MAI COMPANY CONFIDENTIAL FIELD INFORMATION BULLETIN
SERVICE GROUP(S): MICRO,MINI,LARGE MFG: BASIC FOUR
NACS TYPE(S): 4120,4128 MFG. MODEL(S): 4020,4120,4128
CATEGORY: CPU DESC: GPx70 Multiprocessor
*** TABLE OF CONTENTS AS OF 08/29/95 ***
FIB 00001.General Information.........................................04/13/95
FIB 00002.The Use of Remote Diagnostics on the GPX70..................07/02/90
FIB 00003.System hangs on boot or incorrect system date message.......07/31/90
FIB 00004.Problems with GCR Tape Drive when Connected to GPx70 Systems07/31/90
FIB 00005.Garbage on System Console, Unable to Boot...................08/17/90
FIB 00006.AC power cable information..................................08/08/90
FIB 00007.O.S. rel. 1.0A has a problem with 7270 terminal configuratio10/05/90
FIB 00008.Incorrect Switch Settings for SCSI Interface GCR Tape Drives10/15/90
FIB 00009.DO NOT Cycle Power on a GCR Tape Drive with GPx System ON-LI11/21/90
FIB 00010.GPx70 Filesystem Space Notes & Cautions [ WPSF 653 ]........02/22/91
FIB 00011.Disk Overhead...............................................11/27/90
FIB 00012.Setform at Boot. [ WPSF 651 ]..............................11/27/90
FIB 00013.525MB Tape Length Determination.............................11/27/90
FIB 00014.Possible Short on Differential SCSI Adapter [ WPSF 647 ]....11/29/90
FIB 00015.Unable to access GCR Tape Drive.............................11/29/90
FIB 00016.SSM Issues [ WPSF 655 ]....................................11/30/90
FIB 00017.Excessive Async Errors Logged [WPSF 757 A1 ]................12/24/91
FIB 00018.VME 16-way upgrade rev F [ WPSF 677R ].....................04/02/91
FIB 00019.SCSI TAPE DEVICES MTS,HCC,GCR..............................03/19/91
FIB 00020.DATAPRODUCTS Printer Problems SSM [ WPSF 692 ]..............04/02/91
FIB 00021.PPC causes possible system crash [ WPSF 693 ]..............04/11/91
FIB 00022.UPS Warning Cable Incorrect [ WPSF 691 ]....................04/15/91
FIB 00023.SMD Disk Format Information [ WPSF 712 ]....................05/15/91
FIB 00024.Possible system crash using on-line DIVE [ WPSF 719 ]......06/19/91
FIB 00025.Exide UPS Information for the GPx70 [ WPSF 720 ]............07/17/91
FIB 00026.SCSI Device Termination.....................................07/05/91
FIB 00027.Potential AC to Frame Ground Short!!!.......................10/29/91
FIB 00028.ERR "No Commands Match 247" During Boot or DEMON PROCESSOR T04/17/92
FIB 00029.Adding Memory [ WPSF 794]...................................05/04/92
FIB 00030.SSM/VME I/O Space Corruption - Patch 57 [ WPSF 772 ]........05/28/92
FIB 00031.LEVEL 2 CONFIGURATION WITH TERMINAL SERVERS.................08/06/92
FIB 00032.Installing 16 Ways, Check NPROCs in DCT [WPSF 777].........08/18/92
FIB 00033.DCT changes for systems running OpenBasic [ WPSF 805 ]......01/14/93
FIB 00034.Memory bank chart for analyzing ECC errors..................02/01/93
FIB 00035.Screen mesh above air filter clogged........................04/21/93
FIB 00036.Fan modification to replace unreliable 12V fans.............03/23/94
FIB 00037.Release of 1.2A*08 for GPx 70, Enhancements/Fixes...........04/05/94
FIB 00038.Panics, System Hangs and external tape drive errors.........04/28/94
FIB 00039.Extra line feeds, garbled characters, duplicated lines, disk07/14/94
FIB 00040.SSM battery can be replaced - P/N YY030330..................11/18/94
CPU-BASIC FOUR-4020,4120,4128-----------------------Table Of Contents Pg01 of 01
FIB 00001 MAI COMPANY CONFIDENTIAL 04/13/95
*** General Information ***
The GPx Series 70 systems are symmetrical multiprocessor systems based on an
expandable set of system components. They utilize from two to ten 32 bit CPUs
running parallel, and have main memory capacities up to 96 MB.
An optional Disk ECA (Expansion Cabinet Assembly) can be added to the Central
Cabinet Assembly (CCA) for extra disk capacity, disk capacities up to 7.9 Gbytes
are supported.
There are two basic models differing only in physical size and capacity:
5x70 - six system bus card slots and 12 VME (I/O) bus card slots
6x70 - fourteen system bus card slots and sixteen vme (I/O) card slots
The x denotes the number of DPU cards. Each DPU (Dual Processor Unit) contains
two CPUs. Therefore a 5170 has two CPUs, and a 6570 has ten CPUs.
*** NACS MACHINE TYPES AND FEATURE NUMBERS ***
4120 - GPx 5x70 Series Systems
4128 - GPx 6x70 Series Systems
*** AVAILABLE DOCUMENTATION ***
M0232XA BOSS/VX Diagnostics Manual for GPX Series 70
M8226XA GPx Series 70 Installation and Maintenance Manual
M0187XB System Administrators Guide
M0188XB BOSS/VX Reference Manual
M0189XA BOSS/VX User Guide
M0205XA BOSS/VX Transportable Batch Comm (27xx/37xx) User Guide
M0217XX BOXX/VX TBC (27xx/37xx) User Guide
M0223XA BOSS/VX Programming Tools for the GPx70
M0225XA sysadm Utilities Manual
M0226XX BOSS/VX Commands Manual
M0236XA GPx Series 70 TCP/IP User Guide
P0001 UNIX System V/386 User Guide
P0002 UNIX System V/386 System Administrator Guide
P0003 UNIX System V/386 System Administrator Reference Manual
P0004 UNIX System V/386 Programmer Guide Volume I
P0005 UNIX System V/386 Programmer Guide Volume II
P0006 UNIX System V/386 Programmer Reference Manual
*** AVAILABLE TRAINING ***
Special schools on request.
*** SPECIAL TOOLS ***
MM597250:903826-001 :16-WAY LOOPBACK PCBA
MM597420:916807-001 :CABLE,LOOPBACK TEST J3/J4
MM597425:916808-001 :CABLE,LOOPBACK TEST PPCJ5
MM597375:400857-001 :PCBA VME BYPASS GPX-70
W514802-001 :System Serial Number tape (necessary if backplane or
serial number chip on the backplane P/N MM590014 is
replaced)
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB001 Pg001
*** SPECIAL ADMINISTRATIVE PROCEDURES ***
Diagnostic tapes are NOT shipped with the OS unless specifically ordered.
Diagnostic tapes can be ordered using the following codes:
W514785-103 - Diagnostic software
907851-003 - 1/4" tape media
To load the System Serial Number tape W514802-001, at --> or ---> prompt type:
bs tm(48,0)
ORIGINATOR: Norm Jones
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB001 Pg002
FIB 00002 07/02/90
*** The Use of Remote Diagnostics on the GPX70 ***
Prompt DIVE DEMON
------ ---- -----
Tape Boot : -> bs tm(48,2) bs tm(48,2)
--> bs tm(48,3) bs tm(48,3)
---> bh 3tm(48,4) bs tm(48,6)
: tm(48,30)
Disk Boot : -> bs bs
--> bs bs
---> bh wd(0,0)stand/dive bs wd(0,0)stand/dmn/exec
REMOTE DIAGNOSTICS:
1) Local console should be at level 1 boot prompt (->). It is suggested that
all auto-boots be turned off (monAuto = 0 / autoBoot = 0).
2) Verify calling modem and SSM NVRAM remote parameters are the same.
Example: remote (SSM NVRAM) Calling modem
baud = 2400 baud = 2400
stop = 1 stop = 1
size = 7 size = 7
parity = odd parity = odd
flow = xoff
3) Dial into GPX70 from calling modem, a connect signal will appear on the
calling terminal (remote console).
4) From local console, perform a level 1 boot.
5) The system should now be at a level 2 boot prompt (-->). Check that the
key switch is in the remote enable position. It should be noted that the
remote enable led is not on at this time, it will come on at the level 3
boot prompt (--->).
6) Type ' remote console ' at the local console and hit <cr>. At this point,
the remote terminal will have control. The level 2 boot prompt will now
appear on the remote terminal screen.
7) Execute a level 2 boot from remote terminal.
8) With a level 3 boot prompt displayed on the remote terminal, a DEMON or
DIVE boot may now be executed.
9) If DEMON is selected, both terminals (local & remote) will have control
(with screen vision) when the DEMON screen appears. If DIVE is selected,
only the terminal that is selected as 'console' will have control and
screen vision.
10) When system is re-booted, 'console' will go back to local control.
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB002 Pg001
ORIGINATOR: J.O'Brien
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB002 Pg002
FIB 00003 07/31/90
*** System hangs on boot or incorrect system date message ***
SYMPTOM: If the console port cable is not correctly wired, the system may
hang during the boot.
Another symptom may be that the console log may indicate that the
system date is incorrect.
Carefully observing the SSM log during the boot, you may find the
following message:
" Warning: console dcd not present "
FIX: The console cable MUST be wired per MAI Basic Four specification.
The correct part number for the console port cable is 916484-00x.
