home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Simtel MSDOS - Coast to Coast
/
simteldosarchivecoasttocoast2.iso
/
ncsatlnt
/
tel23doc.zip
/
CHAP.7
< prev
next >
Wrap
Text File
|
1991-07-01
|
40KB
|
1,146 lines
Installation and Configuration 7.1
June 1991
7.1 NCSA Telnet
Installation and Configuration 7.1
National Center for Supercomputing Applications
June 1991
6/30/91
Chapter 7 Installation and Configuration
Chapter Overview
Installing the Distribution Files
Sample Configuration
Required Configuration Information
The TELNET.BAT File
Command Line Options
The Configuration File
Overview
Syntax for Entries
PC Information Entries
Host-Specific Parameters
Convertion of UNIX /etc/hosts Files
Hardware Options
Combined Ethernet Drivers
Performance Tuning
RARP for Dynamic IP Address Assignment
BOOTP for Dynamic IP Address Assignment
Interrupts
Domain Name Lookup
FTP Password Protection
Compatibility Notes
Ping
VT102
FTP
Terminate and Stay Resident (TSR) Programs
NCSA Telnet
Limitations
Chapter Overview
This chapter contains information for system administrators
(and other experienced users) to use in installing and customizing
a telnet system. It describes how to install NCSA Telnet, change
the configuration file, the domain name lookup, how to protect FTP
transfers with passwords, and discusses compatibility issues.
Installing the Distribution Files
Three files are required for the standard configuration of NCSA
Telnet. The combination of these with the use of the DOS path
command allows you to call up NCSA Telnet from any
subdirectory on your PC.
TELNET.BAT
This batch file begins the execution of NCSA Telnet. It must be in a
directory that is in your DOS path. It contains the locations of the
TELBIN.EXE and CONFIG.TEL files. You must update this batch
file (with any standard editor) when you change the location of
TELBIN.EXE or CONFIG.TEL.
TELBIN.EXE
This file is the program itself. You can place it in any directory.
The TELNET.BAT file calls up this program when it runs.
CONFIG.TEL
Config.tel is the configuration file that contains your PC's IP
address, the kind of Ethernet hardware that you are using, and
other information about your network. Gateways, nameservers,
and setup information are stored here. The TELNET.BAT file
tells TELBIN.EXE where it can find this file on your disk.
Sample Configuration
Suppose you want to install NCSA Telnet in a directory called
\ncsa on your hard disk drive, drive C. Copy the TELBIN.EXE
and CONFIG.TEL files to C:\ncsa. Suppose also that you have a
particular directory C:\bat, which you use to store commonly
used batch files. Your DOS path (see DOS manual) contains an
entry for C:\bat already. Copy the TELNET.BAT file into
C:\bat. You must now edit the TELNET.BAT file to update the
locations of the TELBIN.EXE and CONFIG.TEL files. When
finished, it should look something like this:
if .%1 == . goto defh
c:\ncsa\telbin -h c:\ncsa\config.tel %1 %2 %3 %4 %5
goto exith
:defh
c:\ncsa\telbin -h c:\ncsa\config.tel ncsab
:exith
This batch file calls up TELBIN.EXE at a known location and
passes it the -h parameter, indicating the location of the
CONFIG.TEL file. In addition to the file locations, the fifth line
indicates which machine NCSA Telnet should connect to by
default.
Required Configuration
Information
To set up a new configuration file, you will need a standard text
editor. The template provided with NCSA Telnet is set up to reflect
commonly used default values. It may be worth the time to go
through the list of configuration options in this chapter and fine-
tune your configuration to match your local computing situation
more closely. In the meantime, you must install the following
fields in the configuration file for NCSA Telnet to operate.
Field Meaning
myip=192.17.20.10 You must provide a unique IP address
for each machine on your network. IP
addresses must be installed in the
configuration file of each computer.
netmask=255.255.255.0 This field is the subnet mask, (in decimal),
and is required if you are using subnets on
your network.
hardware=3C501 NCSA Telnet can support several
different brands of Ethernet adapters.
Each brand has its own unique
identifier which you must set for the
hardware parameter. See the
hardware section in this chapter for a
complete list of settings. Most brands
also require address and ioaddr to be
set.
video=ega If you wish to use the Tektronix
graphics options, you must provide the
correct screen type. Incorrectly
installed, this option can crash NCSA Telnet.
