NetWare Server Driver Installation (Revised 3/5/98)
Copyright 1998 Compaq Computer Corp. All rights reserved.
-----------------------------------------------------------------
-----------------------------------------------------------------
The CPQNF3.LAN driver is a high-performance, NetWare v4.x-
compliant server driver that meets the Open Data Link Interface
(ODI) v3.31 specification. It can be used in all of these
environments:
- NetWare v4.x servers (uses NBI.NLM)
- NetWare v3.12 servers (uses NBI31X.NLM and MSM31X.NLM)
- Servers running SFT III in NetWare v4.1 (as an IPX link, not
as an MSL link)
This NIC supports PCI Hot Plug technology. To take advantage of
this feature, you must use the NIC in a system that supports PCI
Hot Plug, such as Compaq's Proliant 6500 or 7000 series, or in a
non-Compaq system that supports PCI Hot Plug. To implement PCI
Hot Plug, use the configuration instructions that come with your
system. You can find documentation for configuring and installing
PCI Hot Plug devices in Compaq systems in Compaq's SmartStart
application, which ships with Compaq systems. If you use a non-Compaq
system that supports Hot Plug, refer to the system's documentation
or contact tech support for the non-Compaq vendor.
How to Install the Driver
Install the driver and its support files as appropriate for your
environment:
- For NetWare v4.x servers, install from the server using
NetWare's INSTALL.NLM.
- For NetWare v4.x (optional) and v3.1x (required) servers,
install from a workstation by copying files manually to your
server.
TIP: This procedure requires that you have at least one
network controller and driver operating to enable the file
copying. If your server has only one controller and it is
the controller you plan to upgrade, it is easiest to perform
this file copying from a workstation before you power down the
server to upgrade your hardware.
See also:
- Help with LOAD and BIND Commands
- LAN Driver Statistics
- Custom Statistics
- Driver Error Codes
- Advanced Network Fault Detection and Correction
-----------------------------------------------------------------
Installing with the NetWare v4.x INSTALL Utility
This section explains how to load the controller driver using the
NetWare v4.x INSTALL.NLM utility. For procedures not explained
here, refer to the NetWare INSTALL help or your Novell
documentation.
NOTE: Before you deselect your current driver, copy or make
note of the LOAD/BIND statements from your existing
installation; you may want to use the net addresses, names,
frame types, etc., with the new module.
1. Load INSTALL.NLM from the server console.
2. Edit the STARTUP.NCF file, and edit the statements as
recommended below. (For NetWare v4.x servers running SFT
III, you must take the steps shown at each server's IO
Engine. Refer to your Novell documentation for more
information.)
SET MINIMUM PACKET RECEIVE BUFFERS=100 (or more)
SET MAXIMUM PACKET RECEIVE BUFFERS=200 (or more)
SET MAXIMUM PHYSICAL RECEIVE PACKET SIZE=1514 (or greater)
NOTE: If installing the driver on the Netelligent Dual
10/100 TX PCI Controller and when the Orion PCI Bridge
chipset is present on the system, use a packet size of
1545 for increased performance. See also the section
titled "Maximum Physical Receive Packet Size" later in
this file.
These statements let you optimize performance, provided that
you have sufficient server memory. See the section Edit
STARTUP.NCF for details.
3. Select Driver options> Configure network drivers. If you
have an existing driver, first choose Deselect a driver to
remove it, and click Continue. Choose Select a driver. Press
<Ins> to load an unlisted driver. Then press <F3> to specify
the driver's location.
4. When prompted, insert your copy of the Compaq controller
diskette into the server's floppy drive and the enter the
full path to CPQNF3.LAN, such as A:\NWSERVER.
5. When a menu displays showing CPQNF3.LAN controllers, select
your Compaq controller and press <Enter>. INSTALL copies
the driver and its support files (MSM.NLM,ETHERTSM.NLM,
NBI.NLM, CPQNF3.LDI) and prompts you to save copies of any
existing files before they are overwritten.
TIP: If you also have a token ring or FDDI controller in
the server, Compaq has provided compatible versions of
TOKENTSM.NLM and FDDITSM.NLM on the controller diskette.
6. The next menu shows the parameters that can be used with the
CPQNF3.LAN driver. Select a parameter and press <Enter> to
view the list of supported options.
Help for the parameter is displayed in the lower text box.
For parameter descriptions and explanations, see the section
LOAD Command and Parameters.
NOTE: Enter the slot number that is correct for your
machine. If needed, you can press <Alt>+<Esc> to go to the
server console and load CPQNF3 with no parameters (LOAD
A:\NWSERVER\CPQNF3.LAN) to see the slot number displayed.
Press <Esc> to cancel and <Alt>+<Esc> to return to INSTALL.
7. After you have made selections for all controller driver
parameters, press <F10> to save them and load CPQNF3.LAN.
8. Repeat these steps to load and bind additional ports or
Compaq controllers using this driver. Then follow the
prompts to complete the server installation and exit
NetWare.
9. If you changed the STARTUP.NCF file or if you are upgrading
the hardware at this point, power down the server, make any
hardware changes, and restart the server.
10. Restart NetWare.
See also
- Using the BIND Command
- Loading TCPIP.NLM
- Example LOAD and BIND Commands
- UNLOAD and UNBIND Commands
-----------------------------------------------------------------
Installing Manually from a Workstation
This section explains how to install the driver and its support
files on a file server running either NetWare v3.1x or v4.x.
These are the general installation steps:
1. Remove existing drivers, if any.
2. Edit AUTOEXEC.NCF (LOAD and BIND statements) and STARTUP.NCF
SET statements).
3. Determine if modules need to be updated.
4. Load and bind the new driver and modules so you can connect
to the server from a workstation.
5. Copy the support modules and driver to file server.
Detailed instructions follow. After these instructions, you can
find these topics to help you structure your commands:
- LOAD Command and Parameters
- Using the BIND Command
- Loading TCPIP.NLM
- Example LOAD and BIND Commands
- UNLOAD and UNBIND Commands
-----------------------------------------------------------------
Edit STARTUP.NCF
Optimizing Performance
For best performance, increase the number of packet receive
buffers as your server's memory allows. The default minimum
number of buffers is 50 (in v4.x). Load the INSTALL.NLM to add
the following commands to the server's STARTUP.NCF:
SET MINIMUM PACKET RECEIVE BUFFERS=100 (or more)
SET MAXIMUM PACKET RECEIVE BUFFERS=200 (or more)
(Adjust both parameters; these values define an upper and
lower range.)
