitor multiple systems. The system that is
performing the monitoring function must be up and operational, or logging of the
other systems console is lost.
To enable this console logging mode, take a standard serial cable and connect one
end to the Sun Enterprise Server x000 ttya port on the clock board, then connect the
other end of the cable to any serial port on any other local workstation.
Once the cable is connected, a user on this monitoring system can issue the tip
command (subject to configurational issues mentioned below) and be connected to
the other systems console. Note that prior to issuing the tip command, the user
must enable some form of logging, (i.e. using the log to file option of an Xterm
session, etc.).
Chapter E
System Console Logging
Sun Proprietary/Confidential: Internal Use Only
E-5
E-6
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
CHAPTER
F
LED Status Indicators
F.1
Overview
The following tables show the various LED indicators on the xx00 servers. There are
LEDs on the servers’ Front Panel and on the Clock Board assembly that show overall
system or server status. These LEDs are shown in TABLE F-1 and TABLE F-2.
TABLE F-3 show LEDs on the CPU/Memory Boards, I/O Boards and Disk Board
LEDs. The LEDs are either yellow or green in color. How they are illuminated
determines the specific state or condition of that individual FRU assembly.
TABLE F-4 show LEDs on the power supplies.
TABLE F-1
Sun Enterprise Server Front Panel and Clock Board LED Status
Power LED
Service LED
Cycling LED
Condition
Off
Off
Off
No Power
Off
On
Off
Failure Mode
Off
Off
On
Failure Mode
Off
On
On
Failure Mode
On
Off
Off
Hung in POST/OBP
On
Off
On
Hung in OS
On
On
Off
Hung in POST/OBP
Hung in OS/Failed Component
On
On
On
Hung in POST/OBP
Hung in OS/Failed Component
On
Off
Flashing
OS Running Normally
F-1
Sun Proprietary/Confidential: Internal Use Only
TABLE F-1
Power LED
Service LED
Cycling LED
Condition
On
On
Flashing
OS Running with Failed
Component
On
Flashing
Off
Service LED Flashing
Slow Flash = Executing POST
Fast Flash = Executing OBP
On
Flashing
On
Service LED Flashing
OS or OBP Error
TABLE F-2
Notes for Sun Enterprise Server Front Panel and Clock Board LED Status
LED Name
Location
Note
Power LED
Left (Green)
Should always be on. If all three LEDs are off,
suspect power problem. If this LED is in any
other state than on and steady, it indicates a
problem.
Service LED
Middle (Yellow)
This LED should be off in normal operation.
If on, a component is in an error state and
you should check check individual board
LEDs. A lit service LED does not imply there
is an OS-related problem.
Cycling LED
Right (Green)
This LED should be flashing -- this is the
normal state.
TABLE F-3
F-2
Sun Enterprise Server Front Panel and Clock Board LED Status
Sun Enterprise CPU/Memory, I/O, and Disk Board LED Status
Power LED
Service LED
Cycling LED
Condition
Off
Off
Off
No Power
Off
On
Off
Failure Mode
Off
Off
On
Failure Mode
Off
On
On
Failure Mode
On
Off
Off
Hung in POST/OBP
On
Off
On
Hung in OS
On
On
Off
Hung in POST/OBP
Hung in OS/Failed Component
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
TABLE F-3
Sun Enterprise CPU/Memory, I/O, and Disk Board LED Status
Power LED
Service LED
Cycling LED
Condition
On
On
On
Hung in POST/OBP
Hung in OS/Failed Component
On
Off
Flashing
OS Running Normally
On
On
Flashing
OS Running with Failed
On
Flashing
Off
Service LED Flashing
Slow Flash = Executing POST
Fast Flash = Executing OBP
On
Flashing
On
Service LED Flashing
OS or OBP Error
NOTES: Low Power Mode - If the status of the LEDs on the board is off-on-off, this means
the board is in low power mode. This occurs when the board is disabled because it failed
POST, or if the board was just inserted. Low power mode is the only state in which you
may unplug the board while the system is running.
Disk Boards - The amber LED on disk boards installed in Sun Enterprise servers will
remain on when the Sun Enterprise server is running Solaris 2.6 5/98 or above. This is
normal, and it indicates the board is in low power mode (the board can be removed from
the system provided the disks have been idled).