The alternative is to install the tek=no
option. Legal types are cga, ega,
hercules, no9, mono, ega43, vga50.
gateway=1 If you are going to be using gateways
to reach other networks, at least one
machine must be designated as a
gateway, using this gateway option
after the name field. The entry for each
gateway must include a hostip field,
according to the configuration file
directions in this chapter.
The TELNET.BAT File
The actual .exe file that the PC executes is not named telnet.exe. A
batch file called TELNET.BAT loads the software in several steps
that are normally transparent to you. Here is a typical entry in
TELNET.BAT:
C:>\apps\telbin -h c:\apps\config.tel %1 %2 %3 %4
The TELNET.BAT file, which must be in the DOS path, specifies
completely where the program is and where the configuration file
is found. If you do not explicitly specify a machine name, the
system connects to a previously assigned default machine name.
Command Line Options
NCSA Telnet goes through two stages as you load it. The program
first finds the TELNET.BAT. (This file must be in the DOS path so
it can be found first.) Secondly, in the batch file, the location of the
TELBIN.EXE file and any command line defaults are used to call
up the program. We've included many command line arguments
for you in the batch file, TELNET.BAT; you can add others when
you run the telnet batch file. This following command gives a full
description of those options. The syntax for loading NCSA Telnet
with options is:
C:\ > telnet [options] [machine1 machine2 ...]
The options must precede the list of machine names on the
command line. You may order the options in any way, but you
must include the parameters for each option. Include one space
between the options and between each option and the parameter for
that option. The options are:
Option Function
? Shows command line options. This option causes
NCSA Telnet to display a summary version
of the command line options instead of
initiating the program.
-c colorcode Sets the default color for screens. The
colorcode string is six digits long and follows the
coding scheme documented later in this chapter.
-h filename Specifies where to find the configuration file.
Normally, this option is included in TELNET.BAT,
and you never have to use it explicitly. If you do
not specify the configuration file there or on the
command line, there must be a file named
CONFIG.TEL in the default directory, or you must
set an environment variable named CONFIG.TEL
to your configuration file. If a suitable
configuration file cannot be found according to
these rules, NCSA Telnet terminates.
For example, if you name your configuration file
"TELNET.CFG", and the file resided in the
directory C:APPS, you would put the following
command in your AUTO.EXEC.BAT file:
"set CONFIG.TEL=C:\APPS\TELNET.CFG"
NOTE: filename should include a
pathname. For example,
C:\bat\config.tel
NCSA Telnet uses the filename specified for the -h
parameter first. If that filename isn't available, it
first looks in an environment variable, then the
current directory, then all the directories specified
by the "PATH" environment variable in order to
find the filename.
-s Enters server mode. NCSA Telnet waits for
external FTP and rcp requests. It allows you to
leave your PC and access files there from a remote
machine.
-t Disables direct writes to the screen. Without
the -t option, NCSA Telnet defaults to
writing directly to screen memory. This option
greatly enhances the screen speed, but makes
NCSA Telnet incompatible with Topview,
Microsoft Windows, and other windowing
systems. To make NCSA Telnet completely
compatible with BIOS-level programs, use the -t
option. You can install this option permanently
in the TELNET.BAT file or the CONFIG.TEL file
(bios option).
The Configuration File
Overview
The configuration file contains information on local operating
parameters (such as which type of graphics screen your PC has,
and your IP number), plus a list of commonly accessed hosts and
optional network tuning parameters for each of those hosts.
The configuration file is accessed once when the program is
initiated and is not used again. The program reads all of the
machine names into memory, so it may save memory to limit the
number of computers you specify in the file.
The configuration file is typically named CONFIG.TEL. If this
file is in the default directory, it will be found automatically, but
the use of the -h flag in the TELNET.BAT file is recommended.
Syntax for Entries
The configuration file is a list of keywords and values for those
keywords. The one overall requirement for the entries in the file is
that they alternate, keyword then value, keyword then value, and
so on. There are many different formats possible using any of the
allowable delimiters. The delimiters are the colon (:), semicolon
(;), equal sign (=), and any of the whitespace characters. To
include delimiters in a value field, enclose the field in double
quotes. Quotes cannot be a part of the actual data field. Wherever a
pound sign (#) is found, everything from there to the end of line is
a comment. Though you can use multiple formats in the same file,
you will probably want to find and keep a consistent format. For
example, each of the entries in Figure 7.1 specify the same
information.