If the number of users simultaneously accessing the server
exceeds 100, add one more buffer for each user beyond 100. When
using Packet Burst, configure at least 4 buffers for each user.
MAXIMUM PHYSICAL RECEIVE PACKET SIZE
On a NetWare server with a Pentium Pro processor, a Netelligent
Dual 10/100 TX PCI UTP controller, and the Intel Orion chipset,
add the following line to STARTUP.NCF:
SET MAXIMUM PHYSICAL RECEIVE PACKET SIZE=1545
NOTE: When the driver is loaded, it checks whether the server has
a Pentium Pro processor and the Intel Orion chipset. If so, it
displays a message prompting you to increase the MAXIMUM PHYSICAL
RECEIVE PACKET SIZE to this value.
Supporting High Network Usage
Even if you do not increase the minimum number of buffers, you
may need to adjust the maximum number of buffers if the server
experiences high network use. The default maximum number of
buffers is 100. If the number of buffers in use approaches 100,
you can change the value from the NetWare server console using
this command:
SET MAXIMUM PACKET RECEIVE BUFFERS=200 (specify your new value)
Use the INSTALL.NLM to add the above command to the server's
STARTUP.NCF or AUTOEXEC.NCF file so that the new value will be
set each time the server is restarted.
Monitoring Statistics to Check Settings
To examine statistics for a particular LAN controller, load
NetWare's MONITOR.NLM and select Available Options> LAN
Information.
To ensure that the number of buffers is properly set, load
MONITOR.NLM to watch the No ECB available count statistic; if
that count is increasing, adjust the minimum number of buffers
appropriately.
NOTE: Be sure to adjust the maximum number of buffers
whenever you change the minimum; these values define an
upper and lower range.
-----------------------------------------------------------------
Deciding if Modules Need to be Replaced
The controller diskette contains the driver file, support files,
and NetWare Loadable Modules (NLMs) required for all NetWare v4.x-
compliant server drivers. You must use these NLMs, or more recent
versions, with CPQNF3.LAN. You can download the most recent files
through Novell's Web Site (http://support.novell.com/search);
however, keep copies of your current versions as a fall-back.
IMPORTANT: If your server also supports a non-Ethernet MAC-
level protocol, update your topology support module when you
update the Novell media support module (MSM*.NLM). For those
whose server has a token ring or FDDI controller, we have
provided the compatible versions of TOKENTSM.NLM and
FDDITSM.NLM on the controller diskette. For compatibility
issues with these modules, contact your controller hardware
manufacturer.
The table below shows the names and locations of the support
modules provided on the controller diskette:
Location of Support Modules
NLM Directory
NetWare v4.x MSM \NWSERVER
NBI
ETHERTSM
Topology TOKENTSM \NWSERVER
updates, if FDDITSM
needed:
NetWare v3.12 MSM31X \NWSERVER\NW31XSUP
NBI31X
MONITOR
CAUTION: Using earlier versions of MSM, MSM31X, and
ETHERTSM with the most recent version of CPQNF3.LAN will
prevent the driver from loading.
Finding Your Current Module Versions:
1. At the server command prompt, type MODULES and press
<Enter>.
2. The resulting display shows what drivers and modules are
currently running on the server. Locate the entries for
MSM.NLM and ETHERTSM.NLM. (MSM31X.NLM appears as MSM.NLM.)
3. If the MSM and ETHERTSM modules are not currently loaded,
type LOAD MSM and LOAD ETHERTSM to load the modules in order
to check their versions.
If any of the currently running versions is earlier than those
listed in the table above, you should replace modules
MSM31X.NLM, MSM.NLM or ETHERTSM.NLM.
-----------------------------------------------------------------
Loading Files from the Server
In order to copy the new support modules and driver to the file
server from a workstation, you may need to use the steps below to
unload the old files and load the new ones:
1. Using the UNLOAD command from the console command prompt,
unload any existing server drivers that depend on the
support modules you need to replace. This command completely
unloads the drivers from memory and terminates communication
with currently-attached network users.
UNLOAD [driver-name]
2. Unload the support modules by entering the commands in this
order:
UNLOAD ETHERTSM
UNLOAD MSM (or MSM31X)
UNLOAD NBI (or NBI31X)
3. Load the support modules from the controller diskette. Type
the following commands:
v4.x:
LOAD A:\NWSERVER\NBI
LOAD A:\NWSERVER\MSM
LOAD A:\NWSERVER\ETHERTSM
v3.12:
LOAD A:\NWSERVER\NW31XSUP\NBI31X
LOAD A:\NWSERVER\NW31XSUP\MSM31X
LOAD A:\NWSERVER\ETHERTSM
4. Load the controller driver, adding any parameters you need:
LOAD A:\NWSERVER\CPQNF3
5. Bind the driver to a protocol.
-----------------------------------------------------------------
Copying Support Modules and Driver to File Server
Follow this procedure to copy support modules and the controller
driver to the file server. This ensures that the module versions
provided will load whenever you load the driver.
TIP: This procedure requires that you have at least one
network controller and driver operating to enable the file
copying. If your server has only one controller, and it is
the controller you plan to upgrade, it is easiest to perform
this file copying from a workstation before you power down the
server to upgrade your hardware.
Requirements
- At least one server driver must be loaded and bound to a
protocol.
- The SYS volume on the file server must be mounted.
- You must have sufficient rights to copy files to the
SYS:SYSTEM directory.
Copying the files to the server:
1. Locate a workstation with a floppy drive that is also
capable of logging into the file server to which you will
copy the NLMs.
2. Log in to the file server.
3. Locate the files you will be replacing in the SYS:SYSTEM
directory, and change any read-only attributes to read/write
(to allow them to be overwritten).