Shown below is the pin-out diagram for cable 916484:
PA PB
----- -----
| 2 |-----------| 2 |
Jumper pins | 3 |-----------| 3 | Jumper pins
1 to 6 to 8 | 7 |-----------| 5 |--| 5 to 6 to 8
\ | 5 |-----------| 7 | | /
\ |--| 8 |-----------|20 | | <-------/
\--> |--| 6 | | 8 |--|
|--| 1 | | 6 |--|
----- -----
DB-9 DB-25
ORIGINATOR: Harvey Mitchell
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB003 Pg001
FIB 00004 07/31/90
*** Problems with GCR Tape Drive when Connected to GPx70 Systems ***
SYMPTOM: Using dump command, uses less than a full reel (around 1" into the
reel), then asks for another tape.
FIX: The GCR driver defaults are currently incorrect, they may be corrected in
a future OS release but the following command will always work:
/etc/dump 0ufsd /dev/rmt/tg0h 2400 6250 /usr2
NOTE: options mean: 0 = dump entire file system (could use 0 - 9)
u = update /etc/dumpdates file with date
f = device file name will be supplied
(/etc/rmt/tg0h)
s = tape length, 2400 is shown in example, use
correct tape length if other than 2400'
d = tape density, 6250 is shown in the example
which is high density for the GCR
Options must be grouped together immediately following the dump
command, the corresponding values must follow in the same order
the options are entered, and the file system to be saved must be
the last entry.
SYMPTOM: Any GCR tape save operation fills the / (root) partition and all system
activity is interrupted (normally there are messages stating that the
root partition is full). dfspace shows no space in the root partition.
PROBLEM DETERMINATION: List tape device files:
ls -l /dev/rmt
1/2" tape device files are:
/dev/rmt/tg0 - standard 1/2"
/dev/rmt/tg0n - standard 1/2", no rewind
/dev/rmt/tg0h - high density (6250)
/dev/rmt/tg0hn - high density (6250) and no rewind
/dev/rmt/tg0m - medium density (1600)
/dev/rmt/tg0mn - medium density (1600) and no rewind
The device files are 0 block length files which have a
'b' or 'c' in the first column of the mode field. Any
file in the /dev/rmt listing with a '-' as a first
character in this field is a data file which has been
created accidentally while attempting to save files to
the tape.
FIX: Delete any accidentally created data files in the /dev/rmt directory. This
should free up the root directory and allow normal system activity again.
If the files listed above are missing, the system failed to recognize the
tape drive during the last boot process. This will occur if the GCR is
not powered up at boot time or if cables are disconnected, etc.
Verify cable connections, power the tape drive on, and re-boot the system.
The 1/2" tape device files will be re-created during boot if the system
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB004 Pg001
recognizes the tape drive. If the files are still missing verify the tape
subsystem with DEMON.
ORIGINATOR: Norm Jones
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB004 Pg002
FIB 00005 08/17/90
*** Garbage on System Console, Unable to Boot ***
SYMPTOM: Garbage displays on the system console (or possibly no display at
all) during system boot although other indicators and disk activity
appear normal. This problem will probably occur during the first
boot after the system has been powered on, however it can occur at
other times particularly if problems occur during nvram configuration.
FIX: The nvram has lost its settings and is attempting to communicate with
the system console at unknown baud rate and word structure settings.
1. Try all possible VDT port settings, don't overlook the possibility
of 38.4 baud. If a successful match is found go to step 3.
2. If unsuccessful at determining the nvram setting:
A. Power the system off.
B. Use ground strap and remove the SSM PCBA.
C. Install a jumper on J7 (if no extra jumper is available you can
move the jumper from J6 to J7 temporarily) to scrub the nvram.
D. Set your VDT as follows:
9600 baud, 8 bits, 1 stop bit, no parity, xon/xoff flow control
E. Install the SSM PCBA in the system and power-up. All system
console messages should now appear normally.
3. Set the nvram to correct settings
A. Get the system into a level 3 prompt by rebooting or by CTL-P,
displays --->
B. Enter "pr local"
The port settings should match your current VDT settings.
C. Enter "local size=7"
D. Change your VDT to 7 bit
E. Enter "local parity=odd"
F. Change your vdt to odd parity
G. If the baud rate is other than 9600 enter "local baud=9600",
change your VDT to 9600 baud
H. Correct all other nvram settings
I. If jumper J7 was installed on the SSM PCBA to scrub the nvram,
power the system down (use shutdown command if CTL-P was used to
get into the level 3 prompt), remove the SSM PCBA using proper
grounding procedures, remove J7 (re-install on J6 if it was moved
from J6), re-install the SSM PCBA and power the system up.
ORIGINATOR: Norm Jones
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB005 Pg001
FIB 00006 08/08/90
*** AC power cable information ***
The GPx70 INSTALLATION AND MAINTENANCE MANUAL, M8226A has an error concerning
the power cord receptacle required for the 6X70 Series.
The manual shows that an IG6-15R Flat Blade receptacle is required.
The receptacle actually required is a L6-15R, catalog no. IG 4550 or IG 4560.
This is a TWIST-LOCK, 15 AMP, 250V, ISOLATED GROUND receptacle.
ORIGINATOR: Harvey Mitchell
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB006 Pg001
FIB 00007 10/05/90
*** O.S. rel. 1.0A has a problem with 7270 terminal configuration ***
SYMPTOM: System experiences severe slowdown, "framing" errors log on random
serial devices, "ARCP timeout" warnings are posted on the console
terminal, serial port processes may hang.
PROBLEM DETERMINATION: Check to see if the application software is issuing
OPENs to ports configured as 7270.
FIX: This is a known problem with release 1.0A. If the application software
OPENs a port that is configured as 7270 and then writes to that port
and you are experiencing the above symptoms, contact Product Support for
assistance.
ORIGINATOR: Harvey Mitchell
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB007 Pg001
FIB 00008 10/15/90
*** Incorrect Switch Settings for SCSI Interface GCR Tape Drives ***
The SCSI Interface Conversion Instructions (M8219A) which cover conversion of
the industry standard GCR to SCSI has incorrect switch settings on page 4.
The GPx Series 70 Installation and Maintenance Manual M8226A also has incorrect
switch settings on page 2-44.
Correct switch settings for the CS card (SCSI interface) are:
Switch MH07 (Switch B)
___________________________________________________________
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 |
| off | off | on | on | on | off | on | on | off | off |
|_____|_____|_____|_____|_____|_____|_____|_____|_____|_____|
Switch MH11 (Switch A)
___________________________________________________________
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 10 |
| on | off | on | on | off | off | on | on | off | off |
|_____|_____|_____|_____|_____|_____|_____|_____|_____|_____|
ORIGINATOR: Norm Jones
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB008 Pg001
FIB 00009 11/21/90
*** DO NOT Cycle Power on a GCR Tape Drive with GPx System ON-LINE!! ***
SYMPTOM/TEXT: Powering a GCR Tape Drive off/on which is connected to a GPx70
while the system is on-line can cause a Panic Crash. DO NOT POWER THE GCR OFF
THEN BACK ON or UNPLUG SCSI CABLE while the system is on-line!
FIX: The cause is not known at this time, but suspect it has to do with the GCR
sharing.the same SCSI bus with the Disk Drives.
Anytime the GCR tape drive requires service, the GPx system should be shutdown
properly.
ORIGINATOR: Barry Matthews
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB009 Pg001
FIB 00010 02/22/91
*** GPx70 Filesystem Space Notes & Cautions [ WPSF 653 ] ***
GPx SERIES 70
FILE SYSTEM SPACE NOTES & CAUTIONS WALLY MOORE 12/90 653
TYPE: Informational/Problem
PURPOSE: This field bulletin talks about File System space issues that
might arise on the GPx Series 70.
SYMPTOM: After a filesystem is installed on a partition and before any data
is restored, approximately 12% of the disk is already used.
CAUSE: The filesystem used on the GPx Series 70 (Berkley Fast Filesystem)
reserves 10% of each partition for the superuser space. Normal
users can't use this space so the disk space utilities don't
show it as free. The other approximately 2% is partition
overhead, inode tables and other partition overhead.
The reserved space isn't normally used by the root user. It
consists of unallocated file system blocks just like any other
empty portion of the file system. The effect is that the file
system looks bigger to root. The reason for this is to prevent
the fragmentation of the file system's free space into small bits
and pieces. If no space is reserved and the file system becomes
very full, files which grow or are created will tend to be
allocated in very small pieces on widely separated areas of the
disk. When 10% of the free space is set aside, files are more
likely to be allocated in contiguous, or at least nearly
contiguous, blocks.
NOTE: The 'dfspace -S' command will give you the amount of disk
space including the super-user reserved space.
SOLUTION: NOTE: Reserving at least some space which can be used only by
root can be a safety measure. This allows the system
administrator to have normal system functions while fixing a
system which, to the users, has run out of disk space. In
addition, there is a possibility of a reduction in
performance if the 10% is reduced. MAI development is in the
process of testing with different amounts of reserved space
to see what, if any, the reduction in preformance would be.
Until the results are available the following should be used
with caution.
The amount of space reserved for the super-user on each partition
can be changed with the tunefs command. This command can be used
on any partition, even if there is data on the partition.
NOTE: Even though the data will not be damaged, it is recommend
that the customer has a current backup of the partition
before using the command.
To use the command, first unmount the target partition. Next
use the command:
/etc/tunefs -m X /partition
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB010 Pg001
where X = the percentage of disk to reserve for the super-user.
This can be 0.
/partition = the unmounted partition you are changing the
value on.
Remount the partition
If after lowering the reserved super-user percentage value,
the customer sees a reduction in preformance and wants the
default value restored, use the above command and put 10 for
the percentage of disk to reserve.
NOTE: If the value is raised above the current usage level,
users will be unable to allocate files until enough
files have been deleted to get under the higher
threshold.
WARNING: It is HIGHLY RECOMMENDED that the root partition
super-user reserve size NOT be changed. If the
root partition becomes full, the system can become
unusable.
SYMPTOM: Customer has an add-on 670 MB disk drive, but 'dfspace' shows
the drive as a 635 MB drive.
CAUSE: The 635 MB returned is one Megabyte count, not a million
count. A Megabyte is 2 to 20th, or 1,048,576. This
multiplied by 635 is 665,845,760 bytes. Commonly thought of
as 665 MB. The other 5 MB is taken up with disk and OS over
head, such as diagnostic partition and firmware partition.
SOLUTION: If a customer has a concern over the size of the disk using
the 'dfspace' command the 'prtvtoc' or 'df' command can be
used to show the value of the disk space. The 'prtvtoc'
command shows the disk partitioning information. To use the
command enter:
prtvtoc wdx | pg
where x is the drive number. For the first add-on drive use
'wd1'. This will display all 16 partitions and the number of
512 byte sectors that are being used. For more information
on the command see the 'manpage' for this command.
To use the other command enter:
df -t /partition
where /partition = directory or partition to look at. If
this is not added the command will show all the partitions
on the system.
For more information on this command see its 'manpage'
ORIGINATOR: W. Moore
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB010 Pg002
FIB 00011 11/27/90
*** Disk Overhead ***
Symptom
Customer has an add-on 670 MB disk drive, but 'dfspace' shows the
drive as a 635 MB drive.
Cause
The 635 MB returned is one Megabyte count, not a million count. A
Megabyte is 2 to 20th, or 1,048,576. This multiplied by 635 is
665,845,760 bytes. Commonly thought of as 665 MB. The other 5 MB is
take up with disk and OS over head, such as diagnostic partition and
firmware partition.
Solution
If a customer has a concern over the size of the disk using the
'dfspace' command the 'prtvtoc' or 'df' command can be used to show
the value of the disk space. The 'prtvtoc' command shows the disk
partitioning information. To use the command enter:
prtvtoc wdx | pg
where x is the drive number. For the first add-on drive use 'wd1'.
This will display all 16 partitions and how many 512 byte sectors are
being used. For more information on the command see the 'manpage' for
this command.
To use the other command enter:
df -t /partition
where /partition = directory or partition to look at. If this is
not added the command will show all the partitions on the system.
For more information on this command see its 'manpage', or administrator
guide to unix.
ORIGINATOR: W. Moore
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB011 Pg001
FIB 00012 11/27/90
*** Setform at Boot. ***
Subject : GPx Series 70 Author : Dale Jensen Date : 11/05/90
Issue : Setting Printer Form
During System Load
Type : Informational
Purpose
Several customers have requested having non-standard forms set on
their printers without operator intervention after a system reload.
The following proceedure can be used to accomplish that.
Procedure
A shell program can be created that will set non-standard forms on
specified printers during a system load into a multi user run level.
Its requires two command files to be created by the user. The
first command file will contain one line and will be placed in the
'/etc/rc2.d' directory. The file name MUST start with 'S' and a
two digit number that does not exist within the directory. An
example of the file name would be '/etc/rc2.d/S99prtformchg'. The
file only contains one entry which will start the form change
command file in background mode and will look like this :
/etc/ptrformchg &
The second shell program '/etc/ptrformchg' will test to make sure
the printer despooler process is started and then set the specified
form for that printer. A sample program is shown below :
until
/usr/mbf/bin/lpmaint list=p2 -fc form=newform -go 2> /dev/null
do
echo "working" > /dev/null
done
The above commands should be duplicated for each printer that
requires a form change at boot time.
These shell scripts must be maintained by the user and may require
modifications when printers are added or deleted or form assingments
are changed. Testing of these scripts should be done prior to a
system reload since the '/etc/ptrformchg' script could loop
indefinatly and impact system performance. To test the script,
give it file execute privilages first and then enter the script name
from command mode.
Related Documentation
UNIX System V/386 Release 3.2 Users Guide
ORIGINATOR: D.JENSEN
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB012 Pg001
FIB 00013 11/27/90
*** 525MB Tape Length Determination ***
Type: Informational
Purpose:
This field bulletin gives the proper tape to use with the 525MB tape drive.
Symptom:
Data spans two tapes when the data written to a 525MB tape drive is less than
525MB. Customer is using a 1020 ft tape.
Cause/Solution:
The 525MB tape drive will run as either a 120MB or 525MB tape drive. The
mode the tape drive will run in is determined by the size of the tape being
used. On the tape leader is two holes, the tape drive can tell from the
distance between these two holes the size of the tape. The tape must be a DC
6525 tape or the tape drive default to 120MB mode.
ORIGINATOR: W. Moore
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB013 Pg001
FIB 00014 11/29/90
*** Possible Short on Differential SCSI Adapter ***
Rancho Technology PCBA "SCSI Converter PCBA" shorting problem.
MM597070 / PN 400814-001 (GPx70 to GCR SCSI tape drive)
PURPOSE:
Review of the PCB layout reveals two problem areas which
may result in shorting when mounting hardware is installed.
SYMPTOM:
Unknown. Could cause a wide variety of SCSI bus problems or only affect
GCR operation.
CAUSE:
1) Component side of the PCB near J2 pin 50. A wide
etch along the end of the termination resistors
can possibly short to mounting hardware installed
in the adjacent mounting hole.
To check for this condition, SHUTDOWN the system
and turn system OFF. Remove cable from J2, remove
W2 jumper at the rear bottom of the PCBA. Using an
Ohm meter, check for a short between the TOP pin
and chassis. If this condition exist, also check for
the possiblity of F1 fuse being blown, located near
jumper W2. F1 1amp Pico fuse, MBF PN 137003-001.
2) Solder side of the PCB near J2 pin 2. An etch
runs diagonally near the adjacent mounting hole
which can possibly short to mounting stand-offs
in our chassis.
To check for this condition, SHUTDOWN the system
and turn system OFF. Remove the differental cable
from J2 of the PCBA. Using an Ohm meter, check for
a short berween J2-pin 3 to chassis.
SOLUTION:
Fabricate an insulating washer and install between these
two locations to prevent future problems.
ORIGINATOR: Don Luque
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB014 Pg001
FIB 00015 11/29/90
*** Unable to access GCR Tape Drive. ***
SYMPTOM/TEXT: Inability to access a GCR tape drive connected to a GPx70 system
is usually due to the device files for the GCR not being created properly. Every
time a GPX70 is booted it looks to see if a GCR is attached, if it finds one the
system automatically creates the proper device files (the GCR must be powered on
during.system load). If it does not find a GCR tape drive these files will be
deleted (this will occur if the GCR is powered off during boot time).
To test for the above cause perform an alternate load of the Demon Diagnostics
and run the Tape tests, if you can access the drive with demon your problem
most likely involves the GCR device files. You can also test for the existance
of these files with the system on-line by entering the following command
"# ls -l /dev/rmt |pg", and look for these files:
Major,Minor File Name
. ----------- ---------
68,0 tg0
68,3 tg0h
68,131 tg0hn
68,2 tg0m
68,130 tg0mn
68,128 tg0n
FIX: There are two methods to correct this problem:
1.) Shutdown the GPx70 system, power off the GCR tape drive, then boot the GPx
with the GCR off (this will delete all the GCR device files). Shutdown the
GPx once again, power the GCR tape drive on, then boot the GPX system (this
will re-create the device files from scratch).
WARNING - Never Cycle The Power To The GCR With The GPx On-Line.
2.) login as "root" on the system console and enter the following two commands:
# cd /dev <cr>
# ./MAKEDEV tg0 <cr>
These commands will create the device files for the GCR tape drive
ORIGINATOR: Barry Matthews
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB015 Pg001
FIB 00016 11/30/90
*** SSM Issues ***
TYPE INFORMATIONAL / PROBLEM (DATE: 11/2/90)
PURPOSE
To provide current information on the SSM PCBA's and it's
relationship to system issues.
There are three (3) issues:
. 1) VME Bus hang.
2) REMOTE ENABLE during system loading.
3) Printer data lose or SSM hang.
SYMPTOM (1)
- The VME Bus may hang which will prevent the usage of the 16ways.
This will prevent the usage of the system.
- When the system is turned ON, the real time clock no longer has
the correct time. Intermittent problem.
CAUSE
SSM PCBA, PN 903683 (Old) and PN 903829 (New) not at or above
revision 'B'.
SOLUTION
Obtain revision 'B' SSM.
SYMPTOM (2)
During system load with the key in the REMOTE ENABLE state, the
system will appear hung with no response to 'T0' console except
with the single display of:
........................ vdt display ...............
. !
. 5 seconds to boot, press any key to stop:
. booting SSM from 'wd(0,32)'
. loading 246320 bytes
If the system has been loaded with the REMOTE in the ENABLE state
and the system appears to be hung, to determine if this is the
cause, turn the REMOTE key to DISABLE. A line of print will be
displayed to the screen, but may not be the same verbiage, as follows:
........................ vdt display ......................
. . /ccs/gpx/1.0c/sr/ssm/v1.70/build/scr/common/uart.C:459 ASSERT
. batch gerchr: intr
This problem will occur from a Tape Load or Disk load. The system
. did actually load except that there is no connectivity to 'T0'. The
action of powering the system OFF or using NMI RESET does not allow
proper 'shutdown' of the system.
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB016 Pg001
In some cases:
. 1) Console 'T0' may come up as 'T1' and other terminal will be
active.
2) Depending on system PCB configuration, the system may NOT
load, Run LED is OFF.