TABLE F-4
Power Supply LED Status
Green LED
Yellow LED
Note
Off
Off
No AC input or key switch is turned off
On
Off
Normal Operation
On
On
Fan failure or one or more voltages out of
specification.
Off
On
One or more DC outputs failed, or voltages out
of specification, or system in low power state
Chapter F
LED Status Indicators
Sun Proprietary/Confidential: Internal Use Only
F-3
F-4
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
CHAPTER
G
Decoding xx00 Device Paths
Understanding Sun Enterprise Server 6x00 - 3x00 device paths is critical in
identifying messages logged, mainly by Solaris, regarding various I/O components
in these systems.
G.1
Device Driver Acronyms
The following are the device driver abbreviations that are used in device path names
to define various device drivers. These driver abbreviations will be found in the
second part of the device path name, as shown below.
fas - driver for fast/wide SCSI controllers (on-board)
hme - driver for Fast Ethernet
isp - driver for differential SCSI controllers and the SunSwift card
glm - driver for UltraSCSI controllers
scsi - driver for Small Computer Serial Interface (SCSI) devices
sf - driver for soc+ or socal Fiber Channel Arbitrated Loop (FCAL)
socal - driver for SPARC Storage Array (SSA) controllers
G.2
Device Path Example
The following example uses a device path from an Sun Enterprise 3000. The path
name is broken down into segments to identify the various locations slot, location on
the I/O board, etc. For this example, the following device path is used.
/[email protected],0/SUNW,[email protected],0/[email protected],0
G-1
Sun Proprietary/Confidential: Internal Use Only
Part 1 - [email protected],0
The first portion of the device path indicates the I/O board slot location. In the
first table below, we see that [email protected],0 correlates to the I/O board in slot 1,located
on the back of the machine (UE 3000’s only use the rear, odd number slots
because there are internal disks in the front).
Part 2 - SUNW,[email protected],0
The second portion of the device path indicates the type of onboardcontroller and
the Sys I/O bus it uses. I/O boards are basically split in half, each half is
controlled by a Psycho or Sys I/O chip (A or B). We see that SUNW,[email protected],0
correlates to the onboard (built-in) fast/wide SCSI controller at Sys I/O B on this
board.
Part 3 - [email protected],0
The next portion of the device path ([email protected],0) correlates to the SCSI disk (sd) set to
target id 0 (in this case an internal disk, since only internal disks should be
controlled by the onboard SCSI controller of the I/O board in slot 1).
Another Example:
/[email protected],0/SUNW,[email protected],0/[email protected],0/[email protected],0 (ssd27)
This device path correlates to an I/O board in slot 3 of a UE server ([email protected]), the
onboard socal controller ([email protected]), the GBIC port ([email protected]). The
[email protected],0 (ssd27) is a disk inside of an A5x00 array. The long number
after the ‘@’ sign is the world wide (unique) number of this particular disk.
Note – [email protected] is the GBIC port on the right and [email protected] is the GBIC port on the left when
looking at an I/O with soc+ (socal) board. On an Enterprise 3000 & 3500, [email protected] is the
GBIC port on the bottom and [email protected] is the GBIC port on the top.
G-2
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
G.3
Device Path Decoding Tables
TABLE G-1
On-Board Controller Definitions - (Part 1 segment)
FRONT
REAR
Sysio A
Sysio B
Sysio A
Sysio B
[email protected]
[email protected]
[email protected]
[email protected]
[email protected]
[email protected]
[email protected]
[email protected]
[email protected]
[email protected]
[email protected]
[email protected]
In addition to the on-board interfaces, each I/O Board has two SBus slots. TABLE G-2
indicates how the SBus slots are numbered, depending on which I/O Board slot the
SBus cards are installed.
TABLE G-2
I/O Board SBus Slot Assignments
I/O Board
Slot
SBus #
I/O Board
Slot
SBus #
0
[email protected]
[email protected]
1
[email protected]
[email protected]
2
[email protected]
[email protected]
3
[email protected]
[email protected]
4
[email protected]
[email protected]
5
[email protected]
[email protected]
6
[email protected]
[email protected]
7
[email protected]
[email protected]
8
[email protected]
[email protected]
9
[email protected]
[email protected]
10
[email protected]
[email protected]
11
[email protected]
[email protected]
12
[email protected]
[email protected]
13
[email protected]
[email protected]
14
[email protected]
[email protected]
15
[email protected]
[email protected]
Chapter G
Decoding xx00 Device Paths
Sun Proprietary/Confidential: Internal Use Only
G-3
G-4
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
CHAPTER
H
Use of prtdiag
The prtdiag(1M) command is used to display the system hardware configuration,
details of certain system failures and Power On Self Test results.