4. Insert the controller diskette containing the support files
and enter the following command(s) (assuming that your
floppy drive is A and the volume SYS is mapped to F):
v4.x:
COPY A:\NWSERVER\*.* F:\SYSTEM
v3.12:
COPY A:\NWSERVER\CPQNF3.LAN F:\SYSTEM
COPY A:\NWSERVER\NW31XSUP\*.* F:\SYSTEM
-----------------------------------------------------------------
LOAD Command and Parameters
The LOAD command loads the driver on the server. You can enter
the LOAD command from the server's console command prompt, or you
can include it in your AUTOEXEC.NCF file, which will
automatically load the driver when you start SERVER.
This is the format of the command:
LOAD (path)\CPQNF3 [parameter-list]
where (path) is the full pathname where CPQNF3.LAN is
located, if not SYS:SYSTEM
These are the parameters that can be used with CPQNF3.LAN:
SLOT
SLOT=<slot number>
This parameter identifies the slot in which the controller is
installed. If you are installing more than one controller or
connecting to an additional port, enter the slot number for the
controller or port. If you are using a server that does not have
PCI BIOS version 2.1 or later, the slot number you will have to
use will not match the slot number on the back of the server.
Enter the slot number that is correct for your machine. To see
the slot number(s) displayed, press <Alt>+<Esc> to go to the
server console and load the driver with no parameters (LOAD
CPQNF3.LAN).
FRAME
FRAME=<frame type>
Different protocols typically require different frame types, so
you must load the driver once for each frame type you want to
use. The frame type on your server must match the frame type
loaded on your clients.
Frame Type Protocol Environments
ETHERNET_802.2 IPX, IBM, OSI
(v4.x/3.12 default)
ETHERNET_802.3 IPX (v3.11 default)
ETHERNET_II IPX, TCP/IP
ETHERNET_SNAP IPX, APPLETALK
If you do not specify a frame type, it defaults to Ethernet 802.2
(or 802.3, for v3.11); if you are taking the default, you do not
have to enter the FRAME= parameter. Simply use the LOAD command
to load the driver, as shown below:
LOAD CPQNF3 NAME=SALES
However, to use both the ETHERNET_802.3 and ETHERNET_II frame
types, for example, you must load the driver twice and specify
the frame parameter, as shown below:
LOAD CPQNF3 NAME=ACCOUNTING FRAME=ETHERNET_802.3
LOAD CPQNF3 NAME=ENGINEERING FRAME=ETHERNET_II
NODE ADDRESS
NODE ADDRESS=<locally administrated node address>
Use this statement to set a locally-administered node ID which
will override the globally-administered node ID. It must be a 12-
digit hexadecimal number, in the range permitted under IEEE
guidelines. For example:
NODE ADDRESS=7EFFFFFFFFFEh
NAME
NAME=<board name>
Optional name that you can use for convenience to identify this
controller (or this instance of loading the driver); it is
commonly used when you bind a protocol to the driver. This
parameter is limited to 17 alphanumeric characters, and must be
unique among all logical controllers in the file server. For
example:
NAME=Accounting
Speed and Duplex
10HD or 10FD or 100HD or 100FD
If no speed and duplex setting is specified, the controller will
autonegotiate the speed and duplex with the hub/switch. If the
controller and hub/switch are unable to autonegotiate, the controller
will default to 100Mb half duplex if the controller is capable of
100Mb speed. Full duplex allows for simultaneous transmit and receive
data transfer. Full duplex can be used only when the controller is
connected to a hub that supports and is enabled both for full-duplex
mode and for the specified speed(10Mbps and 100Mbps, respectively).
For example:
LOAD CPQNF3 SLOT=1 FRAME=ETHERNET_802.3 100FD
-----------------------------------------------------------------
Using the BIND Command
After loading the driver, you must bind each controller to a
protocol using the BIND command. Enter the command from the
server console command prompt, or include the command in your
AUTOEXEC.NCF file to automatically bind the driver when you start
SERVER.
This is the format of the command:
BIND {IPX or IP} to {Name} {NET=(number)|ADDR=(number)}
IPX or IP The name of the protocol to
which you are binding the
driver (IPX or IP). If you
specify IP, other parameters
are required; consult your
TCP/IP documentation for more
information.
NAME Optional. The name you
assigned to the controller
with the LOAD command. If you
did not use a controller name,
use the driver name CPQNF3.
NET=[NUMBER] IPX protocol only. The unique
IPX internal network number
you have assigned to this
network; it is a hexadecimal
number and can be up to 8
characters long.
Example:NET=5A
ADDR=[NUMBER] TCP/IP protocol only. The
controller's network address.
The address must be unique on
the internetwork. Example:
addr=192.45.67.8
MASK=[NUMBER] TCP/IP protocol only. The
controller's subnetwork mask.
Example: mask=255.255.255.0
To view the current configuration, type CONFIG and press <Enter>
at the server's console command prompt.
-----------------------------------------------------------------
Loading TCPIP.NLM
This section describes how to load the server driver, CPQNF3.LAN,
and Novell's TCPIP.NLM, which provides TCP/IP support on the file
server. TCPIP.NLM ships with NetWare v3.1x and up. In many cases,
other steps will be required; refer to Novell's documentation for
additional information.
Before loading TCPIP.NLM, you must load a server driver with the
ETHERNET_II frame type, as described earlier. If you also want to
use the Internetwork Packet Exchange (IPX) protocol, you must
load the driver another time with either the ETHERNET_802.2,
ETHERNET_802.3, or ETHERNET_SNAP frame type.
The example below loads CPQNF3.LAN and loads TCPIP.NLM with the
forwarding feature enabled, which allows the server to act as an
internet router:
LOAD CPQNF3 frame=Ethernet_II name=IP_adapter
LOAD TCPIP forward=yes
BIND IP to IP_adapter addr=168.111.112.10 mask=255.255.0.0
-----------------------------------------------------------------
Example LOAD and BIND Commands
The example commands shown below can be entered at the SERVER
command prompt or included in your AUTOEXEC.NCF file for
automatic execution when you start SERVER. The commands must be
entered in the order shown below (the LOAD and BIND commands must
each be entered on a single, separate line, but they must be
shown here on multiple lines because of space limitations).