Note that Figure 7.1.c is not a recommended format, but is
included to show how strange an entry may look.
Figure 7.1.aDisplaying the
Same Information
in Different Entry
Formats
name=nic # comment field to end of line ->
host=sri-nic.arpa
hostip=10.0.0.51
scrollback=300
contime=60
Figure 7.1.b
name=nic; host=sri-nic.arpa; hostip="10.0.0.51";
scrollback=300; contime=60
Figure 7.1.c
name
nic
host sri-nic.arpa : hostip=10.0.0.51; scrollback=300;
contime:60
PC Information Entries
The first entries in the configuration file are the PC environment
entries. They specify what types of hardware are to be used, the
names of graphics output files, and the network configuration
information. Following is a list of these entries:
Entry Meaning
address=d000 entered as four hex digits, address=d000,
is the segment address of the Ethernet
board's shared memory. The address only
applies to boards that have jumpers or
registers which can be used to set the
shared memory address. After setting the
board's jumpers to a particular value, use
the address parameter to configure NCSA
Telnet with the same setting. d000 (hex) is
a common address to use and it rarely
conflicts with other boards.
Entry Meaning
arptime=3 specifies how long to try to reach a host
on the local wire, in ticks, where 18.2 ticks
is equivalent to one second. A value of 20
works fine for the network at NCSA, but
larger values may be needed for hosts
that are slow to respond. Smaller values
are more convenient to use.
autoscroll=yes specifies whether the screen continues to
scroll in scrollback mode when the cursor
reaches the top of the screen or not. (It
will only scroll up one line if this option is
off.) Default is yes.
beep=yes specifies whether the musical note symbol
will appear in a session's indicator box
when NCSA Telnet receives a "beep"
character. This command is useful for
alert conditions.
bios=no indicates NCSA Telnet has two screen-
writing modes. One is direct-to-screen
(bios=no) which is considerably faster on
most machines. The other uses BIOS
(bios=yes) and follows the IBM BIOS
conventions at all times. BIOS mode
makes NCSA Telnet compatible with
popular windowing systems and can
reduce the amount of flicker and snow on
the screens of IBM CGA-equipped
systems.
broadcast=255.255. specifies the network address to use
255.255 for broadcast messages.
capfile= specifies a name for the capture file. For
example,
capfile ="c:\temp\myfile"
When you capture to disk, all data is
appended to this file. If no name is
included in the configuration file, "capfile"
is used by default.
capfile=prn specifies that the printer device (PRN)
should be used for the capture file.
clock=on specifies that the clock should be
displayed on the status line. Default is
clock=on.
concolor=431070 specifies the colors used in the console
window.
Entry Meaning
cursorbottom=8 specifies the bottom of the cursor block
using a number from 0 to 15.
cursortop=11 specifies the top of the cursor block using
a number from 0 to 15.
domain="ncsa.uiuc.edu" affects hostname lookups which are sent
to the domain nameserver. All names
which do not contain a period have the
default domain appended before the
nameserver request goes out. If the
hostname you type contains a period,
then the name is unmodified before NCSA
Telnet sends it to the nameserver. There is
no "ring-down" or multiple lookup feature
to try more than one form of the name.
domainretry=4 specifies number of times to query
domain nameserver(s). Each time a retry
is sent, the timeout value (above) is
doubled. Each time a retry occurs, NCSA
Telnet tries the next nameserver,
wrapping around to the first nameserver
when there are no more.
domaintime=20 specifies time in ticks to wait between the
first domain lookup and the second. If you
only have one nameserver, then this is the
same as a simple timeout. If you want to
rotate nameservers quickly because the
first one may be down, set this to a
smaller number. To force a larger number
of retries, set this to a smaller number. To
wait longer for a domain response,
increase this value.
ftp=yes enables FTP serving by default. Access to
your PC can be controlled by the FTP
password file. To disable FTP serving
completely, change this line to ftp=no.
hardware=3C501 specifies that NCSA Telnet can support
several different brands of Ethernet
adapters. Each brand has its own unique
identifier which must be set for the
hardware parameter. See the hardware
section in this chapter for a complete list
of settings.
hpfile=COM1 sends HPGL plotter codes directly to the
plotter, as long as a HP plotter hooked up
to your PC. Use the mode command to
preconfigure the serial port first.
hpfile=hp.out dumps Tektronix graphics to a file in
several formats from the Graphics menu.