The prtdiag command is available on sun4u (Ultra) and sun4d (SPARCcenter 1000 &
2000) system platforms.
■
For a complete description reference InfoDoc 23479
■
For a complete patch description reference Patch Description 104595
■
For a complete patch description reference Patch Description 105642
The following patches are related to prtdiag, but are not critical.
■
Solaris 2.5.1 - 104595-09
■
Solaris 2.6
- 105642-08
Frequency Recommendation: Prtdiag should be run after every un-planned or
unexpected system interrupt, i.e. System panic, etc.
H.1
prtdiag Command
The prtdiag command is available on sun4u (Ultra) and sun4d (SPARCserver 1000 &
2000) system platforms. The first part of the prtdiag output displays the overall
active hardware configuration of the system (active CPUs and their speed, active
Memory and its size, and active I/O controllers and their type).
Additional information, including firmware levels, diagnostic result information, etc.
are displayed using the verbose or "-v" option.
The diagnostic portion of the prtdiag output displays any failed Field Replaceable
Units (FRU’s) in the system on most sun4u model servers.
H-1
Sun Proprietary/Confidential: Internal Use Only
H.2
Location of the prtdiag Command
(sun4u and sun4d)
The prtdiag command, is available on the sun4u (Sun Enterprise 6x00 - 3x00 )
system platforms, as well as several other Sun server models.
To run the prtdiag command, one will need to know which system platform they
are on because the directory path to the prtdiag command differs slightly, between
a sun4u and a sun4d platform.
To find out the system platform of a particular system, issue the uname -m
command.
If the system is a sun4u platform type, the prtdiag command resides in the
following directory:
/usr/platform/sun4u/sbin
H.3
Use of the prtdiag Command (sun4u)
The following shows how to execute the prtdiag command, on a sun4u class server.
Note that for the most information, use of the -v option is recommended.
# /usr/platform/sun4u/sbin/prtdiag -v
H-2
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
CHAPTER
I
Using the Explorer Utility
The Explorer script can be obtained from Sun Enterprise Services. If there are
specific questions regarding explorer, a Sun service call should be opened so that the
question(s) can be answered, etc.
Frequency Recommendation: To be installed on every system and run once a month,
or when any update is made to the system.
I.1
Installing Explorer
To install and run Explorer please perform the following steps:
Note – Explorer requires root privileges to run and root must have privileges to
write to the directory in which you install the explorer package.
To install the Sun Explorer package, transfer the compressed tar file to a local file
system on the server. Then uncompress, untar, and install the Explorer package.
The following are example steps:
# zcat SUNWexplo.tar.Z | tar xf # pkgadd -d . SUNWexplo
During installation you will be asked several questions, including:
■
Your Company Name
■
Your contract ID
■
This systems serial number, etc.
Sun asks that every effort be made to supply your ContractID, Serial number, and
Company Name, as this enables us to track the explorer output more effectively.
I-1
Sun Proprietary/Confidential: Internal Use Only
I.2
Executing Explorer
To run explorer manually after the install process, perform the following steps as
root:
# cd /opt/SUNWexplo
# ./explorer -mail(To automatically e-mail the results)
-or# ./explorer(To save a copy of the explorer output locally)
A copy of the explorer output is placed in /opt/SUNWexplo for your reference.
If you are asked to send explorer output to Sun with regards to an active problem,
you will be given instructions on how to address the e-mail, which explorer file to
send and where to send it to.
The Explorer output may also be sent to the Sun ftp site if it is too large to be
e-mailed.
I-2
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
CHAPTER
J
FTP Access to Sun
The following describe the procedures to transfer information (i.e. a vmcore file, etc.)
to Sun Microsystems external ftp sever in the USA..
J.1
Instructions for USA
The following are instructions to ftp a file(s) to Sun Microsystems external ftp server
known as sunsolve.sun.com.