NOTE: If you include the LOAD and/or BIND commands in your
AUTOEXEC.NCF file, you must re-enter the commands at the
SERVER command prompt, or exit SERVER and restart it in
order to load and bind the driver.
Examples: One CPQNF3 Controller or Port
These examples show the commands for using CPQNF3.LAN in a
NetWare v4.x or v3.1x server named "Pubs" and enabling full-
duplex operation on a controller installed in slot 401. They
assume that the support files are already loaded.
IPX LOAD CPQNF3 SLOT=401 name=pubs 100FD
only: BIND IPX to pubs net=5
IP LOAD CPQNF3 SLOT=401 name=pubs
only: frame=Ethernet_II 100FD
LOAD TCPIP forward=yes
BIND IP to pubs addr=168.111.112.10
mask=255.255.0.0
IPX and LOAD CPQNF3 SLOT=401 name=pubs_IPX 100FD
IP: LOAD CPQNF3 SLOT=401 name=pubs_IP
frame=Ethernet_II 100FD
BIND IPX to pubs_IPX net=6
LOAD TCPIP forward=yes
BIND IP to pubs_IP addr=168.111.112.10
mask=255.255.0.0
Examples: Two CPQNF3 Controllers or Ports
When your server has more than one Compaq controller or a dual-
port controller using the CPQNF3 driver, you need to assign a
logical slot to each RJ45 connector. The logical slot number does
not need to correspond to where the controller is installed.
The following example uses CPQNF3.LAN in a NetWare v4.x server
with two controllers (or one dual-port controller), with "Pubs"
in slot 401 and "Acct" in slot 402:
IPX only: LOAD CPQNF3 slot=401 name=pubs
BIND IPX to pubs net=5
LOAD CPQNF3 slot=402 name=acct
BIND IPX to acct net=6
IPX and LOAD CPQNF3 slot=401 name=pubs_IPX
IP: LOAD CPQNF3 slot=401 name=pubs_IP
frame=Ethernet_II
BIND IPX to pubs_IPX net=5
LOAD TCPIP forward=yes
BIND IP to pubs_IP addr=168.111.112.10
mask=255.255.0.0
LOAD CPQNF3 slot=402 name=acct_IPX
LOAD CPQNF3 slot=402 name=acct_IP
frame=Ethernet_II
BIND IPX to acct_IPX net=6
BIND IP to acct_IP addr=168.111.112.11
mask=255.255.0.0
-----------------------------------------------------------------
UNLOAD and UNBIND Commands
You can use the UNLOAD or UNBIND commands to remove a driver. The
commands have the format shown below:
UNLOAD CPQNF3
UNBIND IPX CPQNF3 [name]
The UNLOAD command completely unloads the driver from memory. If
you want to reload the driver, you will be required to use the
LOAD and BIND commands.
The UNBIND command requires only that you re-enter the BIND
command, and does not affect the LOAD command. You can
selectively unbind a protocol for a particular logical board by
specifying a board name, in the format shown above.
-----------------------------------------------------------------
LAN Driver Ethernet Statistics
Load NetWare's MONITOR.NLM to view statistics to help you
diagnosis network problems. The statistics, which estimate
network and controller activity, supplement the diagnostic
information that would normally be obtained with a protocol
analyzer or network monitor.
NOTE: Most of these statistics will increase during normal
operation. Occasional events do not necessarily indicate a
problem with the network or the network controller board.
These are key statistics that relate to Compaq controllers:
- Send OK Single Collision Count
- Send OK Multiple Collision Count
- Send OK But Deferred
- Send Abort from Late Collision
- Send Abort from Excess Collisions
- Send Abort from Carrier Sense
- Send Abort from Excessive Deferral
- Receive Abort from Bad Frame Alignment
- Total Send/Receive OK Byte Count
Send OK Single Collision Count
The number of single collision packets: This counter contains the
number of packets that are involved in a single collision and are
subsequently transmitted successfully.
These events show that the network has light to moderate traffic.
If single collisions become more frequent, the count for multiple
collisions escalates.
Send OK Multiple Collision Count
The number of multiple collision packets: This counter contains
the number of packets that are involved in multiple collisions
and are subsequently transmitted successfully.
These events mean that the network is experiencing moderate to
heavy traffic. If multiple collisions become more frequent, the
count for excessive collisions escalates.
Send OK But Deferred
The number of packets deferred before transmission: This counter
contains the number of packets whose transmission was delayed on
the first attempt because the medium was busy. Packets involved
in any collisions are not counted. Frames that wait before
transmission are counted. This statistic will be incremented
often during normal operation on a busy network.
Deferred transmissions occur when the network is extremely busy:
so busy that the controller did not try to transmit. High counts
of multiple collisions and excessive collisions also occur.
Deferred transmissions indicate that this segment of the LAN is
overcrowded. Reduce the traffic by reorganizing the LAN. For
example, if you have 100 stations on one Ethernet segment, add a
controller to your server and create two 50-station segments to
balance the load. If a few isolated stations create the traffic,
put them on a separate segment.
Send Abort from Late Collision
Late collisions can indicate cabling problems. A late collision
is one that occurred 64 bytes or more into the packet.
A station will believe it has control of the cable segment if it
has already transmitted 64 bytes. If another node at the far end
of the segment has not yet seen the packet, and transmits, this
packet will collide with the first transmission after the first
64 bytes have been sent. Ensure that your segment length does not
exceed the maximum length allowed.
Because the location of cabling problems can be very difficult to
detect on an Ethernet network, you may want to "shorten" an
Ethernet segment (remove portions of the network to isolate
problems) until the problems are no longer seen, and then expand
the network until the problem recurs.
If this counter increments quickly in a short period of time, it
may mean that the network card is running in half duplex mode,
but your hub port is configured for full duplex mode. Compare
your network card's configuration with the hub port's
configuration and correct any differences.
Send Abort from Excess Collisions
The number of packets aborted during transmission due to
excessive collisions: This counter contains the number of packets
that, due to excessive collisions, are not transmitted
successfully. A station may attempt to transmit up to 16 times
before it must abort the attempt. Once the abort occurs, this
counter increments.