CAUSE
The reason for this problem is the difference between the SSM FW
onboard the SSM PCBA (2.0.1) and the SSM Level 2 FW (1.7.6) and
SSM Operational FW (1.7.6) that is provided by the software.
The SSM onboard FW (2.0.1) was implemented by revision 'B' of the
PCBA. This revision is MANDATORY for the 1.1A OS or above, at which
time the SSM Level 2 FW and SSM Operation FW will be compatable with
the 2.0.1 SSM onboard FW.
SOLUTION
During system loads, INSURE that the REMOTE Key switch is in the
DISABLED position until the 1.1A OS is installed.
If you determine that the system has been loaded with the REMOTE
ENABLED, the options are:
1) If the Run LED is OFF, then there are no active process and
the system can be reloaded.
2) If the Run LED is ON, then check to see if there are terminals
that are active. If so, then you can 'shutdown' the system
from that terminal. You must 'cd' to the 'root' (/) and enter
'superuser' (#) mode. Since there will not be a 'T0' console,
you will not know when the system is down in the level 3 (--->)
state. Insure that you wait alittle longer then the 'grace
time' (-g?) before reloading the system.
SYMPTOM (3)
- a) Loss of data will occur on the parallel printer supported by the
SSM (lp) if it is configured as 'Dataproducts', and it is powered
OFF before or during printing of data.
- b) If a Xerox printer is attached to the parallel port supported by
the SSM (lp), the SSM PCBA will 'Hang' with no warning message.
CAUSE
- a) This will occur on both SSM's, PN 903683 (Old) PN 903829 (New).
This is not a hardware issue but a software issue.
- b) This is only an issue with the New SSM PN 903829.
SOLUTION
. - a) Insure that the printer is not powered OFF during printing of
data. There will be a change to the software to correct this issue
in a future release.
- b) Revision 'C' of the New SSM 903829 will correct this issue. If
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB016 Pg002
revision 'C' is not available and this is a problem, then you
may also use the Old SSM 903683 at revision 'B' that does not
have this problem.
ORIGINATOR: Don Luque
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB016 Pg003
FIB 00017 12/24/91
*** Excessive Async Errors Logged [WPSF 757 A1 ] ***
Type : Problem
Purpose
There have been several calls from the field reporting problems and asking
for information regarding async errors. The purpose of this field
bulletin is to explain why some sites see such a large number of async
errors reported and provide some possible solutions.
Symptom
The system console frequently displays messages reporting async port
errors and the system errorlog is constantly logging async errors. The
errors reported are usually async parity or framing errors. A terminal or
printer hang may have also been reported.
Cause
There are several reasons that the number of async errors reported may be
so high. These errors can be induced by faulty cables, cables connected
to the system but not to a device, turning devices off and on or hitting
break key sequences on the key board. When a modem is first connected
several errors may occur. Disconnecting a modem from a phone connection
can also cause a flurry of errors with some modems. Other errors may occur
due to bad phone lines when a modem is connected. A 16-way firmware problem
which multiplies the actual number of errors occurring when they get logged
into the the errorlog has also been discovered.
Solution
Check all cabling to ensure proper shielding and grounding. Disconnect
any cables acting as antennas to the system because they are not connected
to a device. This will greatly reduce the number of errors being logged
in most cases.
Expect certain errors to occur when turning devices off and on or when
the break key sequences are entered. The only solution for these
errors is to avoid doing those operations while the system is running.
At this time there is no way of disabling the logging of async errors,
but this feature is planned for a future release. To reduce the amount of
errors being logged to the console, the errorlog file can be redefined
with a larger amount of records. Currently 10% of the errorlog file is
used for async errors. If the errorlog file is redefined with 10000
records versus 1000, it will then hold 1000 async errors versus only
100. This can be done using the command : 'mbf_errlog -initial s=10000'.
It is recommended to print the errorlog information prior to redefining
it if there are other hardware problems.
For information regarding the firmware correction, see FIB 18 in this FIB
group.
A new ECN, number 14863, has been released for the purpose of updating
the 16-way firmware. This new firmware will change the 16-way revision
level from 'D' to 'E'. Along with this new firmware, the 1.0A*63 Operating
System has been released. The combination of new 16-way firmware and the
1.0A*63 Operating System level update will correct several async problems,
such as port hangs, which are not caused by faulty cabling, port
configurations or the device connected to the 16-way.
NOTE: The 1.1A Operating System introduces a new option to the mbf_errlog
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB017 Pg001
command, -noasync, which turns terminal errorlogging off. Refer to the
1.1A Software Announcement (FIB 28) for details.
ORIGINATOR: Dale Jensen
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB017 Pg002
FIB 00018 04/02/91
*** VME 16-way upgrade rev F [ WPSF 677R ] ***
FM: D.Luque DTD 3-29-91 **REVISED-B**
PRELIMINARY FIELD BULLETIN
GPx70 SERIES 70
VME 16WAY UPGRADE; REV F
TYPE: Problem
PURPOSE: -STOP 2.2.18 UPGRADE/ INSTALL 2.3.1 'NEW'-
REWORK ALL ASSEMBLIES; IMPLEMENTATION CODE = 6
The VME 16way PCB FW is being upgraded to help reduce/eliminate
async GETTY errors. This HW upgrade to the 16way is not tied to an
OS level, but with the FW upgrade and the 1.1A OS, a 100% fix can be
achieved (outside of cabling or peripheral problems causing errors).
SOLUTION: A new ECN 15011 has just been released to upgrade the 16way
FW to 2.3.1, PCB=rev F. FW can be obtained using MM# 597077 (one per 16way)
NOTE: 1) There have only been FW changes to the 16ways, no wire
jumps or etch cuts have been implemented. ALL 16ways
are upgradable to this revision with only the FW change
and insuring that JMP JJ is in the correct position.
2) Jump JJ selects the FW (PROM) size. The rework instruction
location of Pins # 1, 2, 3 are different than the location
shown in the GPx Series 70 Maintenance Manual, yet THEY ARE
BOTH CORRECT ! Each indicates the selection of the 256K
Prom size with the jumper in the same orientation. Reference
Figure #1.
To determine the FW level of the 16way(s) in the system, DEMON
may be used. Select 16way Interaction test; Zero out selections and
then select #1 Function Select. The next screen display will have
"Boards found during probe:"; this will indicate the revision of the
PCB as "board 0 rev: 02.03.01 board 1 rev: 02.03.01" if at rev F.
REWORK INSTRUCTIONS
Rework all assemblies 400736-001 as follows:
1. A wrist strap and antistatic mat must be used during rework.
2. Remove 2.2.3 or 2.2.18 firmware set in locations E1 and E2.
3. Install new firmware set (400736-003) as follows:
a) 180-002-2337A Rev. 2.3.1 goes into location E2.
b) 180-022-2338A Rev. 2.3.1 goes into location E1.
4. INSURE that jumper JJ is jumpered from pins 2 to 3 and
inspect EPROMS for bent pins.
5. Update the revision tag on the PCB extractor tab to
reflect rev. 'F'. Add FW revision label on the solder
side (SS) of the PCB to reflect FW level; same location as
the EPROMS on SS.
** 6. Return old EPROMS in the same IC container the new ones came in.
Use Normal Return Procedures. Send them back in a batch shipment,
do not return individually due to shipping cost.
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB018 Pg001
__
| |
-------------------------------==|__|
PIN #1---> .-- |
|E1| |
|__| |
| PCB FRONT
PIN #1---> .-- |
|E2| | Figure #1
|__| |
. ._. |
ECN JMP JJ ----> 1-2-3 |
MANUAL JMP JJ -> 3-2-1 |
^-^ = 256K
=========================================================================
ORIGINATOR: D. Luque
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB018 Pg002
FIB 00019 03/19/91
*** SCSI TAPE DEVICES MTS,HCC,GCR ***
As of 1.1a the MTS HCC and GCR are supported on GPX 70
All tape drives are suported off of the SCSI differential adapter.
This SCSI buss support up to 7 devices,with up to 4 of them on the
SCSI adapter.
The address of the devices are in decending order.
ID 5 is named /dev/rmt/tg0
ID 4 is named /dev/rmt/tg1
ID 3 is named /dev/rmy/tg2
ID 2 is named /dev/rmt/tg3
ORIGINATOR: JOHN WHALEN
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB019 Pg001
FIB 00020 04/02/91
*** DATAPRODUCTS Printer Problems SSM [ WPSF 692 ] ***
Luque (4/1/91) Preliminary FIELD BULLETIN #nnnn
SUBJECT: GPx SERIES 70
DATAPRODUCTS PRINTER ON SSM
TYPE: Informational/Problem
PURPOSE: Dataproducts printer configured on the old version of the
SSM PCBA (PN 903683), with the 1.0A*63 OS, has a problem
with the DP interface. This is not a problem with the
Centornics printer interface nor the new version of SSM
PCBA PN 903829.
SYMPTOM: Dataproduct printers will exhibit erratic priting, data loss
and garbage print when connected to the 903683 SSM PCBA with
the 1.0A*63 OS.
CAUSE: The SSM Firmware (Operational) is provided via the OS. The
DP interface provided by the SSM Firmware (Operational) has
a problem, version 2.0.6 does have a problem; versions
2.0.0 thru 2.0.6 are suspect.
SOLUTION: The BEST solution for this problem is to upgrade the system
to the 1.1A OS that contains the 2.0.9 SSM Firmware.