Important: After ftp’ing any file it is necessary to inform the Sun Solution Center
and/or Sun Engineer you are working with that the file has been sent. This is
necessary because the files sent to Sun’s anonymous ftp server are erased every two
days. Thus files can be deleted before they are retrieved by Sun support engineers.
1. What files to send.
Depending on the problem, you will be directed to gather certain files and prepare
them to be sent to Sun via the anonymous ftp server, sunsolve.sun.com.
Note that the problem category definition section of this document also has
recommendations on the specific files that should be made available and be
prepared to send based on the Sun Solution Center engineer or Sun field engineer
request.
2. How to prepare the files for sending:
The best way to do this is by creating a tar file named <SunCase#>.tar containing the
required files and then compressing the file. Compressing the files is important,
especially if you are sending system dumps (vmcore files) as these can be hundreds
of MB’s in size.
The following is an example of how to create a tar file archive, assuming a Sun
assigned problem case number of 62221234:
J-1
Sun Proprietary/Confidential: Internal Use Only
# tar -cvf 62221234.tar file_names_to_send
After the tar file is created, please compress the tar file. The preferred method is to
use GNU Zip, which uses a tighter compression algorithm.
# gzip 62221234.tar
If you do not have gzip, use the standard Solaris compress command
# compress 62221234.tar
3. ftp the files to the Sun external ftp site:
Note that the standard ftp command may not work in all firewall situations or there
may be ftp proxy servers in place. Simply use the ftp procedure specific to your
location for external (outside the company) ftp transmissions.
# ftp sunsolve.sun.com
or
# ftp 192.18.99.148
Once connected, you can use "anonymous" login.
login: anonymous
password:
your e-mail address
ftp> cd cores
ftp> binary
ftp> put Case#.tar.gz(for example: 62221234.tar.gz)
-orftp> put Case#.tar.Z(for example: 62221234.tar.Z)
ftp> quit
J-2
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
CHAPTER
K
SunVTS
SunVTS, Sun Validation Test Suite, tests and validates Sun hardware by verifying the
connectivity and functionality of hardware devices, controllers and peripherals.
SunVTS is used by OPs, OEMs, SunService/Field, Design Engineering, SQA and end
users to ensure a "clean bill of health" of the overall system under test and its
underlying hardware. The tool is used for both Hardware validation and repair
verification. Several features are incorporated into SunVTS to enhance the ability of
the tool to diagnose systems. SunVTS makes an effective diagnostics tool as it
strives to stimulate, detect and identify hardware faults.
Frequency Recommendation: SunVTS should be loaded on every system and be
used to certify a system prior to entering production. SunVTS may also be used as
necessary when instructed by a Sun service engineer.
K.1
SunVTS Versions
When installing SunVTS it is important to realize this validation tool is OS version
and revision specific. TABLE K-1 shows the current versions and patches necessary
for each Solaris release.
TABLE K-1
SunVTS Version Matrix
Solaris Release
Sun VTS Version
SunVTS Patches
Solaris 8 1/01 (Update 3)
SunVTS 4.2
N/A
Solaris 8 10/00 (Update 2)
SunVTS 4.1
110353-xx
Solaris 8
SunVTS 4.0
110041-xx
110046-xx
K-1
Sun Proprietary/Confidential: Internal Use Only
TABLE K-1
K-2
SunVTS Version Matrix
Solaris Release
Sun VTS Version
SunVTS Patches
Solaris 7 11/99
SunVTS 3.4
109930-xx
110010-xx
Solaris 7 8/99
SunVTS 3.3
110039-xx
Solaris 7 5/99
SunVTS 3.2
110040-xx
110160-xx
107732-xx
Solaris 7 3/99
SunVTS 3.1
107890-xx
Solaris 7
SunVTS 3.0
107543-xx
Solaris 2.6 - 5/98
SunVTS 2.1.3
106387-xx
106810-xx
107135-xx
107542-xx
Solaris 2.6 - 3/98
SunVTS 2.1.2
106140-xx
106483-xx
Solaris 2.5.1 - 11/97
SunVTS 2.1.1
105799-xx
106073-xx
106247-xx
106528-xx
107237-xx
Solaris 2.6
SunVTS 2.1, REV=37.97.07.16
106114-xx
Solaris 2.5.1 - 8/97
SunVTS 2.1, REV=37.97.06.27
106250-xx
Solaris 2.5.1 - 4/97
SunVTS 2.0.1
105061-xx
105221-xx
105331-xx
Solaris 2.5.1
SunVTS 2.0
104774-xx
105673-xx
Solaris 2.5 SHWP (HW 1/96)
SunVTS 1.0
103097-xx
106703-xx
Solaris 2.5
SunVTS 1.0
103097-xx
106703-xx
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
CHAPTER
L
Forth Debug
IMPORTANT! Enabling Forth debug mode should ONLY be set based on specific
direction from Sun Microsystems support engineers. Setting Forth debug mode
should NOT be default on all systems.