If you see an increase in deferred transmissions as well as
excessive collisions, your network is extremely busy and this
segment of the LAN is overcrowded. Reduce the traffic by
reorganizing the LAN. For example, if you have 100 stations on
one Ethernet segment, add a controller to your server and create
two 50-station segments, to balance the load. If a few isolated
stations create the traffic, put them on a separate segment.
NOTE: Faulty components or cabling can cause excessive
collisions. See Diagnosing Excessive Collisions for help.
Send Abort from Carrier Sense
The number of packets transmitted with carrier sense errors: This
counter contains the number of times that the carrier sense
signal from the physical layer interface was not asserted or was
de-asserted during transmission of a packet without collision.
The carrier sense signal is an ongoing activity of a data station
to detect whether or not another station is transmitting. Carrier
sense errors are detected when a station transmits a frame and
does not detect its own signal on the wire.
If you receive carrier sense errors, see Diagnosing Carrier Sense
Errors.
Send Abort from Excessive Deferral
The number of packets aborted due to excessive deferrals: This
counter contains the number of packets that, due to excessive
deferrals, are not transmitted successfully. This counter is
incremented when the number of transmission attempts is greater
than 16.
Receive Abort from Bad Frame Alignment
The number of packets received with alignment errors: The
alignment of a frame is checked by the receiver after the packet
has failed the Cyclical Redundancy Check (CRC). Misaligned
packets do not end on an 8-bit boundary. All packets contain a
set number of bytes and must end after a defined number of bytes.
Packets that do not end on a byte boundary fail the alignment
check.
If your network continues to experience an increasing number of
CRC/Alignment errors, check for the problems listed in Send Abort
from Excess Collisions.
Total Send/Receive OK Byte Count
These counts are defined as 8 byte values. The LOW count is the
lower 4 bytes of the count. The HIGH is the upper 4 bytes of the
count. The HIGH will increment once every time the LOW reaches 4
billion. At that point, the LOW will start back at zero. Thus, in
a normal situation, you will see a very active and very high
count for the LOW statistic. The HIGH count statistic will be
much smaller. An example of these types of counters is the Total
Send OK Byte Count High and Total Send OK Byte Count Low.
Diagnosing Excessive Collisions
Common Cabling Problems (non-fiber-optic)
- Send Abort from Excess Collisions will increment with each
transmit frame if you are using coaxial cabling (on the 10B2
connector) and the cable is disconnected.
- Segment too long: Nodes at the far end of the cabling system
transmit, unaware that a station at the other end has
already gained control of the medium by transmitting the
first 64 bytes of a frame.
- Failing cable: Packet data traveling through shorted or
damaged cabling may become corrupt before reaching the
destination station.
- Segment not grounded properly: Improper grounding of a
segment may allow ground-induced noise to corrupt data flow.
- Improper termination: If a cable segment is not properly
terminated, allowing the signal to be absorbed upon reaching
the end of the segment, a partial signal will bounce back
and collide with existing signals.
- Taps too close: Follow the minimum recommended spacing
between cable taps to ensure minimal reflection build-up and
data distortion.
- Noisy cable: Interference or noise produced by motors or
other devices can distort the signals and cause
CRC/Alignment errors.
Common Component Problems
- Deaf/partially deaf node: A faulty station that cannot hear
the activity is considered a deaf node. If you suspect a
deaf node, replace the controller or transceiver.
- Failing repeater, transceiver, or controller card:
Repeaters, transceivers, and controller cards can disrupt
the network signal, transmit erroneous signals on the wire,
or ignore incoming packets. Perform the following steps:
* If your controller is continuously transmitting, it causes
erroneous signals or "jabber." Replace a jabbering
transmitter to ensure proper network performance.
* Swap out the transceiver, transceiver cable, and
transceiver attachment point, one at a time. If you find a
faulty component, replace it.
* Check your switch hub. This component may be at fault.
Diagnosing Carrier Sense Errors
If you receive carrier sense errors, use this list to help you
diagnose and treat the cause:
- Failing cable: Packet data traveling through shorted or
damaged cabling may cause a signal loss. Ensure that your
cable is working and plugged in properly.
- (non-fiber-optic) Segment not grounded properly: Improper
grounding of a segment may allow ground-induced noise to
interrupt the signal. Ensure that you have properly grounded
all segments.
- (non-fiber-optic) Noisy cable: Interference or noise
produced by motors or other devices can interrupt the
signals.
- Deaf/partially deaf node: A faulty station that cannot hear
the activity is considered a deaf node. If you suspect this,
replace the controller or transceiver.
- Failing repeater, transceiver, or controller: Repeaters
switches, hub, transceivers, and controllers can disrupt the
network signal, transmit erroneous signals on the wire, or
ignore incoming packets. If you suspect this, swap out each
device one at a time, and replace any faulty component you
find.
If none of the items listed above help you isolate the problem,
you may need to replace the switch or hub. Use a network analyzer
to isolate the problem area.
-----------------------------------------------------------------
Custom Statistics
The controller driver provides these custom statistics:
- TLAN Device Revision (X.X)
- Adapter Check Count
- Transmit Timeout Errors
- Send Abort Link Failures
- Transmit Packets Copied
- Oversized Receive Errors
- Driver Operating Time (minutes)
- Link Speed (bits per second)
- Promiscuous Mode (0=Off, 1=On)
- Network Link Status (0=Up, 1=Down)
- VG Transitions into Training (N/A)
- VG Training Failures (N/A)
TLAN Device Revision (X.X)
This statistic is a two-digit number that represents the
version of ThunderLAN that is used on your controller. For
instance, if this value is 23, then the TLAN version in use is
v2.3.
Adapter Check Count
Counts the number of adapter checks that occur and indicates a
possible hardware problem. A console error message that details
the problem will be generated. The adapter is reset and operation
should continue. The Adapter Reset statistic also increment.
At the next scheduled maintenance, bring down the server and run
the Compaq Diagnostics Utility to determine if the board should
be replaced.
Transmit Timeout Errors
This counter is incremented when a frame does not transfer within
the expected amount of time. The controller will be reset and
operation should continue. The Adapter Reset statistic also
increments.