DATAPRODUCTS
---------------
OS | 1.0A | 1.1A |
----------------------
903683 | X | OK | X=problem
----------------------
903829 | OK | OK |
----------------------
ORIGINATOR: D. LUQUE
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB020 Pg001
FIB 00021 04/11/91
*** PPC causes possible system crash [ WPSF 693 ] ***
PRELIMINARY FIELD BULLETIN
Sj: GPx Series 70 Author: Dale Jensen Date: 04/08/91
System Crash using PPC
Type: Problem
Purpose:
Inform the field of a problem which can cause a GPx Series 70 system to
crash when there is a Parallel Printer Controller installed.
Symptom:
A system crash with a message 'vme Int Map Class :: vme Int Handlr - vector
255 not mapped' can occur when printing to a printer connected to the PPC
board (P/N 903841-001).
Cause:
This problem occurs when the PPC and a 16Way interrupt the System Services
Module (SSM) at the same time, the PPC can hang the VME interrupt bus.
The GPx Series 70 Installation and Maintenance Manual (M8226) shows the PPC
located to the right of the 16Way Controllers. The PPC board has a priority
of 2 and the 16Way has a priority of 3. When the PPC and 16Way send an
interrupt at the same time, the SSM responds to the 16Way first because of
its higher priority. Due to a design error in the PPC, the PPC does not
pass the acknowledgement on to the 16Way. This causes the SSM to time out
and the system to crash.
Solution :
The solution is to place the PPC at the end of the VME bus, making it the
left most board in the VME card cage (the farthest from the SSM).
A new ECN will soon be released and will correct this problem. An update
to this Field Bulletin will be sent out when it becomes available.
ORIGINATOR: D. JENSEN
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB021 Pg001
FIB 00022 04/15/91
*** UPS Warning Cable Incorrect [ WPSF 691 ] ***
NOTE: THIS IS AN UPDATE TO THIS 'PRELIMINARY FIELD BULLETIN'. CHANGES ARE
PRECEEDED BY AN '*'. ECN 15084 HAS BEEN AMENDED (15084A) IMP 6B.
============================================================================== .
D.Luque PRELIMINARY 'FIELD BULLETIN' #nnn DATE 4/12/91
SUBJECT : GPx SERIES 70
UPS WARNING CABLE INCORRECT
TYPE: Problem
PURPOSE: To inform the field of a mis-wire problem with the 'Warning
cable' used to interface the EXIDE UPS with GPx 70 Automatic
Shutdown Feature. Without a proper 'shutdown' prior to loss
* of power, data loss/corruption can occur. ECN 15084A has been
released to address this issue; IMPLEMENTATION CODE = 6B.
SYMPTOM: When the site loses AC power and the UPS is running on
'battery power', the system does not have the capability to
undergo an automatic shutdown because this condition is not
sensed.
CAUSE: The warning cable, PN 916076-004, was incorrectly wired.
* The '-004' version is now obsolete, rework to '-005'.
SOLUTION: ECN 15084A was released to correct this problem. Rework the
warning cable as follows:
1) Remove pin-13 at the "P2" end (DB-25 connector) with a
pin extractor and insert pin into pin-15 location.
* 2) Update cable revision level to 'F' and re-identify cable
PN to 916076-005.
P2 (DB-25 to UPS) P1 (RAMP Connector to PDU)
____ _
| 15 |--------------------|2|
| 14 |--------------------|1|
---- * 916076-005 -
ORIGINATOR: D. LUQUE
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB022 Pg001
FIB 00023 05/15/91
*** SMD Disk Format Information ***
GPx Series 70 Wally Moore 05/01/91
FORMATTING SMD DRIVE UPGRADES
Type: Informational
Purpose:
When a SMD disk drive is bought over to a GPx Series 70 during an upgrade from a
MPx/AS it must be reformatted on the GPx Series 70 before it can be used. This
is due to the differences in the way the BOSS/VS and BOSS/VX interface with the
drive. Since the drive is being reformated the Manufacturer's Defect data has
to be re-entered. This requires that a non-standard format be used. This field
bulletin provides the steps to recreate a new defect map and format the drive.
NOTE: In all other cases the standard format option will be used.
Procedure
A quick overview of the steps required are as follows:
1. Set the switches on the drive per the service manual and install the
drive in the SMD add-on chassis.
2. Create a Manufacturer's Defect Map File.
3. Run the on-line version of dive.
4. Select the drive to be formated
5. Load the defect map file.
6. Change the format parameter to use Manufacturer's Defect Map
7. Format the drive.
8. Use diskadd to set-up the drive for system use.
NOTE: Steps 1 and 2 can be exchanged, that is the Defect map file can be
created before the drive is installed.
Following is the details of each of the above steps.
1. Switch settings
SW1 SW5
Drive Delay in
8 7 6 5 4 3 2 1 Address 4 3 2 1 Seconds
--------------------------- -----------------
1 1 1 1 0 1 1 1 Disk 0 1 1 1 1 0
1 1 1 1 0 1 1 0 Disk 1 1 1 1 0 1
1 1 1 1 0 1 0 1 Disk 2 1 1 0 1 2
1 1 1 1 0 1 0 0 Disk 3 1 1 0 0 3
1 0 1 1 4
SW2 : All positions OFF 1 0 1 0 5
SW3 : positions 5 and 6 ON, 1 0 0 1 6
all the rest OFF 1 0 0 0 7
SW4 : 2 on, all the rest OFF
NOTE: 1 = OFF, 0 = ON
Read the GPx Series 70 Service manual M8226B for more information on the
switch settings.
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB023 Pg001
2. Creating a Manufacturers Defect Map File
During the format procedure you will be required to supply a file that contains
the Manufacturer's Defects. The Manufacturer's Defect information should be
with the drive on a sheet of paper, if you are missing this information you will
need to contact Manufacturing to get a new copy of the list. To create this
file use the 'vi' editor. The file must contain the following information:
cylinder, head, pos/bytes(bytes from index) and len/bytes(len in bytes). This
information is found on the Manufacturer's Defect sheet. NOTE: The MPx altracks
listings doesn't provide this information. To enter the data go into vi using
the command:
vi /za12345 <RETURN>
Where /za12345 is the name of the file. 12345 can be the drive serial
number.
After entering 'vi' enter 'i' to go into insert mode. Next enter the cyl, head,
pos/bytes and len/bytes information. Use a space to separate the fields. Use
the decimal data from the Defect sheet, not the hex data. The decimal data will
be the information in the ()s. Don't put headings on the fields, just enter the
data. When you have finished editing the file enter '<ESCAPE> :wq' to save and
exit 'vi'. An example of the file is as follows:
MANUFACTURER'S DEFECT LISTING SHEET
NO. CYLINDER HEAD POS/BYTES LEN/BYTES
0001 0003(0003) 13(19) 7653(30291) 0004(0004)
0002 0008(0008) 11(17) 46C0(18112) 0014(0020)
0003 001E(0030) 0F(15) 12DC(04828) 0006(0006)
0004 0026(0038) 16(22) 8F72(36722) 0008(0008)
0005 0037(0055) 13(19) 9F2F(40751) 0004(0004)
$ vi /za25346 <RETURN>
0003 19 30291 4
0008 11 18112 20
0030 15 04828 6
0038 22 36722 8
0055 19 40751 4
Since this is a standard UNIX file it can be created on any system and then
moved to the target system. If this file is created on another system use
'tar' to back it up and restore it on the target system. This file can be
created at any time before the format.
3. RUN On-line DIVE
Formatting the drive requires that the On-line version of DIVE be used. To use
On-line DIVE you must install it from the Diagnostics Product tape using
mbfinstall. NOTE: Running DIVE requires a product key. After DIVE has been
installed on the system, as superuser, 'cd' to '/usr/mbf/diag'. Next enter
'./dive' The DIVE screen will come up. Enter the public key when asked for it.
The DIVE MAIN MENU will then be displayed.
4. Select the drive to be formatted
At the DIVE MAIN MENU select option 1. 'Select a Device'. Since the
program will not be able to read the drive information you will need to
select the type drive as a 'F569'.
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB023 Pg002
5. Load the Defect Map File
After selecting the drive to format, go back to the DIVE MAIN MENU. Next
select option 2 'Format'. From the FORMAT MENU enter option 2
'Non-Standard Format'. From the 'NON-STANDARD FORMAT MENU' select option 2
'Defect List Management'. From the 'DEFECT LIST MENU' select option 5
'Load the Manufacturer's Defect List'. In the 'LOAD THE MANUFACTURER'S
DEFECT LIST MENU' select option 1 'Load the List from a File'. When
prompted for the File with bad block list information enter the complete
path name for the file that you created in step 2.
6. Change the format parameter to use Manufacturer's Defect Map
After returning to the 'NON-STANDARD FORMAT MENU' select option 1 'Edit
Format Parameters'. In the 'EDIT FORMAT PARAMETERS' screen answer 'n' to
the prompt 'Are these entries correct (y/n) ?'. <RETURN> to the option
'Use MFG Defect List' change this to 'y'. Note when this option is changed
to 'y' the option 'Use Bad Block List' will automatically be changed to
'n', since only one of these options can be used.
7. Format the Drive
After returning to the 'NON-STANDARD FORMAT MENU' select option 3 'Begin
Formatting the Disk'. Answer the prompts and the program will then format
the drive using the Manufacturer's Defect Map file you entered.
8. Use diskadd to set-up the drive for system use
After the format of the drive is complete exit the On-line DIVE program.
Next use the utility 'diskadd'. This utility will allow you to set-up the
drive for system use.
After running 'diskadd' the drive will be ready for use with the GPx Series
70 System.
Related Documentation
o GPx Series 70 Installation and Maintenance Manual - M8226B
o BOSS/VX Diagnostics Manual for GPx Series 70 - M0232A
o Software Announcement 'GPx SERIES 70 SYSTEM' - SA175
o manpage for diskadd 'man diskadd'
o On-line diagnostics documentation 'more /usr/mbf/diag/doc/dive.fs'
ORIGINATOR: Wally Moore
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB023 Pg003
FIB 00024 06/19/91
*** Possible system crash using on-line DIVE [ WPSF 719 ] ***
GPx SERIES 70
USING DIVE ONLINE DON LUQUE 6/91 719
TYPE: Problem.