L.1
Forth Debug - Explanation and Setting
Forth Debug is a tool that is used while in OBP. Forth Debug is roughly analogous to
adb/kadb.
For the system to have this tool enabled, there must be entries made in the
/etc/system file in the user-settable variables section. The system to be analyzed
must also be running a debug kernel.
set forthdebug=1
set obpdebug=1
A subsequent system reboot will show the following strings which verify that forth
debug is enabled. The command, /usr/bin/dmesg, is one way to verify.
obpsym: symbolic debugging is available.
Read 72209 bytes from misc/forthdebug
With this feature enabled, additional analysis techniques are available to whomever
is attempting system fault analysis on the Sun Enterprise Server. As in the case of
adb, additional user-defined ‘‘forth debug macros’’ can be created and used.
L-1
Sun Proprietary/Confidential: Internal Use Only
L-2
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
CHAPTER
M
Sun Enterprise xx00
Firmware Levels
The following information details how to determine the level of Sun firmware
(OpenBoot and POST) that is installed on a server. Also included is a matrix of the
latest released firmware versions based on server platform type.
Frequency Recommendation: Keep the flashprom up to date on all boards in every
machine.
M.1
Purpose of Flash Prom Updates
Flashprom updates offer problem fixes, feature enhancements, and changes in the
following areas:
■
Ability to support new hardware (i.e. CPUs, Memory, etc.)
■
Changes that affect prom level messaging.
■
Fixes to areas such as OpenBoot (prom level code) and the Power On Self Test
(POST).
As such, flashprom patches are considered critical patches and should be updated
when new versions are available.
It is possible to determine the Open Boot Prom (OBP) revision of most
SPARCstations from the UNIX prompt using the prtconf or prtdiag -v
commands. This allows the firmware version to be checked without system
downtime.
The following is an example of the flashprom section of the prtdiag -v output. In
this example, the flashprom is at the 3.2.28 level.
# prtdiag -v
M-1
Sun Proprietary/Confidential: Internal Use Only
<extra output deleted for readability>
System Board PROM revisions:
Board
11:19
Board
11:18
Board
11:19
Board
11:18
Board
11:19
Board
11:19
Board
11:19
Board
11:19
0:
OBP
3.2.28 2000/12/18 11:15
1:
FCODE 1.8.28 2000/12/18 11:14
iPOST 3.4.28 2000/12/18
2:
OBP
POST
3:
FCODE 1.8.28 2000/12/18 11:14
iPOST 3.4.28 2000/12/18
4:
OBP
3.2.28 2000/12/18 11:15
POST
3.9.28 2000/12/18
5:
OBP
3.2.28 2000/12/18 11:15
POST
3.9.28 2000/12/18
6:
OBP
3.2.28 2000/12/18 11:15
POST
3.9.28 2000/12/18
7:
OBP
3.2.28 2000/12/18 11:15
POST
3.9.28 2000/12/18
3.2.28 2000/12/18 11:15
POST
3.9.28 2000/12/18
3.9.28 2000/12/18
For complete details regarding the problems that a flashprom upgrade resolves,
reference the firmware patch 103346-xx README file.
For questions regarding Sun flashprom versions, please contact Sun Microsystems
for additional information.
M-2
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
M.2
Flashprom Version Matrix
TABLE M-1 show the firmware level matrix for a variety of Sun systems. This table is
current as of 12/11/00.
