                       NETWORK DIAGNOSTIC PROGRAM

                Copyright 1993 Invisible Software, Inc.



       [See GUIDE.TXT for a list of all manuals and text files.]



     The NETDIAG program lets you test and analyze the operation of the
network hardware.  NETDIAG has three different functions:

     *  Polling - Displays a list of the stations in the network, and
        verifies that the stations are able to communicate.

     *  Status - Displays traffic and error statistics that indicate how
        much network traffic has been sent and received by each station,
        and how many errors each station has detected.

     *  Stress - Sends and receives network packets as rapidly as
        possible, and logs the number that are lost.

     NETDIAG is independent of the other programs in Invisible LAN.  You can
use NETDIAG regardless of whether or not Invisible LAN is running.  Therefore,
NETDIAG can be used to verify the hardware installation before you start
using Invisible LAN.



------------------------------------------------------------------------



TABLE OF CONTENTS:

 1.  Starting NETDIAG
 2.  Polling
 3.  Status
    3.1.  Error Counters for Invisible Network
    3.2.  Error Counters for Ethernet
    3.3.  Error Counters for Token Ring
 4.  Stress



-----------------------------------------------------------------------



1.  STARTING NETDIAG


     To start NETDIAG, type the command:

        NETDIAG

     When NETDIAG is started, it first checks that the TransBIOS/NetBIOS
software is loaded in your computer.  The TransBIOS/NetBIOS software can
be loaded either from the network board's ROM, or from disk.  If the
software is not loaded, NETDIAG asks you what type of network hardware
you have, and then automatically loads the TransBIOS/NetBIOS software.

     The following sections describe the functions that NETDIAG
performs.  At any time, you can get on-line help by pressing F1, and you
can exit from NETDIAG by pressing Esc.

     Note:  The NETPOLL program that was included in old versions of
Invisible LAN has been discontinued.  It is replaced by the polling
function in NETDIAG.



------------------------------------------------------------------------



2.  POLLING


     When you first start NETDIAG, it begins with the network polling
function.  You can also activate the polling function by pressing F2.

     When you activate the polling function, NETDIAG tries to contact
all the other stations in the network, and displays a list of the
stations it is able to contact.

     For each station in the list, NETDIAG displays the following
information:

     *  Station address or serial number.

     *  A flag which indicates if the station is local or remote.

     *  The type of network hardware installed on the station.

     *  The configuration of the network hardware.  The information
        varies according to the type of hardware installed, but
        generally includes things like I/O address and interrupt level.

     *  The version number of the TransBIOS software.

     *  The amount of time the station has been running, in minutes.
        This is displayed in hexadecimal (base-16) notation.

     *  The Invisible LAN name assigned to the station.

     The display is automatically updated every few seconds.  If there
are more stations than fit on the display, you can use Home, End, PgUp,
and PgDn to scroll through the list of stations.

     When you first install your network hardware, you should run
NETDIAG on all your network stations.  If each station's display shows
all the other stations in the network, then your network hardware is
operating correctly.



------------------------------------------------------------------------



3.  STATUS


     You activate the status function by pressing F3.

     When you activate the status function, NETDIAG displays a list of
all the stations in the network.  For each station, it shows the
following information:

     *  Station address or serial number.

     *  Number of packets transmitted and received by the station.

     *  Error counters for the station.

     *  The number of active network sessions.

     *  The name of the station (as defined by Invisible LAN).

     The NETDIAG status display is updated automatically every few
seconds.

     The error counters vary depending on the type of network hardware
you are using.  The following sections describe the error counters
for each supported type of hardware.



3.1.  ERROR COUNTERS FOR INVISIBLE NETWORK


     For the Invisible Network products (Models 100, 200, 300, 200/A,
and 300/A), the following error counters are shown:

     RTx (Retransmit) - The number of session packets that had to be
        retransmitted because the other station failed to respond in
        time.

     Pty (Parity) - The number of parity errors detected by the network
        hardware.  If you get a lot of these errors, it may indicate a
        problem with the network cable, the terminators, or one of the
        network cards.

     Ovr (Overrun) - The number of DMA overrun errors during receive.
        This means that the computer's DMA controller was too slow in
        responding to the network card.

     CkS (Checksum) - The number of packets received with invalid
        checksum, length, or address.  If you get a lot of these errors,
        there may be a problem with the network cable, the terminators,
        or one of the network cards.

     Sql (Squelch) - The number of times that the computer did not
        respond to a network interrupt in time.  It is normal to have
        one or two of these errors when you first start the network.

     LBk (Loopback) - The number of times that the received data did not
        match the transmitted data.  This may indicate a problem with
        the network cable, or the line driver chips on the network card.

     SAb (Session Abort) - The number of network sessions that were
        aborted because the other station did not respond.  The most
        likely cause is that the other station was turned off or
        rebooted.

     LoD (Lost Data) - The number of times that the station failed to
        receive session data which was sent by another station.

     Srv (Service) - The number of times that a transmission was delayed
        due to heavy network traffic.  This may indicate a very heavily
        loaded network, or it may indicate that you turned off the
        master station.



3.2.  ERROR COUNTERS FOR ETHERNET


     For Ethernet products, the following error counters are shown:

     Col (Collision) - The number of times that the station encountered
        a collision when attempting to transmit a packet (i.e., another
        station tried to transmit at the same time).  It is normal to
        have a large number of collisions.

     RTx (Retransmit) - The number of session packets that had to be
        retransmitted because the other station failed to respond in
        time.

     LoD (Lost Data) - The number of times that the station failed to
        receive session data which was sent by another station.

     HOv (Hardware Overflow) - The number of times that the on-board
        static RAM was filled up, thereby causing a loss of received
        packets.

     CRC (Cyclic Redundancy Check) - The number of received packets
        that were corrupted, and were discarded by the hardware.  It is
        normal to have a few of these errors.

     Aln (Alignment) - The number of received packets that did not end
        on a byte boundary, and were discarded by the hardware.  It is
        normal to have a few of these errors.

     AbT (Aborted Transmissions) - The number of packets that could not
        be transmitted because of excessive collisions.  This error may
        indicate an overloaded network.

     Res (Reset) - The number of times that the network hardware failed
        and had to be restarted.  This number should be very small.

     SA (Session Abort) - The number of network sessions that were
        aborted because the other station did not respond.  The most
        likely cause is that the other station was turned off or
        rebooted.

     BP (Bad Packet) - The number of received packets that were
        discarded by the software because they were corrupted.  This
        number should be very small.

     F (FIFO) - The number of times that the network hardware failed
        to access the on-board static RAM in time.  This number should
        be very small.



3.3.  ERROR COUNTERS FOR TOKEN RING


     For Token Ring products, the following error counters are shown:

     RTx (Retransmit) - The number of session packets that had to be
        retransmitted because the other station failed to respond in
        time.

     LoD (Lost Data) - The number of times that the station failed to
        receive session data which was sent by another station.

     NCp (Not Copied) - The number of transmitted packets that were not
        copied by the intended destination station, either because the
        destination station was not on the network, or because the
        destination station had no free buffer space on the adapter.

     Drp (Dropped) - The number of received packets that were discarded
        because TransBIOS had no free data buffers.  If you get a lot of
        these errors, you should increase the TransBIOS data segment
        size.

     SA (Session Abort) - The number of network sessions that were
        aborted because the other station did not respond.  The most
        likely cause is that the other station was turned off or
        rebooted.

     BP (Bad Packet) - The number of received packets that were
        discarded by the software because they were corrupted.  This
        number should be very small.

     TF (Transmit Failure) - The number of packets that could not be
        transmitted onto the ring.  This number should be very small.

     RF (Receive Failure) - The number of receive operations that
        failed and were aborted.  This number should be very small.

     AC (Adapter Check) - The number of times that the token ring
        adapter required a hardware reset because of an adapter check
        error.  This number should be very small.

     RS (Ring Status) - The number of times that the token ring adapter
        required a hardware reset because of a fatal ring status change.
        This can be caused by pulling out the network cable, or some
        other disruption of the network.

     PC (PC Error) - The number of times that the token ring adapter
        required a hardware reset because the LAN Support Program
        detected a fatal error.  This number should be very small.



------------------------------------------------------------------------



4.  STRESS


     In addition to displaying status information, NETDIAG can also
perform a stress test of the network.  The stress test can reveal
network problems by sending a large number of network packets in a short
period of time.

     To begin the stress test, press F4.  Within a few seconds, NETDIAG
begins transmitting packets as fast as it can.  NETDIAG also counts the
response packets that it receives from other stations.  If any of the
packets are lost, NETDIAG displays the number that were lost.

     In order for the count of lost stress-test packets to be correct,
it is necessary that you run NETDIAG on only one station, while all the
other stations are quiescent.  In this case, "quiescent" means:  (1) the
station is not running NETDIAG or any other diagnostic program;  (2) the
station has no active drive or printer mappings (either as a server or
as a workstation);  (3) the station has no active network sessions.

     To terminate the stress test and return to the configuration or
status screen, press F2 or F3.  To return to DOS, press Esc.