PURPOSE:
Inform the field of a 'crash' condition when online DIVE
is used.
SYMPTOM:
A crash can occur when two or more SMD 8-inch disk drives
(F621) are being formatted online.
CAUSE:
Formatting two or more SMD 8-inch disk drives concurrently
on a 'live' system.
SOLUTION:
Format only one SMD 8-inch disk drive at a time. This
is NOT a problem with the 5 1/4-inch SCSI disk drives.
ORIGINATOR: Don Luque
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB024 Pg001
FIB 00025 07/17/91
*** Exide UPS Information for the GPx70 [ WPSF 720 ] ***
GPx SERIES 70
EXIDE UPS INFORMATION DON LUQUE 7/91 720
TYPE: Informational
PURPOSE: To provide information on the EXIDE UPS that is provided through
MAI SYSTEMS under the Market Codes provided.
o MARKET CODES:
MAI SYSTEMS offers two versions of UPS's, a 3KVA and 5KVA, that
match the power requirements for the GPx 5070 and GPx 6070 series
systems under full load.
UPS MKT SYS SYS POWER UPS POWER REMAINING BTU'S BATTERY
CODE TYPE REQUIRED DELIVERED POWER PER HR BACKUP
_________________________________________________________________
ZA093AA GPx5070 2.0KVA 3.0KVA 1.0KVA 5,124 13 MIN.
ZA057AA GPx6070 3.0KVA 5.0KVA 2.0KVA 8,540 6.5 MIN.
Other Peripheral Ratings:
VDT's in general 0.12KVA n/a n/a 410 n/a
SCSI ECA (5 1/4) 0.6KVA n/a n/a 1,430 n/a
SMD ECA (8-inch) 1.6KVA n/a n/a 3,800 n/a
UR012AA 3KVA/5KVA UPS Interface Cable, Automatic Shutdown Feature
NOTE: Though the GPx6070 base system requires 3.0KVA, you would NOT
want to substitute the 3KVA UPS for the 5KVA UPS. To do
so would not allow for other peripherals required for
operation during a power outage condition. Nor would it
provide the proper receptacle for an ECA add-on.
o REMAINING POWER:
Remaining Power is the power that can still be provided by the
UPS, outside the base system requirement, that may be used by
other peripherals. The peripherals that would use this remaining
power would be those required during a power outage condition;
e.g.
- Console terminal. (5-15R2 receptacle)
- Modems/Mux's. (5-15R2 receptacle)
- Printer. (5-15R2 receptacle, normally)
- Drive Expansion Cabinet Assembly (ECA).
(8-inch; L6-15R receptacle, only on 5KVA UPS.
5-inch; L5-15R2 receptacle, 3KVA and 5KVA UPS.)
NOTE: At no time should the TOTAL 'KVA' required by the equipment
attached to the UPS 'EXCEED' the rated KVA output of the UPS.
o AUTOMATIC SHUTDOWN:
UPS setup for the GPx70 Series systems are provided by EXIDE
service personnel. The UPS needs to be set up for proper 'relay'
configuration to provide proper 'Automatic Shutdown' interface.
The Exide AS/400 option provides the proper relay configuration.
The UPS Interface Cable is required to take advantage of the
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB025 Pg001
'Automatic Shutdown' feature. This cable must be ordered
separately and is installed as follows:
UPS GPx70
---- 916076-005 ------
J2 | P2-------------P1 | J2, Pins 1,2,3
UR012AA
o SHUTDOWN SOFTWARE:
The 'Shutdown Software' comes with the UNIX Base package. It
resides in the '/etc/powerfail' script. With the proper Exide UPS
setup and cable installation, upon detection of powerfail the
system will receive an NMI (non-maskable interrupt) relating to
powerfail, that will invoke the powerfail script. The following
warning message will be broadcasted via 'wall' (write all):
nmiClass::nmiUPS - UPS STATUS CHANGE
Broadcast Message from root (systty) on <sys name/date/time>
Power failure: on standby power - shutdown in 4 minutes.
After 4 minutes another message is broadcasted.
Broadcast Message from root (systty) on <sys name/date/time>
Power Failure: on standby power, shutting down, log off now.
Within the next minute the system will start shutting down the
system. If main power is regained to the UPS prior to the
shutdown, the following message will be broadcasted:
nmiClass::nmiUPS - UPS STATUS CHANGE
Broadcast Message from root (systty) on <sys name/date/time>
Power back on, power fail shutdown canceled.
o RELATED DOCUMENTATION:
Marketing Announcement #1019 US/C "New UPS Products Available"
GPx Series 70 Installation and Maintenance Manual, M8226B
Systems Site Planning Guide, M0174
ORIGINATOR: Don Luque
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB025 Pg002
FIB 00026 07/05/91
*** SCSI Device Termination ***
There have been many errors in various documents regarding SCSI device
terminator orientation, this FIB is intended to give an explanation of how the
termination is accomplished and how to check for proper termination on any
internal (single-ended) SCSI bus. This FIB does NOT apply to differential SCSI
busses such as the external SCSI bus used on the external tape drives connected
to the GPx systems.
The SCSI (Small Computer System Interface) bus must be terminated at both ends
of the bus. The SCSI controller has built in termination, however the other end
of the bus MUST BE TERMINATED at the last device and ONLY THE LAST DEVICE on
the bus. Terminator packs must be removed from any other SCSI devices on the
bus.
The terminators used are either SIP (Single Inline Packs) or DIP (Dual Inline
Packs) which contain 220/330 ohm networks with the 220 ohm resistor tied to +5V
and the 330 ohm resistor tied to ground and where each signal on the bus is
connected into an individual network.
+5V
|
2
2
0
signal ------* +3V inactive state
3
3
0
|
-----
---
-
The signals are negative (ground) active and when properly terminated each
inactive signal should be approximately +3V.
To check for proper termination, disconnect the SCSI bus cable, power up all
SCSI devices and measure the voltage at each SCSI device between pin 1 (ground)
and pin 2 (Data bit 0). The signal voltage on terminating devices should measure
approximately +3V and the voltage at non-terminating devices should measure
approximately ground. Measurements at a device with the terminator packs
reversed will read approximately +2V.
Measurements can be made with a DVM or an oscilloscope. It may be easier to
connect the SCSI bus cable to the device being checked and measure the voltage
on an open bus connector, however you must be sure only one SCSI device is
connected to the cable at a time.
Use the following SCSI cable signal table to verify that all signals are
properly terminated.
(continued on next page)
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB026 Pg001
SCSI Bus Signals
Ground Pins: ALL ODD NUMBERED PINS
Signal Pins: 2 - DB0 SCSI BUS bit 0
4 - DB1 SCSI BUS bit 1
6 - DB2 SCSI BUS bit 2
8 - DB3 SCSI BUS bit 3
10 - DB4 SCSI BUS bit 4
12 - DB5 SCSI BUS bit 5
14 - DB6 SCSI BUS bit 6
16 - DB7 SCSI BUS bit 7
18 - DBP SCSI BUS Parity
20 - OPEN
22 - OPEN
24 - OPEN
26 - +5V Terminator Power (from controller only)
28 - OPEN
30 - OPEN
32 - ATN Attention
34 - OPEN
36 - BSY Busy
38 - ACK Acknowledge
40 - RST Reset
42 - MSG Message
44 - SEL Select
46 - C/D Control/Data
48 - REQ Request
50 - I/O Input/Output
ORIGINATOR: Norm Jones
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB026 Pg002
FIB 00027 10/29/91
*** Potential AC to Frame Ground Short!!! ***
SYMPTOM: Potential AC to frame ground short in the PDU.
PROBLEM DETERMINATION: Inspect the AC input wires at the terminal block in the
PDU, the straight crimp-on lugs used may cause the wires
to be pressed against the CPU's frame which may in time
wear through the insulation due to vibration.
To inspect:
1. Open the front door of the system
2. Using a flashlight, observe the AC input wires
through the slot in the PDU. The wires should NOT be
touching the CPU frame.
FIX: If the AC input wires are pressed against the frame:
I. Remove the PDU.
A. cut off the straight wire lugs.
B. crimp on 90 degree lugs.
C. reconnect to the terminal block.
OR
A. form the lugs 90 degrees out from the terminal block.
NOTE: The lugs may break when bent, be sure to have some spare lugs,
preferrably 90 degree lugs, before bending.
II. Reinstall the PDU and observe the AC wiring again.
NOTE: Systems should be checked for this condition on the next PM.
It would be a good idea to have a crimp tool and some 12 guage 90
degree wire lugs on-hand.
AC INPUT WIRES AS VIEWED THROUGH SLOT IN THE PDU
BAD GOOD
______ ______
| || | ||
| TB |-S | TB |-S
|______|| |______||*** <-- wire clear of
* frame using
* <-- wire forced against frame 90 degree lug
--------------- ---------------
| FRAME | FRAME
ORIGINATOR: Norm Jones
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB027 Pg001
FIB 00028 MAI COMPANY CONFIDENTIAL 04/17/92
*** ERR "No Commands Match 247" During Boot or DEMON PROCESSOR Tests ***
SYMPTOM: Message "no commands match 247" during system boot or during DEMON
processor tests.
PROBLEM DETERMINATION: Turn off power to the internal modem.
FIX: Insure the REMOTE key switch is in the DISABLED position and leave the
internal modem powered off unless remote diagnostics are being run.
ORIGINATOR: Norm Jones/Don Luque
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB028 Pg001
FIB 00029 MAI COMPANY CONFIDENTIAL 05/04/92
*** ADDING MEMORY ***
***************** PRELIMINARY FIELD BULLETIN ********* PART 1 ****************
GPx 70 Series Don Luque 03/nn/92 n
Adding Memory
Type: Informational; Model 4020 (GPx5070) and 4028 (GPx6070).
Purpose
Provide sufficient information to install the Memory PCBA and
explain other requirements when memory is added, i.e., DCT,
System Configuration Record (SCR), Swap Space, +5 VDC Converter
module.
Procedure
First, try to understand WHY the memory is being added. Is it
because they were getting messages indicating system resources
have been reached (process killed due to lack of dynamic swap
space or buffer allocation messages) ? Is it due to them getting
ready to add more processes (users) ? Or is it 'just because' ?
Knowing this before hand will help you determine what other actions
are required besides adding the hardware.
o MEMORY
There are two types of memory that come in two different versions:
- Memory Controller Board's (MCB)
8MB MCB PN 903689-001, MM 597025
16MB MCB PN 903689-002, MM 597030
- Memory Expansion Board's (MEB)
8MB MEB PN 903691-001, MM 597035
16MB MEB PN 903691-002, MM 597040
- Memory Expansion Board Cables
Cable, 50 Pin, Memory (two required) PN 916685-001, MM 597275
Cable, 40 Pin, Memory (one required) PN 916551-004, MM 597280
There are no jumpers or switch settings on either the MCB's or the
MEB's. A MEB requires an MCB for operations by; they are cabled
to each other. A system could only have MCB's installed but all MEB's
will always require a MCB to cable to. MCB's and MEB's are installed
into the system CPU chassis, in the front of the system.
NOTE: The 'Board Stiffener' at the rear edge of the MCB is closer
to the rear edge than the MEB, this difference can cause a problem
depending upon the system model and slot location.
o PRESITE SURVEY
Determine the model of the system, the Memory version(s) already
installed and what slot the New Memory will go into. Determine if the
correct Memory PCBA (MCB/MEB) has been ordered, remember that a MEB
cannot be installed without a MCB to cable up to.
o SYSTEM CONFIGURATION RECORD (SCR)
The SCR can be installed prior to the HW. This will not impact the
existing configuration. To install the SCR, all users must be off
the system.
- Insure that the users are off the system and perform a shutdown:
# shutdown -i0 -g0 -y
- At level 3 (--->) boot onto the miniroot:
---> bh 'wd(0,0)unix -M -r wd0s8 -s wd0s8'
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB029 Pg001
- Go into the System Maintenance mode by entering the root password.
- Go into the install options:
# ./INSTALL
- Select option 'I - Configuration Record'.
- After the SCR has been installed, use the 'E - EXIT' option. At
the '#' enter the following:
# sync (this will flush the buffers to disk)
# uadmin 2 (this will bring the system to level 3; --->)
---> bh doOs (this will reload the system without going back
to the miniroot).
o MEMORY INSTALLATION
Memory will always go to the right of other system PCBA's in the System
Bus Card Cage.
If the Memory is going into a 4020, the rightmost slot in the System
Bus Card Cage has the Control Panel Cable Connector next to it. This
connector WILL cause interference if a MCB is going to be added and this
is the only slot open. A MEB must be used and cabled to a MCB. If
there is already a MCB with a MEB attached to it and there is going to
be another Memory added, then the Memory will need to be rearranged so
the last two slots to the right contain the MCB/MEB set. A MCB will
need to be used for the upgrade and installed to the left of the
existing Memory set.
If the Memory is going into a 4028, the last two slots in the System
Bus Card Cage only provides power. The MEB's do not interface to the
System Bus, they interface to the system through the MCB. Therefore
only MEB's can go into these last two locations.
o DCT VALUES
Understanding WHY the memory is being added will determine what
action to take with the DCT table. If there is going to be be a
change to the DCT table, INSURE the the original DCT value is saved
prior to making changes. The DCT tables are under the '/etc' directory
o DCT (cont.)
- If memory is being added to the system in preparation for adding
more users/tasks, then the DCT table should be changed to match
the one that reflects the value of total memory.
This would also be the case if the system was displaying system
resource messages referring to system table overflows.
- If the system was displaying system resource errors such as
'lack of dynamic swap space', then just installing the memory
and letting it go to the users should be all that's needed.
There may be the chance that increasing the 'swap area' will also
cure this kind of error BUT, to much swapping will decrease the
performance. Excessive disk access (swapping) will decrease
system performance. See SWAP AREA.
This would also hold true if there appears to be a performance problem
where screen refresh displays are slow (just adding the memory).
NOTE: Installing memory for 'system resource errors' may not be
necessary. There may already be enough memory in the system
where 'tuning' the DCT values will cure the problems. Checking
the 'sar' values and tuning the DCT table should always be the
first step prior to suggesting more memory. This action should
be the responsibility of the SYSTEM ADMINISTRATOR. The system
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB029 Pg002
is not broken, it just needs taken care of.
o SWAP AREA
The 'swap area' must be large enough to hold all of the system main
memory, it can be displayed using the command 'swap -l'. The swap
area size is displayed in 512-byte blocks. If the swap area is large
enough to hold the system main memory, most likely it will not need to .
increased.
If the swap area requires expansion because it will not contain the
total system main memory, it should be done before the memory is
installed. The SYSTEM ADMINISTRATOR should perform this task. It
will require backing up the partition(s) that will be effected.
o +5 VDC CONVERTER
The Power Supply Control Module should be monitored after installing
the memory. Insure that there are no 'Out of Tolerance' LED's ON.
Use the outline in the GPx Series 70 Installation and Maintenance
Manual M8226, section 4.3.3. If the +5 VDC cannot be adjusted to it's
value, then another +5 VDC Converter will be required.
ORIGINATOR: D LUQUE
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB029 Pg003
FIB 00030 MAI COMPANY CONFIDENTIAL 05/28/92
*** SSM,VME I/O SPACE CORRUPTION ***
*************************PRELIMINARY FIELD BULLETIN**************************
Subject: GPx 70 - Patch 57 Author Date
Corrects SSM/VME I/O Dale Jensen 5/26/92
Space Corruption
Type: Problem
Purpose
This field bulletin is to inform the field about the availability of Patch 57.
Symptom
Analysis of several system crashes have shown that the memory in the SSM,
which is used for I/O by controllers on the VME bus, has been corrupted. The
crash will indicate a VME BUS ERROR on the system console or in the 'dmesg'
display after the crash.
Solution
Patch 57 has been released and should be installed on all systems experiencing
VME BUS ERROR crashes. This patch corrects one of potential causes for this
system crash but there are other conditions, such as failing hardware,
that can also cause a VME BUS ERROR and will not be corrected by this patch.
This patch only applies to systems that have more than one type of controller
in the VME Bus card cage (such as PPC & 16way or 16way and LAN controller).
This patch is intended only for systems with the operating system level 1.1A
installed. This problem has been corrected in operating system release level
1.2A.
The installation instructions are provided with the patch tape and are :
1. 'shutdown' the system to single user mode
2. cd /
3. rm unix
4. mt -f /dev/rmt/tm0 noret rew
For sites with TCP installed :
5. tar xvf /dev/rmt/tm0 unix.tcp.920420
6. ln unix.tcp.920420 unix
for sites without TCP installed :
5. tar xvf /dev/rmt/tm0 unix.920420
6. ln unix.920420 unix
7. sync
8. init 6
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB030 Pg001
A new version of the unix kernel will then be installed and the system can then
be used for normal operations.
Patch 57 should be ordered through software distribution.
ORIGINATOR: D JENSEN
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB030 Pg002
FIB 00031 MAI COMPANY CONFIDENTIAL 08/06/92
*** LEVEL 2 CONFIGURATION WITH TERMINAL SERVERS ***
When TERMINAL SERVERS are installed on GPX 70 DO NOT modify
any ports using configuration.
The ports on the servers will be locked out as soon as the operator sign
off. all server diagnostics will work but you won't be able to sign
on.
if this happens you have to do a manual shutdown command
i.e shutdown -i6 -g0 -y.
ORIGINATOR: J WHALEN
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB031 Pg001
FIB 00032 MAI COMPANY CONFIDENTIAL 08/18/92
*** Installing 16 Ways, Check NPROCs in DCT ***
***************** PRELIMINARY FIELD BULLETIN *************************
GPx70 Series Systems
Installing 16 Way's vs DCT.
Type: Problem
Problem: Running out of available system resources.
Symptom: System resource messages "WARNING: proc: table is full".
Cause: Added 16 ways most likely means that there will be more concurrent
users on the system. This will require more available system
resources to handle the extra system load.
Solution: The DCT (Dynamic Configuration Table) allocates the system resources.
The DCT has a variable that represents active process called
'NPROC'. The NPROC value can be changed by using the 'mkdct'
command.
Prior to installing any 16 ways, 'sar -v' should be run on a daily
basis espically during hours of high system activity. The sar -v
command is used to determine if system tables are over or under
configured. The column under the heading 'proc-sz' will display
the amount of processes (proc) in use and the current size (sz)
of NPROC in the DCT. The 'proc' value should always stay below
the 'sz' value. The 'ov' heading indicates how much the 'proc'
value went over (ov) the declared size.
One 16 way port/task will take 2 to 5 proc's depending upon it's
use. If 'sar -v' indicates there are plenty of proc's not in use
during high activity times (sz - proc = not used), then the added 16
way may not require the need to up the NPORC value.