When you write out HPGL (HP plotter)
codes, this file name is used. For example,
hpfile="c:\hpgrafs\hp.out"
The default is hp.out in the current
directory.
interrupt=3 specifies several choices for interrupt
vector on the Ethernet board. NCSA
Telnet defaults to IRQ3, but can be set to
use different interrupts by indicating the
interrupt number here. Match this
number with the interrupt setting from
your Ethernet board.
ioaddr=360 specifies the I/O base address of the
Ethernet board in hex. Some Ethernet
boards have selectable I/O addresses.
These generally range from 200 to 400
(hex) and must be entered as two, three,
or four digits.
keyfile=filename specifies an additional keyboard mapping
file to provide move key definitions. This
file over-rides the definitions in the
telnet.key file.
myip=BOOTP indicates that this entry is a special form
of the myip entry. Setting myip to
BOOTP will cause NCSA Telnet to query
a BOOTP server to get its IP address.
myip=RARP indicates that this entry is a special form
of the myip entry. Setting myip to RARP
will cause NCSA Telnet to query a RARP
server to get its IP address. See the section
on RARP in this chapter.
myip=10.0.0.51 specifies the IP address to use for the PC.
This number must contain all four parts
of the IP address. This field is required.
myname="hawkind.
ncsa.uiuc.edu" specifies the network name used to
identify the PC on the network.
netmask=255.255.255.0 specifies the subnet mask for your local
network. It is optional for networks
without subnets.
outputfile=filename specifies that this entry is the same as
keyfile option, except it specifies the
output mapping file to over-ride the
default telnet.out mappings.
Entry Meaning
passfile= specifies the file in which FTP usernames
and passwords can be found. For
example,
passfile="c:\bat\ftppass"
No default name exists for this file. If the
file is specified, then FTP will require a
username and password for all attempted
FTP connections. If the file is not
specified, then there is no password
checking for FTP. Use the program
TELPASS.EXE to encrypt new
passwords. See the FTP Password
Protection section later in this chapter.
psfile=ps.out prints PostScript commands to several
types of available laser printers. NCSA
Telnet's PostScript output has been tested
on the Apple (Sun) LaserWriter, which is
connected to Sun workstations. The
default filename is ps.out in the current
directory.
rcp=yes specifies that rcp serving is enabled by
default. To disable rcp serving, change
this line to rcp=no.
tek=yes specifies there is no overhead to unused
Tektronix graphics emulation. However,
you may wish to disable graphics
support. tek=no will disable Tektronix
graphics emulation.
tekfile=tek.out specifies the filename to use when you
write Tektronix codes to disk. The default
name is tek.out.Tektronix 4014 graphics
commands are a very compact way to
store a Tektronix graphics image.
video=ega specifies Tektronix 4014 graphics
emulation is included in NCSA Telnet for
a variety of video devices. Supported are:
hercules (Hercules monochrome
graphics), cga (IBM and compatibles
original Color Graphics Adapter), ega
(IBM Enhanced Graphics Adapter and
compatibles), ega 43 (ega 43 - line mode),
vga 50 (vga 50 - line mode) and no9
(Number Nine Computer's Revolution
512 x 8 board).
windowgoaway=yes specifies that a session doesn't wait for a
keypress after it has been closed. This
defaults to windowgoaway=no.
Entry Meaning
wire=thin specifies which type of ethernet cable to
use on the ethernet board. Options are
thin and thick. Use this command only for
3Com3C503 ethernet cards.
Host-Specific Parameters
Following the PC configuration options, you may specify zero or
more hosts, and host-specific information for each host. By
making the host entries in the configuration file, you can specify
color, backspace, scrollback, connection timeout, and several
tuning parameters for each host. You may want to have more than
one session name associated with a single computer, such as
purple cray and red cray with different screen colors to fit your
mood of the day.
Typically, the first host listed will be name=default, which stores
the default values for the other hosts. Any keyword listed under
later hosts will override the default setting for that host session.
NOTE: The keyword name is special because it separates entries.
The parameters following name up to the next keyword name are all
associated with the session name. The parameters are installed
whenever a connection is opened with that session name.
Entry Meaning
copyfrom=nic copies all unspecified parameters from a
previous session name. Note that the
session name which you want to
reference with copyfrom must appear
before the session name that uses a
copyfrom directive. Parameters that are
specified for a session name along with a
copyfrom directive override that
copyfrom directive.
clearsave=yes updates scrollback when the screen is
cleared. When clearing the screen, all of
the visible lines are saved into the
scrollback region. If you prefer not to
have the text saved when the screen
clears, set clearsave=no. In the case of
host programs which clear the screen one
line at a time, the lines are never saved
into the scrollback region.
contime=10 gives the connection timeout in seconds.
When you are making a connection
attempt, after this amount of time has
elapsed NCSA Telnet gives up on opening
the connection and deletes the window.
For congested or slow networks, this
value should be made larger.
Entry Meaning
crmap=4.3BSDCRNUL specifies a special compatibility option for
4.3 BSD UNIX. Now, a UNIX bug fix to
take care of the problem, but some hosts
may still want crnul to be used for end-of-
line. The default is crmap=crlf, which
sends CRLF when you press RETURN.
In line mode, CRLF is always used.
duplex=half applies only to hosts that negotiate non-
echoing mode and does not expect local
line editing. All character keys are sent
and echoed to the screen immediately.
This parameter has no effect in echo
mode.
erase=delete sets the backspace translation for this
host. Some hosts prefer the backspace key
to be DELETE and some prefer the
backspace key to be BACKSPACE. Set
this value erase=delete or
erase=backspace.
gateway=1 specifies the gateway precedence for this
host. To reach hosts not connected to your
local network, you must have at least one
gateway entry. The hostip keyword must
be present for this host. Gateway
numbers must start at 1 and increase by
ones. Gateway 1 has the highest
precedence, but the first gateway to
respond to an ARP will be used. ICMP
redirects can affect how gateways are
used, but not permanently.
host=sri-nic.arpa specifies the hostname or alternate name.
If you want to associate both a session
name and a hostname with a particular
set of parameters, you may include both.
Note that the name parameter is required,
while the host parameter is optional. The
rule of thumb is: When you have only a
hostname, insert it as name=hostname. If
you have both a session name and a
hostname, enter both name=sessionname
and host=hostname. When you want to
open a new connection, either hostname
or sessionname works.
hostip=10.0.0.51 gives the IP address of the host. If this is
not present, the domain nameserver must
be queried to get the IP number of the
host. For efficiency, include the IP
addresses of all commonly accessed hosts.
IP addresses of gateways and
nameservers must be in the configuration
file.
maxseg=512 gives a byte count (1-1024) of the largest
TCP segment that can be received. This
value can control the size of packets that
are sent over the connection. Reducing
this value can eliminate IP fragmentation
that we cannot reassemble. maxseg=512
should force the sending host to never
fragment.
mtu=512 specifies a byte count (1-1024) of the
largest amount of data to put in the
packets that are sent. If you are sending to
the ARPANET you should use mtu=512. If
you are sending to local hosts, you should
use mtu=1024.
name=nic specifies the name that you want to
appear on the bottom line of the screen
when the connection opens. It is the
primary name associated with a list of
parameters. It is common to have more
than one session name for a host, each
with different parameters, perhaps with
different colors or different amounts of
scrollback. This parameter is required
because it separates entries.
nameserver=1 specifies the nameserver precedence for
this host. NCSA Telnet uses UDP to query
domain nameservers for machinenames
that are not in the configuration file. Each
machine that is to be used as a
nameserver must have this keyword
listed. The hostip keyword must be
present for this host. Nameserver 1 has
the highest precedence. Nameserver
numbers must start at 1 and increase by
ones.
nfcolor=white specifies a normal, foreground
nbcolor=black specifies a normal, background
rfcolor=black specifies a reverse, foreground
rbcolor=white specifies a reverse, background
ufcolor=blue specifies a underline, foreground
ubcolor=black specifies a underline, background
You can use these new color options
instead of the more complicated color
code from previous versions. For each
value, specify a string from the following
list of colors available on IBM-PC class
machines with color displays:
Also for machines with EGA or better
graphics adapters, the following colors
are available:
BLACK
BLUE
GREEN
CYAN
RED
MAGENTA
YELLOW
WHITE
The colors are in all caps, and for the
foreground colors they are the
highlighted version of the lowercase
colors. For background colors, they make
the foreground blink.
The foreground and background colors
are combined to create the text
appearance on the screen. You may wish
to experiment with the Parameter menu
(Chapter 3), which interactively allows
you to select a color scheme. The previous
version's color= entries are still supported.
retrans=7 gives the initial retransmit timeout in
18ths of a second. Increasing the value of
this parameter may help in reducing the
initial burst of retries that is typical of
connections with high round-trip times.
rwin=512 gives a byte count (1-4096) of the largest
size of the TCP window to advertise to
other hosts. Unfortunately, some
hardware and networks, the 3COM
3C501 Etherlink for example, cannot
handle receiving back-to-back packets.
This situation requires us to limit the TCP
receive window that we advertise to other
hosts. For communicating to slower hosts,
or when using the other brands of
Ethernet boards, a larger window (4096)
may work better.
scrollback=100 specifies the number of lines of scrollback
for this session. Be aware that scrollback
occupies at least 86 bytes per line saved.
There can be a different number of lines
of scrollback for each session. Plan your
use of scrollback wisely unless you have
memory to spare.
Converting UNIX /etc/hosts
Files
Included with the NCSA Telnet distribution is an awk script called
newh. Used with the following command under 4.X BSD UNIX, it
will convert the /etc/hosts file format into a format compatible
with NCSA Telnet's configuration file. Note that domain name
lookup should make this operation obsolete, or apply to only a small
subset of your /etc/hosts file.
% awk -f newh /etc/hosts >config.temp
After creating this new file, prepend the PC-specific information
and download the file to the PC.
Hardware Options
Combined Ethernet Drivers
Only one distribution program for NCSA Telnet Version 2.3
exists. All of the Ethernet and video device drivers are combined
into one executable program. You must use the hardware entry in
the configuration file to inform NCSA Telnet which Ethernet board
you have installed. Choose from the list of supported Ethernet
boards in Table 7.1; the value to use is case insensitive. The
requirements for interrupt, address, and ioaddr entries in the
configuration file are listed.
Table 7.1 Ethernet Boards
Supported by NCSA
Telnet
Value Ethernet Board Requires╩╩╩╩╩╩╩╩╩╩
For PC bus
3C501 3COM 3C501 Etherlink interrupt,ioaddr
3c503 3COM3C503 interrupt, address,
ioaddr, wire
3c505 3COM 3C505 interrupt, address,
ioaddr
atalk PCAppletalk Card (software)
interrupt,
address, ioaddr
bicc BICC 411x Ethernet Cards interrupt, address,
ioaddr
decnet DEC Decnet Protocol interrupt, address,
ioaddr
NI5210 MICOM NI5210 interrupt, address,
ioaddr
packet Any packet driver interrupt
**also for
PS/2 MCA Bus
PCNIC Ungermann-Bass PC-NIC address,ioaddr
or IBM Baseband adapter
starlan AT&T Starlan IO Card address, oiaddr
WD8003 Western Digital WD8003 address,ioaddr
Table 7.1 Ethernet Boards
Supported by NCSA
Telnet (continued)
Value Ethernet Board Requires╩╩╩╩╩╩╩╩╩╩
WD8003e Western Digital WD80033e address,ioaddr
For PS/2 MCA bus
NICps2 Ungermann-Bass NICps/2 address
3C523 3COM 3C523 Etherlink/MC address,ioaddr
WD8003a Western Digital WD8003a address, ioaddr
Performance Tuning
The values of maxseg, mtu and rwin in the configuration file must
be set correctly to get maximum data transfer throughput between
machines. Some rules of thumb to use when setting these values
include the following:
Ñ The maximum reasonable values for these parameters are:
rwin=4096
mtu=1024
maxseg=1024
Ñ The best setting for local network use, with a high performance
Ethernet board is:
rwin=4096
mtu=1024
maxseg=1024
Ñ The best setting for ARPANET use, or any situation with a lot of
unknown gateways, but with a high performance Ethernet board
is:
rwin=4096
mtu=512
maxseg=512
Ñ The setting required for the 3COM 3C501, and any other
troublesome network situation, also the most conservative
setting, is:
rwin=512
mtu=512
maxseg=512
rwin specifies what the other computer is allowed to send you at any
one time, so it depends mostly upon your local Ethernet board. If the
board can handle it, always set rwin=4096. maxseg is used to
avoid fragmentation. If you get fragmented packets, lower the
value of maxseg for that host until fragmentation stops occurring.
RARP for Dynamic IP
Address Assignment
When the following special form of the myip entry is used,
myip=RARP, NCSA Telnet looks up your local machine's IP address
with a broadcast request to the network. The "Reverse Address
Resolution Protocol" is used to attempt to translate the local
machine's assigned 48-bit Ethernet address to a corresponding IP
address. From our tests with the RARP server from Sun
Microsystems' SunOS 3.X, the RARP lookup takes about 3 seconds.
You must install the 48-bit Ethernet address from each PC in the
RARP hosts' tables before RARP will work. The value of arptime,
in seconds, multiplied by three, is used for a timeout value.
NOTE: Sun's RARP lookup will not work if you have a double-zero
in your Ethernet address. Other leading zeros can be dropped as
well. Use a single zero, i.e. 8:0:20:1:5a:90, instead of
08:00:20:01:5a:90.
BOOTP for Dynamic IP
Address Assignment
The Bootp process, myip=RARP, closely resembles the RARP process.
For instance, the former performs the same function as RARP by
getting an IP number from a Bootp server which has the PC's
Ethernet address registered in the Bootp Host table. Bootp and
RARP time-out procedures are also the same.
Appletalk Hardware
Using an Appletalk network involves some special
considerations. First, you must load the Appletalk driver into
memory. Version 1.0 of the "ATALK.EXE" driver was used in the
development of NCSA Telnet.
The second consideration involves the "interrupt=" line. The
"interrupt=" line in your CONFIG.TEL line refers to the
software interrupt the Appletalk driver is using, not the hardware
interrupt the card is set to. For example, if your Appletalk card is
set to IRG2, you would not set the "interrupt=" line to "2".
Instead, the value should be set to the software interrupt, usually
"interrupt=60" or "interrupt=5C".
Static IP addressing does not work at the current time in NCSA
Telnet 2.3 using the Appletalk driver. Therefore, NCSA Telnet
ignores any IP address you set in your CONFIG.TEL file , and
assigns an IP address to your PC by the Appletalk gateway.
There has been no success in getting the Appletalk packet driver to
work with NCSA Telnet. In addition, NCSA Telnet has not been
tested with Appletalk cards for MicroChannel machines, such as
IBM's PS/2 line.
Interrupts
When you configure the 3COM hardware, an important option is
the choice of interrupt request number (level). 3COM boards arrive
set for IRQ3, which corresponds to the PC-DOS definition of the
COM2 communications port. If you have a port configured as
COM2, you cannot set your Ethernet board to IRQ3. Another
common value to use is IRQ5, which is the same interrupt that DOS
uses for LPT2. Make sure that your PC hardware configuration
does not conflict in any way.
Interrupt Conflicts
The most common hardware problem is interrupt conflicts. PCs
have a limited number of interrupts, and you cannot have two
devices which expect to use the same interrupt. PC Ethernet boards
almost always consume an interrupt, even though we do not use
interrupts for all Ethernet boards. Interrupts are also used by COM
ports and LPT ports in your machine. You must check for
conflicts.
Intermittant Interrupts
Interrupt conflicts don't always show up right away. They are
software controlled. If you do not activate the interrupt on your
COM2 port, it may not conflict with a 3C501 board on IRQ3 until
next week when you try to use a modem terminal emulator and
then switch to NCSA Telnet. NCSA Telnet might work fine until
you try to FTP to your hard disk; then disaster strikes.
Interrupts on Hard Disks
Hard disks use interrupts too. We have traced two serious hard
disk problems to interrupt conflicts. IRQ2 is the default setting for
some Ethernet boards when they come out of the box. On a PC-AT,
this interrupt is used for the second drive controller, and a conflict
can shut off the whole second controller which includes the hard
disk. On the PC/XT and PC compatible computers, IRQ5 is
sometimes assigned to the hard disk rather than LPT2. Strange
hard disk problems like disk write errors may be caused by the
Ethernet conflicting with the hard disk.
Moral: Set your Ethernet board to an unused interrupt.
Domain Name Lookup
When NCSA Telnet cannot find a name in the configuration file,
it can still find the IP number if you are running a domain
nameserver. To use the domain system, at least one host must be
running the nameserver and the configuration file must have a
nameserver entry for that host. There may be more than one. If
one nameserver fails to respond, the one with the next higher
precedence is queried. As soon as a response is received, NCSA
Telnet will add its name to the status line and attempt to open a
telnet connection.
When you enter a name to open a connection, NCSA Telnet follows
specific search order:
1. It looks up the name as a session name from the configuration
file.
2. It looks up the name as a hostname from the configuration file.
3. It sends the name as a domain query to the first nameserver.
4. If the domain request times out, the program repeats the query,
but to another nameserver. NCSA Telnet repeats this process
until it reaches the maximum number of retries or receives a
response.
With the domain nameserver, the number of hosts in the
configuration file can be kept to a minimum. Each host in the
configuration file will be a commonly used computer that may
have special screen color or scrollback requirements. The IP
addresses for rarely used hosts will be accessible if the domain
name retrieval system can resolve those hosts.
FTP Password Protection
The presence of the passfile keyword in the configuration file
enables FTP password protection. If you have a password file, FTP
will not allow any FTP connections to open without a correct
username and password. You can have several usernames and
individual passwords for each user. The password file can be
easily read by typing out the file. The passwords are encrypted, but
not with a secure encryption system. Only trusted users should be
allowed access to the password file.
NOTE: Precautionary measures are especially important now
that the FTP server allows the person connected to the PC to create
and remove directories and files.
To create or change passwords for users, run a separate program
called Telpass. This program reads in the password file so you
can add and change users and passwords. There are no special
installation instructions for Telpass as it is menu-driven. Start up
Telpass from DOS with the name of the password file that you wish
to edit.
C:\ telpass pwfile
Telpass automatically saves the list of names and passwords when
you exit. Just add a passfile option to your configuration file to
enable FTP passwords.
Compatibility Notes
Ping
NCSA Telnet responds to ping (ICMP echo) requests. Other hosts
use this command by other hosts to determine whether your PC is
online.
VT102
The VT102 emulator is nearly complete. VT102 features not
emulated are double width and double height characters, VT52
mode, origin mode for cursor positioning, and 132 column mode.
Included in the emulator are variable spaced tabs, graphics
characters, keypad modes, and character attributes.
FTP
The FTP server in NCSA Telnet is close to the DARPA
specification of the minimum implementation. Exceptions are the
following:
Ñ The command connection does not perform telnet negotiation.
Ñ The block mode of FTP is not supported.
Ñ Some error conditions may display as command not
understood instead of returning more appropriate messages.
FTP has not been thoroughly tested on low-bandwidth or low-
reliability networks. It should be most successful on local
networks, with unknown throughput for long-haul or satellite
networks.
Terminate and Stay
Resident (TSR) Programs
Programs such as SideKick from Borland International Inc. take
complete control of the PC when they are initiated with key
sequences. The use of these programs does not crash NCSA Telnet,
but it suspends the network communications capability while the
TSR program is active. When network communications are
suspended, whether from the ALT-E DOS escape feature or from
using TSR programs, it takes the host approximately one minute to
decide that your PC has crashed, and then it cuts you off. As long as
these programs are used for short (less than 30 seconds) intervals,
you should have no problem.
NCSA Telnet
The standard telnet protocol has several potential options that can
be invoked if both parties of the telnet connection agree. NCSA
Telnet refuses most of these options, but accepts the following:
echo, option 1; suppress go ahead, option 3, termtype, option 24,
NAWS (Negotiate About Window Size), option 31; and Linemode,
option 34. Some obscure features of NCSA Telnet are not
implemented in this version: out-of-band interrupts are not
available, go ahead signals do nothing, and telnet acknowledge
signals are not acknowledged. If there are any problems with the
limitations of NCSA Telnet, please submit a bug report.
Limitations
There is a 20-session limit for NCSA Telnet. This limit is
somewhat arbitrary and may be altered if there appears to be a need
for it. NCSA Telnet can run out of memory, often before the 20-
session limit. You are notified on the screen if this happens.