If this counter increments rapidly in a short period of time,
check to ensure that the cables are properly connected. The Link
Status LED should be lit if the driver is loaded and the
controller is connected to a hub. If this statistic still
increments, power off the system and check the following:
- Check the seating of the controller in its slot. If these
elements appear to be in working order, continue.
- Swap out the controller cables and controllers one at a
time. If you find a faulty component, replace it.
- Check your hub. This component may be at fault. Use the
diagnostics from the hub manufacturer to help you determine
if a problem exists.
Send Abort Link Failures
Indicates how many transmitted packets have failed to transmit
due to link failures.
Transmit Packets Copied
Indicates how many packets have been copied by the driver before
transmission. This counter enables you to monitor inefficient
NLMs running on the server. NLMs that use up to 16 fragments
transmitted per packet may cause this counter to be incremented.
A typical file server never sees this counter increment.
Oversized Receive Errors
If a packet is received that is larger than the maximum allowed
on Ethernet, this counter increments. A high number of
these errors indicates that a node has malfunctioned and is
transmitting oversized frames. Use a network analyzer to
determine which node on the network is causing the problem.
Driver Operating Time (minutes)
Indicates the approximate time in minutes that the driver has
been operating.
Link Speed (bits per second)
Indicates the speed at which the controller is running.
Promiscuous Mode (0=Off, 1=On)
Indicates whether Promiscuous Mode is currently turned on or off
and can be used to verify the operation of network analyzer
products such as Novell's LANalyzer NLM. (When the LANalyzer is
capturing network data, Promiscuous Mode will be on.)
Network Link Status (0=Up, 1=Down)
Reports the current status of the link. Zero indicates the link
is working (up). One indicates that there is no communication
with the hub (down).
If this statistic is one, check your cables and connections. If
the problem persists, try changing the cable or resetting the
hub.
-----------------------------------------------------------------
Identifying Faulty Network Devices
Faulty network devices can produce oversized Ethernet packets.
The largest allowable packet size is 1514 bytes, except when
using the Netelligent Dual 10/100 TX PCI Controller, where the
maximum allowable packet size is 1545 under certain conditions.
Previous versions of the driver supported packet sizes up to 2154
bytes.
Review these LAN statistics using NetWare's MONITOR.NLM:
- Receive Packet Too Big Count
- Checksum Errors
- Send Abort from Late Collision
- Oversized Receive Errors
If any of these statistics has a non-zero value, there may be a
faulty device on your network supplying oversized packets.
-----------------------------------------------------------------
Driver Error Codes and Messages
During initialization of your controller, several error messages
may occur. Note these messages and refer to the following lists
for actions to take:
Message:
- Adapter Check #: Receive List Error
- Adapter Check #: Transmit List
- Adapter Check #: Receive Ack Error
- Adapter Check #: Transmit Ack
- FATAL 403: Internal code initialization failed
- FATAL 404: Unable to register HSM
- FATAL 405: Unable to parse driver parameters
- FATAL 406: Hardware resource conflict; adapter may be in
use.
- FATAL 409: Invalid TLAN ICFGB!
These errors indicate that you may need to update your driver.
Note the error and contact your service provider to obtain an
update. The Transmit List Error may also indicate a problem with
other NLMs installed on the server.
Message:
- FATAL 411: Interface initialization failed
- FATAL 412: Unable to set event processors
- FATAL 418: Unable to register with LSL
- FATAL 420: Transmit frame buffer assignment error
- FATAL 422: Unable to schedule callback
- ERROR 425: New driver version required for adapter in slot #
- ERROR 201: Some TCBs may have been lost!
- ERROR 202: Some RCBs may have been lost!
These errors indicate that you may need to update your driver.
Note the error and contact your service provider to obtain an
update. The Transmit List Error may also indicate a problem with
other NLMs installed on the server.
Message:
- Adapter Check #: Data Parity Error
- Adapter Check #: Address Parity Error
- Adapter Check #: Master Abort
- Adapter Check #: Target Abort
- Adapter Check #: Internal Ack Counter Overflow
- Adapter Check #: Unknown Failure Code.
- FATAL 424: Slot #: Error #: Cannot read EEPROM!
Be sure the module on your controller is connected properly.
Check to be sure the board is firmly installed in its slot.
Install the controller in a different slot in your system. If the
problem persists, contact your service provider.
Message:
- WARNING 301: Adapter in slot # is disabled and cannot be
used.
- WARNING 302: An adapter on the PCI bus is disabled and
cannot be used.
- ERROR 102: Adapter in slot # not configured.
- ERROR 103: Adapter bus #, device # not configured.
- FATAL 402: Invalid EISA configuration
The controller has been disabled by the System Configuration
Utility. Rerun the System Configuration and ensure that the
controller is configured for use. Run the System Configuration
Utility to configure the system.
Message:
- FATAL 401: ETHERTSM.NLM v#.## or greater required.
Obtain ETHERTSM.NLM and MSM.NLM of the version or later specified
in Deciding if Modules Need to be Replaced. Reload the CPQNF3.LAN
driver after loading the new ETHERTSM and MSM NLMs.
Message:
- FATAL 407: Unable to allocate transmit queue memory.
- FATAL 408: Unable to allocate receive queue memory.
- FATAL 410: Unable to allocate TLAN interface memory.
- FATAL 419: Unable to allocate transmit frame buffers.
Add memory to your server.
Message:
- FATAL 421: Unable to allocate any RCBs.
Add memory to your server. Check the Minimum and Maximum Physical
Receive Packet Size parameter. You may need to increase these
parameters based on the number and type of network controllers in
your system.
Message:
- ERROR 203: Cannot update adapter statistics
- ERROR 204: Adapter close failed
- ERROR 205: Adapter close failed in type # shutdown
- ERROR 206: Adapter reset failed
- ERROR 207: Adapter reset failed type # failed
- ERROR 209: Node address re-setting failed
- ERROR 210: Adapter re-open failed
- ERROR 211: Multicast table update failed
- ERROR 212: Promiscuous mode modify failed
The driver will usually continue to operate. During the next
scheduled maintenance, check the seating of the board and run
Compaq Diagnostics. The controller may require replacing or you
may need a driver upgrade.
Message:
- FATAL 413: Unable to initialize adapter
- FATAL 414: Unable to read Burned in Address
- FATAL 415: Unable to set node address
- ERROR 213: Unknown Network Status Event has occurred
- FATAL 416: Unable to open the adapter.
Run Compaq Diagnostics. Your controller may need replacing.
-----------------------------------------------------------------
Advanced Network Fault Detection and Correction
NOTE: This feature applies to controller installations on Compaq
Servers only.
The Advanced Network Fault Detection and Correction Feature of
CPQNF3.LAN allows a secondary controller to take over network
responsibilities if the primary controller fails. This
secondary controller does not handle any network activity unless
a failure occurs with the primary controller. These two
controllers are referred to as the Controller Pair.
NOTE: The feature is not currently supported under NetWare SMP.
Compaq is actively working with Novell to provide this feature
under SMP, and it will be available soon.
When using a controller pair, CPQNF3.LAN can detect failure
events and recover from these errors, usually without noticeably
disturbing normal operation. Advanced Network Fault Detection and
Correction includes recovery from the following error types:
- Hardware Failure Events
Adapter Checks
Transmit Timeouts
Link Downs
- Network Status Changes
Link Status Changes
Transmit Integrity Check Timeouts
Topics:
- Hardware Failure Events
- Network Status Changes
- Secondary Controller Recovery Process
- Installing the Paired Controllers
- Online Control with NetWare
- Performance Tips
-----------------------------------------------------------------
Hardware Failure Events
Hardware failures can occur, and it is important to be able to
recognize the factors that indicate a problem. The CPQNF3.LAN
driver switches operations from the primary controller to the
secondary controller based on these events.
Adapter Checks
These errors indicate that the hardware experienced a problem. A
console error message that details the problem is generated
and an immediate switch-over to the secondary controller occurs.
Transmit Timeouts
This counter is incremented when a packet does not transfer
within the expected amount of time. The controller from which it
was transmitted is presumed failed and is reset. If this was the
primary controller, the CPQNF3.LAN driver switches over to the
secondary controller.
-----------------------------------------------------------------
Network Status Changes
Changes in the network are detected through transmit integrity
checks and link status changes.
Transmitter Integrity Check
When a secondary controller is installed in your server and idle
time exists for either controller, the Transmitter Integrity
Check feature transmits packets to itself. The secondary
controller, since it is not active, uses this feature
regularly. If the primary controller fails the Transmitter
Integrity Check, the secondary controller assumes control of
network activity. A console alert is generated and the event
is logged in the console log. If at any time the secondary
controller does not pass Transmitter Integrity Check, an alert
is generated and the event is recorded in the console
log.
The Transmitter Integrity Check feature can be disabled by
setting the optional parameter, TXTESTTIME, to 0. Refer to the
Installation section for usage of the TXTESTTIME keyword.
If you receive an alert indicating a link failure, you should
first check the cable connections and the seating of the failed
controller to ensure that everything has been properly installed.
If you still receive errors, replace the cables attached to the
failed controller. At the next scheduled maintenance, run Compaq
Diagnostics on the controller. It may require replacement.
Link Status Change
Link status changes may occur when a cable is unplugged or a hub
problem occurs. If a link status change occurs on the primary
controller, the secondary controller assumes control of
network activity. If the secondary controller is attached to the
same hub as the primary controller, the problem is resolved
if the cause was related to a cabling problem. However, if the
hub is related to the problem, both controllers are marked
failed. An alert occurs, and the event is logged in the
console log.
If you receive an alert, first check the cable connections and
the seating of the failed controller to ensure that everything
has been properly installed. If you still receive errors, replace
the cables attached to the failed controller. Check the operation
of your hub. At the next scheduled maintenance, run Compaq
Diagnostics on the controller. It may require replacement.
-----------------------------------------------------------------
Secondary Controller Recovery Process
When the driver detects failures on the primary controller and
the secondary controller has a valid link, redundant recovery
occurs automatically. At the time of switchover (i.e., when
the primary controller fails the Transmit Integrity Check), the
MAC address of the primary controller is copied to the secondary
controller by the driver. The secondary controller then
begins operation. If the secondary controller has been installed
on a different hub from the primary controller, the controller
pair can continue operation through a single hub power loss or
hub failure on one of the controllers.
If the secondary controller takes over operation for any reason,
an alert, notifying you of the primary controller's failure, is
sent to the NetWare console. You should take corrective action at
the next scheduled maintenance. Run Compaq Diagnostics on the
failed controller.
-----------------------------------------------------------------
Installing the Paired Controllers
Hardware Installation Tips
- Supported controllers: All controllers supported by
CPQNF3.LAN, including the NetFlex-3 and Netelligent
controllers, installed on a Compaq server.
- The secondary controller must be installed on the same
network segment as the primary controller, but you can
connect the secondary controller to a different hub. If
the primary controller failure is related to a hub problem,
the secondary controller can continue operation.
- The paired controllers must be identical in speed and
protocol; for example, do not pair a 100Base-TX module with
a 100VG module or a 10Base-T module.
Software Installation Tips
The NetFlex-3 driver must be configured properly for the network
fault detection feature. Three new command line keywords are
available for configuring a controller pair for fault tolerant
operation. The new keywords are TXTESTTIME, PAIR, and SECONDARY.
TXTESTTIME
The TXTESTTIME parameter controls the frequency that test frames
are transmitted by the controller in order to verify that a valid
network link exists. Using a setting of zero disables this
feature. The default setting is zero (disabled) unless the
Advanced Network Fault Detection and Correction Feature is used.
In this case, the default is one second. The test frames are
transmitted only during periods of inactivity. The minimum
setting is 0 (disabled); the maximum setting is 10. Exercise
caution when changing this parameter. If the TXTESTTIME value is
too large, CPQNF3 may not be able to switch over opportunely
after a failure without workstations losing their connection to
the server.
LOAD CPQNF3 SLOT=3 FRAME=ETHERNET_802.2 TXTESTTIME=2
PAIR
The PAIR keyword identifies the controller pair associated with the
controller and must be used when loading both controllers. This
parameter is valid only when using the Advanced NetWork Fault
Detection and Correction feature and is required to load the
primary and secondary controllers. If the SECONDARY keyword is
not present, this controller is recognized as the primary
controller of the controller pair; the secondary controller
must be loaded prior to using to the SECONDARY keyword.
LOAD CPQNF3 SLOT=3 PAIR=1 SECONDARY (secondary)
LOAD CPQNF3 SLOT=4 PAIR=1 FRAME=ETHERNET_II (primary)
SECONDARY
The SECONDARY keyword is required on the command line for the
driver to load the secondary controller when using the Advanced
network Fault Detection and Correction feature. If this keyword
is present on the command line, the controller is placed in
an inactive state awaiting activation when the primary controller
(of the controller pair) is loaded. This keyword can be used
only with the PAIR keyword. The secondary controller must be loaded
before loading the primary controller of a controller pair.
LOAD CPQNF3 SLOT=3 PAIR=1 SECONDARY (secondary)
LOAD CPQNF3 SLOT=4 PAIR=1 FRAME=ETHERNET_II (primary)
Example Configuration
For example, if Slots 1 and 2 contain 10/100 TX PCI UTP
controllers, use the following syntax in the AUTOEXEC.NCF to
configure the pair for fault tolerant operation and specify
that the transmit integrity test frames be transmitted
every 3 seconds:
LOAD CPQNF3 SLOT=2 PAIR=1 SECONDARY
LOAD CPQNF3 SLOT=1 PAIR=1 FRAME=ETHERNET_II NAME=IPXNET1
TXTESTTIME=3
LOAD CPQNF3 SLOT=1 PAIR=1 FRAME=ETHERNET_802.3 NAME=IPXNET2
TXTESTTIME=3
BIND IPX TO IPXNET1 NET=123
BIND IPX TO IPXNET2 NET=246
Verify Operation of the Secondary Controller
Both controllers must be present and initialize properly when the
CPQNF3.LAN driver is loaded. If CPQNF3.LAN is unable to initiate
startup of the secondary controller, an error message appears on
the console. Make a note of this error and see the Error Messages
section for more information. To ensure that the secondary controller
is operational, unplug the cable from the primary controller.
Operation should switch to the secondary controller within a few
seconds.
-----------------------------------------------------------------
Online Control with NetWare
Advanced Network Control Utility
The Advanced Network Control Utility is located within the Compaq
Online Configuration Utility, CPQONLIN.NLM. This utility is
available only on the Compaq NetWare Software Support diskettes
or CD supplied with a Compaq Server. With this utility you can
control and monitor the status of Controller Pairs configured for
the Advanced Network Fault Detection and Correction feature. The
utility also lets you switch control to the secondary controller
at any time. If you suspect a problem with the primary
controller because MONITOR.NLM shows networking error statistics
incrementing rapidly, use CPQONLIN.NLM to switch to the secondary
controller. At the next scheduled maintenance, run Compaq
Diagnostics on the primary controller.
The Advanced Network Control Utility also has an advanced
statistics screen that displays statistics for both controllers
simultaneously. This screen is activated by pressing <F7> from
the controller pair window.
There are two modes of operation that can be set from within the
advanced network control utility: automatic and manual.
Automatic Operation
When in automatic mode, which is the default, if the primary
controller experiences a failure for any reason, activity
is automatically switched to the secondary controller in the system.
No user intervention is required. You can also manually switch
activity from one controller to the other at any time using the
Advanced Network Control Utility.
Manual Operation
When in manual mode, you must manually switch operation of the
primary controller to the secondary controller. If a failure
occurs, activity will not automatically switch to the secondary
controller and the server may experience a network outage. Use manual
mode only for test purposes.
-----------------------------------------------------------------
Performance Tips
- If you are using the Netelligent Dual 10/100 TX PCI UTP
controller in a ProLiant 5000, you may see the following
message when loading the driver:
"NOTE: For better performance, set maximum physical receive
packet size = 1545"
This message means that in order to get the best performance
out of your Netelligent Dual 10/100 controller, you need to
adjust NetWare's 'MAXIMUM PHYSICAL RECEIVE PACKET SIZE'
parameter to at least 1545. Place this setting in your
STARTUP.NCF file.
- Be aware that some common performance testing software has
not yet been upgraded to handle 100Mbps networking
technology. Be sure you obtain the latest versions of
software test products. Compaq runs various performance
tests and has found with older versions of some software,
such as NetBench 3.0 and PERFORM3, the performance figures
were incorrect. If test results do not appear to make sense,
try running the test again using different parameters or use
a different benchmark test.
- If you are using a 100Mbps controller in your server and the
Intel PRO/100 in clients, you may see a large number of
Receive Abort from Bad Frame Alignment errors. Set the
THRESHOLD parameter in the client NET.CFG file to 200 after
the Link Driver line. For example:
Link Driver e100odi
THRESHOLD 200
You can add any frame types that are necessary after these lines.
- For best performance, note whether your server's primary bus
is PCI-based or EISA-based and choose a network controller
that matches this primary bus. Matching buses means that
bridging across bus types is not necessary.
- Packet Burst Problems - If you notice that the file server's
CPU utilization is at 100 percent when clients are reading
large files, a VLM upgrade (now available from Novell) may
help. This upgrade will enhance the performance of your
100Mbps controller.
- When using LANalyzer for Windows to capture traffic on
100Mbps networks, use the fastest possible processor (minimum
of 486, 586-class recommended) and a PCI-based machine. If
you are not using fast enough equipment, LANalyzer may not
be able to capture all the packets on the network.
Download Driver Pack
After your driver has been downloaded, follow these simple steps to install it.
Expand the archive file (if the download file is in zip or rar format).
If the expanded file has an .exe extension, double click it and follow the installation instructions.
Otherwise, open Device Manager by right-clicking the Start menu and selecting Device Manager.
Find the device and model you want to update in the device list.
Double-click on it to open the Properties dialog box.
From the Properties dialog box, select the Driver tab.
Click the Update Driver button, then follow the instructions.
Very important: You must reboot your system to ensure that any driver updates have taken effect.
For more help, visit our Driver Support section for step-by-step videos on how to install drivers for every file type.