TABLE M-1
Sun Flashprom Version/Patch Matrix
Release Date
Machine Type
Patch ID
3/7/00
Ultra 1E
104288-xx
3/7/00
Ultra 1 (non-E)
104881-xx
3/7/00
Ultra 2
104169-xx
7/19/00
Ultra 5/10
106121-xx
3/7/00
Ultra 30
105930-xx
3/7/00
Ultra 60 E220R
106455-xx
3/7/00
Ultra 80 E420R
109082-xx
5/3/00
UE250
106503-xx
2/2/00
UE450
106122-xx
10/4/00
E3x00 - E6x00
103346-xx
8/7/00
Netra t1
108673-xx
Chapter M
Sun Enterprise xx00 Firmware Levels
Sun Proprietary/Confidential: Internal Use Only
M-3
M-4
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
CHAPTER
N
ESD Handling & Tools
For more information on ESD and procedures, refer to Chapter Three of the Sun
Microsystems Data Center Site Planning Guide (805-5863) or refer to edist.corp web
site.
Frequency Recommendation: Proper ESD procedures should be followed whenever
working on a any Sun system. ESD friendly environments should exist where all Sun
systems are installed and/or service operations occur.
N.1
Electrostatic Discharge (ESD)
Electrostatic Discharge (ESD) can be extremely damaging to electronic components.
By definition, static electricity is an electrical charge at rest. The discharge of this
built-up energy can cause numerous problems.
N.2
ESD Damage
Today’s electronic equipment has a much denser component geometry, and is
composed of thinner, more easily damaged materials. Changes in the design,
manufacturing process, and materials used has improved ESD sensitivity
considerably.
While Sun equipment has been designed to be tolerant of some ESD events, it is
important to take precautions in the design of the computer room to minimize
exposure to discharges. This is particularly important when unprotected
components are being handled during installations or upgrades.
N-1
Sun Proprietary/Confidential: Internal Use Only
Damage caused by ESD can take the form of catastrophic failures, but is more often
low-grade damage that may not show up during initial installation or upgrades.
However, damage caused by ESD can make the system more susceptible to a later
failure. Cumulative degradation of the components can also occur as the result of
repeated, low voltage exposures. These types of problems are very subtle and
extremely difficult to detect.
N.3
ESD Control
A detailed site-specific evaluation should be conducted to determine the most
appropriate ESD program for each controlled area. The following is a list of ways to
control static generation and ESD:
N-2
■
Use appropriate personal grounding equipment. Use of appropriate personal
grounding equipment (wrist straps, heel grounders, etc.) by operators in contact
with sensitive components can lessen the likelihood of human instigation of ESD.
■
Close cabinet covers at all times. Covers should only be opened by trained
personnel using proper grounding when inspections, repairs or reconfigurations
are needed.
■
Have a properly grounded access floor system with static dissipative tile surfaces.
This will provide a proper path to ground.
■
Use appropriate cleaning agents. The use of inappropriate cleaning agents or
excessive build-up of contaminants on the floor grid can allow an insulating
barrier to accumulate, thereby affecting the static dissipative capabilities of the
floor.
■
Use appropriate carts and furniture in the room. Use of appropriate carts and
furniture in the room will significantly decrease the risk of ESD since the
movement of inappropriate chairs or carts can easily generate static charges.
■
Room ionizers may be appropriate to help neutralize static charges in
manufacturing areas or print rooms. An ionizer emits negative and positive ions
that are attracted to charged objects, and can neutralize them.
■
Isolate activities or hardware that are likely to generate static charges. Isolation of
activities or hardware that are likely to generate static charges can decrease the
risks to more sensitive hardware. This is one reason why print equipment
(copiers, printers, etc.) should be kept out of the computer room.
■
Maintain appropriate relative humidity levels. Appropriate moisture levels will
help ease the dissipation of charges, lessening the likelihood of catastrophic
failures. The following chart illustrates the effect moisture levels can have on
electrostatic charge generation.
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
N.4
Electrostatic Voltage At Workstations
Reference: Data Center Site Planning Guide on sun.com
TABLE N-1
Static Voltage
Means Of Static Generation
Relative Humidity 10-20%
Relative Humidity 65-90%
Walking Across Carpet
35000
1500
Walking over vinyl floor
12000
250
Worker at bench
6000
100
Vinyl envelopes for work
instructions
7000
600
Common Poly bag picked up
from bench
20000
1200
Work chair padded with
urethane foam
18000
1500
Chapter N
ESD Handling & Tools
Sun Proprietary/Confidential: Internal Use Only
N-3
N-4
Sun Enterprise xx00 Problem Solving Manual • February 2002
Sun Proprietary/Confidential: Internal Use Only
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement