Ref: 99980157
Title: Customer Care Bulletin, April 1985
Date: 4/1/85

Copyright 3Com Corporation, 1991.  All rights reserved.

.h1;ETHERSERIES 2.4 IS SHIPPING

Here are some of the enhancements ... and bugs.

Enhancements

  * Dedicated Mode has not changed
  * EtherPrint performance is improved
  * EtherPrint includes a /DIRECT option (this eliminates the
    spooler; however, spooled files will not print until the
    DIRECT has been unlinked or times out
  * The /PLOT option on EtherPrint no longer requires an EP
    UNLINK command to start printing
  * A local user on a Standard Mode server can now cross - link
    to other servers under 2.4
  * Standard Mode servers can address more than one hard drive
  * Support for DOS 3.0 and the IBM AT as a server
  * The Standard Mode server can have 3 printers (two parallel
    and one serial HP LaserJet).  We are experimenting with other
    serial printer.  The Dedicated server will support two
    parallel printers.
  * You can have up to 40 servers on one network
  * Maximum number of semaphores has been increased to 150
  * The Standard Mode server will now support EtherMail
  * Featured is EtherMenu, an interactive menu software program
  * A CONFIG program to help in changing DMA, Interrupt Level and
    I/O Base Address
  * The Administrator can now control an individual user's access
    to create users or volumes.
  * When using EtherShare and EtherPrint on a Standard Mode
    server, the minimum RAM required is 320K.  Additional RAM is
    required for third - party hard disk drives or to run
    application software.

  BUGS

  * If your user node is an XT and you want to boot from it,
    FDSETUP is wrong.  You must edit the file FDSETUP.BAT (using
    any editor such as edlin, WordStar, MultiMate) to change all
    of the b: to a: and the a: to c:.  This will copy the files
    onto your hard disk; or you can just use the copy command and
    copy all the files onto the hard disk.

  * If you bring up the server in the Standard Mode but do not
    have it connected to the network and attempt to log in,
    nothing will happen.  You must have your server either on the
    network or have a loopback plug attached to the EtherLink
    card.

For your Information

  * If you change the I/O base address, DMA or the interrupt
    level on the Etherlink card, the software must also be
    changed (see the Administrator's Guide, Appendix B).  After
    running the CONFIG utility you must reinstall the server
    software.

  * The total number of links allowed on a 2.4 print server is
    17.  If you're not using your link, unlink the printer so
    someone else can use it.  It's best NOT to include EP LINK in
    your AUTOEXEC file.

    If you prefer to use EP LINK/HOLD, remember to unlink it
    before linking to another printer.  An EP LINK command to
    another printer won't unlink your current link unless linked
    to another printer on the SAME server.  If you fail to unlink
    a /HOLD link before switching to another server's printer,
    your spool file will never be printed because the link is
    destroyed.

    Before shutting down your computer, unlink everything (ES and
    EP).  Without properly unlinking your EP links, the print
    server still thinks you're linked to it.  One way to make
    sure you unlink properly is to make a batch file which
    includes EP UNLINK and ES LOGOUT.

  * If you bring up your server in Standard Mode and get the
    error message "Not ENOUGH BUFFERS TO CACHE FAT", you must
    increase the CIOSYS buffers.  Refer to the Administrator's
    Guide, Appendix E - 3.

  * With EtherShare Version 2.4, it is possible to add expansion
    peripherals to your Standard Mode server.

The normal configuration would have EtherShare installed to
drive C:, and drives D:, E:, F:, etc., installed as expansion
peripherals. As released, EtherShare 2.4 will not recognize
expansion peripherals which have a lower drive specifier than
where the EtherShare software resides.


.h1;Memory Requirements

Memory requirements for a Standard Mode server are as follows:

  DOS 2.1 + EtherShare    = 219K
          + Print     = 270K

    or + Mail      = 252K
       + Print and Mail =  301K

  Ethermenu             104K
  Third party hard disk drive   64K
  Additional memory is required for application software.


.h1;Compatibility of Releases 2.2 and 2.4

Be careful; 2.2 user software is compatible with 2.4 server
software, and servers running only 2.2 EtherSeries server
software can run on the same network as servers running only 2.4
EtherSeries server software. HOWEVER,each server running on the
network must be consistent and run either all Release 2.2 server
software (EtherShare, EtherPrint, and EtherMail) or all 2.4
server software.  EXAMPLE: EtherShare PC Server software, Release
2.4, should NOT be installed on the same server with EtherMail,
Release 2.2.

We recommend that for ease of use and consistency of network
functions, the entire network be upgraded to EtherSeries 2.4.

.h1;EtherStart

Here's what we have found about Etherstart with ES 2.4 and DOS 3.0:

(1) If the server is running DOS 3.0 and you create a START volume
using DOS 2.0 or 2.1, everything works fine.

(2) If you create a START volume using DOS 3.0, it will work.  It
lets you log in and links the START volume, but then hangs.  On a
PC with a floppy drive, the drive light comes on and stays on.  A
soft reboot is not possible; the system must be powered down to
recover.  You also need a sys volume with DOS 2.0 or 2.1 files.

.h1;Software Interrupt Conflicts

When the server is in Standard Mode, some application software
may cause the following error message to appear at the server
user station:

  "APPLICATION PROGRAM AND SERVER INTERRUPT CONFLICT.  DOS HAS
   BEEN HALTED, SERVER STILL RUNNING"

If the software application package attempts to change this
interrupt, our software notices and prints the interrupt conflict
message.  When this happens, the server goes into "serve - the -
network - only" mode, and the user's program is suspended.

The only solution is to use the software solely from a user
workstation.  The problem is that the software uses interrupts
which cause the server to be confused.  As you know, our software
uses interrupts to signal an incoming packet to the server.  If
the server station receives interrupts from the software package
as well, it 'gets confused' and returns that error message.

3Com uses the following Interrupts:

    60H Enet Entry
    61H Cdriver/Mudriver entry calls
    62H Ciosys entry calls
    63H Cstartup entry calls
    64H Dummy device header address

What we hook in the Operating System includes:

    08H Hardware timer is hooked
    28H Hooked to facilitate foreground - background timeslicing
        when DOS is idle.

Also consider the 3Com hardware interrupt.

.h1;Using Diagnostics - Divide and Conquer

When part or all of your network is experiencing a problem, the
best way to find out what is wrong is to use an ancient battle
plan: divide and conquer.

The object is to find what will work and then proceed step by
step until it doesn't work anymore.  Then the failing component
can be identified. Although this can be time consuming, it is the
most complete method of troubleshooting available.

By following the steps below you can minimize the number of calls
you need to place to 3Com Technical Support and the amount of
time you need to spend on the phone when a call is necessary.

ERROR:  *** Write to network failed.

Is the message occurring on all workstations or just one?

ALL STATIONS:

(a) Check the cable. Make sure that it conforms to spec (thin
ethernet -RG58A/U or RG58C/U; thick Ethernet - 50 ohm Ethernet
spec cable.)

(b) Check the terminators. There must be a terminator (50 ohm) on
each end of the network cabling.

(c) Check the cable connections. Make sure that there the cable
is securely connected at all points. Cable must run in one
continuous line (bus topology) and be terminated on each end.


ONLY ONE STATION:

(a) Check to see if the cabling to this station is different than
the rest of the network. For thin Ethernet, you must use either
RG58A/U or RG58C/U cabling.  For thick Ethernet, you must use 50
ohm Ethernet spec cabling.

(b) Make sure the T-connector or transceiver drop cable is
attached securely to the board connector.

(c) Make a loopback plug (two terminators attached to a T -
connector) and attach it to the board. Run DIAGNOSE L.

If DIAGNOSE PASSES: Double check the cable type and connections to the
board.

If DIAGNOSE FAILS: Move the loopback plug to another machine
and run DIAGNOSE L to make sure that system's board is good.
Move the good board and the loopback plug to the workstation
having a problem.  Run DIAGNOSE L again.

  If DIAGNOSE PASSES: Check to see that the jumpers on both boards
  are the same.

If the SAME: The original board is bad and you should call 3Com
for a return authorization (RMA) number.

If DIFFERENT: Change the jumpers on the board that failed to
correspond to the ones on the board that passed. Place the
failed board in the system with the loopback plug in place
and run DIAGNOSE L again. If it passes, the problem is
solved. If not, the board is bad. Call 3Com for an RMA number
to return it for repair.

If DIAGNOSE FAILS: Take out any additional memory,
communications or other types of boards that are in the
machine. Run DIAGNOSE L again.

If DIAGNOSE passes: Return the other boards to the machine
one at a time, running DIAGNOSE L each time. When it fails
again, you will know that the board you just returned to the
system has a conflict with the EtherLink board.

If DIAGNOSE falls: The PC has a problem on either the mother
board or the bus interface. Run the necessary diagnostics on
the PC to determine the problem.

ERROR: *** Server not ready. Correct login name?

is the message occurring on all workstations or just one?

ALL STATION:

(a) Make sure your cable is RG58 A/U or C/U (thin ethernet) or
standard 50 ohm (thick ethernet).

(b) Check to see if there is a conflict with DMA, Interrupts or I/
O Base Addresses (from 300 to 30F as defaults.)  If there is a
conflict with another board, change the other board or the
EtherLink card.  If the EtherLink card is changed, make the
appropriate software changes as described in Appendix B of the
Administrator's Guide.

(c) If changes were made to the EtherLink board, double check the
corresponding changes to the software.

(d) Make a loopback plug (two terminators attached to a T-
connector), attach it to the EtherLink board in the server and
run DIAGNOSE L.  The server should not be in operation when
DIAGNOSE is run.  NOTE: If you made changes to the EtherLink
board, add the appropriate arguments to the DIAGNOSE command.
See Page C-3 of the User's Guide for options.

If DIAGNOSE PASSES: Go back and re-check items (a), (b), and
(c) above.

If DIAGNOSE FAILS: Move the loopback plug to a user station
and run DIAGNOSE L to make sure the board is good.  Move the
good board and the loopback plug to the server station.  Run
DIAGNOSE L again.

If DIAGNOSE PASSES: The original EtherLink board in the
server is bad.  Carefully remove the Ethernet Address
chip from the board that was originally in the server
system.  (The chip is located beneath the largest chip on
the board and has a white sticker on it with the Ethernet
address.)  Being careful not to confuse the two chips,
remove the Ethernet address chip from the good board and
replace it with the one from the original server board.
CAUTION:  Notice the orientation when replacing the chip.
Your server software has been encrypted with this
Ethernet address and won't operate with a different one.
Call 3Com and obtain a return authorization (RMA) number
for the bad board.

If DIAGNOSE FAILS: Take out any additional memory,
communications, hard disk controller or other types of
boards that are in the machine.  Run DIAGNOSE L again.

If DIAGNOSE PASSES: Return the other boards to the
machine one at a time, running DIAGNOSE L each time.
When it fails again, you will know that the board you
just returned to the system has a conflict with the
EtherLink board.  Change the conflicting DMA,
Interrupt or I/O Base address either on the EtherLink
card or on the add - in board to avoid the conflict.

If DIAGNOSE FAILS: The PC has a problem in either the
motherboard or the bus interface.  Run the necessary
diagnostics on the PC to determine the problem.