NOTE: The NPORC value should stay within a realistic value. This
value can only be determined by running sar. To high a value
can cause system performance problems and to low a value will
cause system warning messages. There should always be a
padding to this value, do not run on the edge because a
task/port that only took 2 proc's yesterday may take 5 later
depending upon the job being run. 16 way ports are only one
device that uses proc's, there are other devices that require
proc's too.
Reference the manpages for 'sar' and 'mkdct'. Reference the 'GPx
Series 70 Installation and Maintenance Manual' for HW installation
of the 16 way.
ORIGINATOR: Don Luque
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB032 Pg001
FIB 00033 MAI COMPANY CONFIDENTIAL 01/14/93
*** DCT changes for systems running OpenBasic [ WPSF 805 ] ***
PURPOSE:
To inform the field that GPx 70 Systems which have converted to OpenBASIC
will get a performance improvement by changing the fixbufpct, fixbinode,
fixbdata, and fixbindblk paramenters to zero in the dct table.
The default settings are: fixbufpct 40
fixbinode 0
fixbdata 0
fixbindblk 1
With these settings 40% of the buffer cache is set aside for small (1K) I/O
requests, most buffer requests are 8K in size. By default, inodes and file
data are read into 8K buffers, while b-tree and indirect blocks are read and
written in units of 1K. When using OpenBASIC, the BFS is running as a user
level application, and all its I/O appears to the system as data requests.
In this case, the 1K section of the buffer cache is vastly underutilized.
PROCEDURE
Log on to the system as 'root', and execute the following steps:
# cd /etc
# mkdct -r -o dct -f dct.asc
Use the 'vi' editor to locate the fixbufpct, fixbinode, fixbdata, and
fixbindblk parameters, set their value to 0 (zero).
Create an updated dct file by typing:
# mkdct -o dct -f dct.asc
These changes will not take effect until all users log off and the system
is rebooted.
NOTE: PERFORM THIS PROCEDURE ONLY ON GPx 70 SYSTEMS RUNNING OpenBASIC
EXCLUSIVELY.
RELATED DOCUMENTATION
BOSS/VX System Administrator's Guide (M0178)
Online 'man' page for mkdct
ORIGINATOR: Chris Adkins
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB033 Pg001
FIB 00034 MAI COMPANY CONFIDENTIAL 02/01/93
*** Memory bank chart for analyzing ECC errors ***
Memory ECC errors are reported on the console by memory banks (ex. bank 5).
Memory is organized in 4 MB banks. The following chart relates bank number to
memory address range:
BANK # MEMORY ADDRESS
------ --------------
0 ...........00 - 04 MEG
1 ...........04 - 08 MEG
2 ...........08 - 12 MEG
3 ...........12 - 16 MEG
4 ...........16 - 20 MEG
5 ...........20 - 24 MEG
6 ...........24 - 28 MEG
7 ...........28 - 32 MEG
8 ...........32 - 36 MEG
9 ...........36 - 40 MEG
10 ...........40 - 44 MEG
11 ...........44 - 48 MEG
12 ...........48 - 52 MEG
13 ...........52 - 56 MEG
14 ...........56 - 60 MEG
15 ...........60 - 64 MEG
16 ...........64 - 68 MEG
17 ...........68 - 72 MEG
18 ...........72 - 76 MEG
19 ...........76 - 80 MEG
20 ...........80 - 84 MEG
21 ...........84 - 88 MEG
22 ...........88 - 92 MEG
23 ...........92 - 96 MEG
ORIGINATOR: Norm Jones
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB034 Pg001
FIB 00035 MAI COMPANY CONFIDENTIAL 04/21/93
*** Screen mesh above air filter clogged ***
SYMPTOM/TEXT: There is a screen mesh attached to the chassis of the GPX70
that will collect dust and prevent adequate cooling of the power converters
and all PCBA'S in the system. The removable filter does not seem to stop the
dust from collecting on the wire mesh.
FIX: The CPU chassis will have to be tilted on its side, to vacuum the screen
mesh. Be sure to have a helper when tilting the system.
Remove the removable filter prior to vacuuming, and inspection.
ORIGINATOR: Allan Mollyhorn
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB035 Pg001
FIB 00036 MAI COMPANY CONFIDENTIAL 03/23/94
*** Fan modification to replace unreliable 12V fans ***
The 12V cooling fans (P/N MM597121 105 cfm) have proved to be unreliable and do
not provide sufficient air flow when working properly.
The 115V 130 cfm fans P/N MM895030 used in the low boy MPx and Advanced Series
systems are the same size and provide a significant improvement in cooling with
the added advantage of greatly improved reliability. There is no 115V AC
distribution provided in the GPx70 cabinet but it is fairly easy to make one for
the AC fans. The GPx6070 series (240V) do not have any 115V available in the PDU
but a seperate distribution with a seperate 115V AC cord can be made easily
using a power strip with circuit breaker and switch.
This change will take about two hours.
GPx5070 (115V power source):
Parts required:
5 - AC fans P/N MM895030
2 - AC line cords (or comparable wire)
1. Remove the existing 12V fans (both in the fan bay and the rear of the main
chassis)
2. Replace with 115V AC fans P/N MM895030
3. Remove the 12V wires from the backplane power terminals
4. Splice a line cord into the fan supply wires and connect to the circuit
breaker in the PDU. Two line cords will be required, one for the fan bay
and one for the fans on the main chassis.
......................................................................
GPx6070 (240V power source):
Parts required:
8 - AC fans P/N MM895030
2 - AC line cords (or comparable wire)
1 - Power strip with circuit breaker and switch
1. Remove the existing 12V fans (both in the fan bay and the rear of the main
chassis)
2. Replace with 115V AC fans P/N MM895030
3. Remove the 12V wires from the backplane power terminals
4. Secure the power strip in the cable bay at the lower rear of the main
chassis using cable ties and connect to an appropriate AC source (from the
UPS if at all possible).
5. Splice a line cord into the fan supply wires and connect to the power strip.
Two line cords will be required, one for the fan bay and one for the fans
on the main chassis.
ORIGINATOR: Norm Jones
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB036 Pg001
FIB 00037 MAI COMPANY CONFIDENTIAL 04/05/94
*** Release of 1.2A*08 for GPx 70, Enhancements/Fixes ***
There is a new release for the GPx 70. It is 1.2A*08. This release will be
sent to all of our GPx70 customers that have service. To upgrade the system
with this release use the 1.2A upgrade instructions. Following is a list of
fixes/enhancements:
1. A dct file was created for a 128MB system.
2. A problem was found that if a printer was set-up with despool wait, the
despooler would tie up a cpu while the printer was open. This was only seen
using OpenBASIC. This problem has been fixed.
3. A problem was found were keyed files would get corruption when using rfa.
This problem was been fixed.
4. The following disk drives are now supported: Seagate ST11200, Seagate
ST11900, and Fujitsu F2694.
5. There is now support for the Exabyte 8500 8MM tape with compression.
6. Directory cache sizing is now looked at in the dct file. It was ignored
before.
7. A fix was put in to correct a directory lookup problem that would cause a
system hang.
8. A warning message will now be displayed on the console if a streams buffer
or message buffer overflows. After this warning is displayed the problem
needs to be corrected as soon as possible to avoid a system hang.
9. The 'sync' command was improved to provide better integrity of the files.
Again please upgrade your customer's systems as soon as possible.
ORIGINATOR: Wally Moore
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB037 Pg001
FIB 00038 MAI COMPANY CONFIDENTIAL 04/28/94
*** Panics, System Hangs and external tape drive errors ***
SYMPTOM: Panics, system hangs, external tape drive errors on systems with
external SCSI devices.
PROBLEM DETERMINATION: Check for proper latching of the external SCSI cables
to the devices. The spring clip should seat tightly and
securely in the cable connector housing.
Some external SCSI cable connector housings have a
problem with interference between the plastic connector
housing and the metal hood, the metal hood has a notch in
it which must be centered in the groove in the plastic
housing to allow the clip to seat fully.
FIX: If the above condition exists, use jointed pliars to carefully form the
metal hood, centering the notch across the groove in the plastic
housing.
NOTE: For a secure connection use long cable ties to wrap around the clips and
connector, this will prevent accidental disconnection.
These cables are very heavy, use cable ties to support the weight of the
cables, tying to available spots (fan grille, cover slit, stick on tie,
etc.).
SCSI Cable Connector
Side view
_________
| __
\___/ | The notches should be clear with no
Interference ________/ | interference.
___ \ |
/ \__|
|_________
ORIGINATOR: Norm Jones
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB038 Pg001
FIB 00039 MAI COMPANY CONFIDENTIAL 07/14/94
*** Extra line feeds, garbled characters, duplicated lines, disk errs ***
SYMPTOM: 1) Strange serial device (VDTs & printers) failures:
extra line feeds, garbled characters, duplicated lines of data
2) strange disk errors
FIX: Replace the memory controller(s).
NOTE: In most of the known cases where these failures have occurred the failure
was caused by the first memory controller.
The reason for a memory controller to cause such odd failures is unknown
at this time. If you find a memory controller causing odd symptoms please
notify National Support and forward the PCBA to engineering for
evaluation.
ORIGINATOR: Norm Jones
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB039 Pg001
FIB 00040 MAI COMPANY CONFIDENTIAL 11/18/94
*** SSM battery can be replaced - P/N YY030330 ***
Subject : SSM BATTERY FOR GPx70
A part number has been assigned for the battery on the SSM for the GPx70. It
is YY030330 (183008-001) and can be ordered.
ORIGINATOR: Wally Moore
CPU-BASIC FOUR-4020,4120,4128-----------------------FIB040 Pg001