htsh/wapsh - Installation and Administration Guide
Copyright exolution gmbh / Michael Kerrisk, Munich, Germany
Version 1.0, last revised 21 Nov 2000
1. Introduction
2. Installation
2.1. Software requirements
2.2. HTTP Server Host Setup
2.2.1. Create a loginhosts file (optional)
2.2.2. Create an htsh service in /etc/services
2.3. htshd (htsh server daemon) Setup and execution
2.3.1. Create an htsh service in /etc/services
2.3.2. Building the htshd application daemon
2.3.3. htshd installation
2.3.4. htshd startup
2.4. htshd initialisation
2.4.1. htshd configuration file
2.4.2. htshd command line options
2.5. Global user initialisation file
2.6. htshd exit status
3. Security Features of htsh
1. Introduction
For an overview of the operation of
htsh, please read the htsh User Guide before reading this document.
2. Installation
2.1 Software requirements
The following software is required to run htsh:
- Apache web server (installed on htsh HTTP Server Host)
- SSL certificate for htsh (optional)
- PHP 4.0.2 or later (installed on htsh HTTP Server Host)
- htsh web-server application (installed on htsh HTTP Server Host)
- htshd (htsh server daemon) (installed on htsh Login Host)
2.2 HTTP Server Host Setup
2.2.1 Create a loginhosts file (optional)
By default, the HTTP Server and the htsh server daemon (htshd) must reside on the same machine (i.e. localhost). This means that users can only login on the machine running the HTTP Server. If you wish to place the HTTP Server and htshd on separate machines, or to allow the HTTP Server to establish login connections to other Login Hosts in addition to localhost, then two steps must be performed:
- When starting the htshd program the set of HTTP Server addresses from which login connections will be accepted must be listed either in the htshd configuration file or on the htshd command line. This step is described in detail later in this guide.
- In the directory on the HTTP Server which contains the htsh web application, create a file called loginhosts in the htsh web directory. This file lists the Login Hosts on which logins are permitted. This step is described in detail in the remainder of this section.
The loginhosts file lists the Login Hosts on which the htsh web application will allow users to login. (These hosts are displayed as a selectable list in the Client Browser on the htsh login page, unless the list consists of just one host, in which case a selectable list would be redundant, and is therefore not displayed.) The format of each line in this file is as follows:
[alias] address
The alias specifies the name to be displayed to the user on the htsh login page. The address is the DNS name or numeric dot address to which this name corresponds. If alias is omitted then the address value is displayed in the htsh login dropdown list. Empty lines and lines beginning with a hash (#) character in the loginhosts file are ignored. Here is a sample loginhosts file:
freyr
localhost 127.0.0.1
mimir 192.168.1.61
jord jord.exolution.lan
2.2.2 Create an htsh service in /etc/services
A service named htsh must be created in the file /etc/services on the HTTP Server host. The port number specified for this service must match the port number specified for the htsh service on the Login Host.
2.3 htshd (htsh server daemon) Setup and execution
The following steps must be performed in order to set up the htshd server.
2.3.1 Create an htsh service in /etc/services
A service named htsh must be created in the file /etc/services on each Login Host. (If the Login Host and the HTTP Server Host are the same machine, then only a single /etc/services file needs to be updated).
2.3.2 Building the htshd application daemon
A short shell script (Build_htshd) is provided which checks the flavour of Unix in use, as well as a few other system options, and then compiles and links the htshd program.
2.3.3 htshd installation
Just copy the htshd executable produced by the previous step into any desired directory, for example /usr/sbin.
2.3.4 htshd startup
There are two ways to start the htsh server daemon.
Startup via inetd(8)
This is the preferred method for starting htshd. Creating an inetd entry for the htsh service, ensures that htshd is automatically started whenever a user tries to connect to the service. To configure htshd to be run by inetd, add a line of the following form to /etc/inetd.conf:
htsh stream tcp nowait root /some-path/htshd htshd cmd-line-args
Where some-path should be the pathname of the directory where htshd resides, and cmd-line-args are any desired command line arguments (as defined below). (Note that some versions of inetd place quite a short limit on the number of command line arguments that may be specified in inetd.conf. If you need to specify more command line arguments than are permitted, you should instead use the htshd configuration file (described below) to specify startup options for htshd.
Manual startup
While logged in as root, it is possible to simply run htshd from a shell prompt. Simply enter the command:
# htshd
In this case htshd automatically becomes a daemon (unless the -D n command line option or the runasdaemon n configuration option is specified). Starting htshd manually may be useful for running htshd in debug mode (command line option -d 7777) or for one-off testing of the operation of htshd.
2.4 htshd initialisation
Upon startup, htshd first processes its configuration file (if present) and then processes the htshd command line options. Running the daemon without specifying a configuration file or specifying any command line options results in the following defaults:
- Login requests are accepted only from an HTTP Server residing on the same host (i.e. localhost) as htshd.
- Any user with a valid username/password may login on the Login Host.
Most initialisation options can appear in either the configuration file, or on the htshd command line. Note that command line options may override configuration file options, since the command line is processed after the configuration file.
Note that certain htshd configuration file options and command line options have corresponding commands which may be used in the (global or user specific) htsh initialisation files to set the same values on a per-user basis. Since the initialisation files are processed at the time a user logs in, settings in these files will override corresponding htshd configuration file and command line options. The htshd options which may be overridden in the htsh initialisation files are noted below.
2.4.1 htshd configuration file
The htsh configuration file can be used to specify configuration options for htshd. By default htshd expects to find the configuration file at /etc/htshd.conf. If desired, an alternate location can be specified using the -f command line option to htshd, or processing of the configuration file can be skipped entirely using the -F option. If the default configuration file is missing, and no alternate file is specified, htshd silently skips configuration file processing.
Configuration file format
- The configuration file is line oriented. Commands consist of series of an initial keyword followed by one or more space-delimited arguments.
- Blank lines, and comment lines (those whose initial character is "#") are ignored.
- Lines (other than comment lines) can be continued onto the following line by preceding the newline with a backslash. (The backslash and newline are removed.)
Configuration File commands
The following commands may be placed in the configuration file. Each command may be repeated multiple times. In the case of commands setting a single value, only the last instance of the command will have effect. For those commands (httpserver and user) which can take a list of arguments, all arguments across all instances of the command are accumulated into a single set.
allowedprotocols proto-name... |
Specify the set of protocols which may be used to login to htsh. Permitted protocol names are (lowercase) "wap" and "http".
The default is to allow logins using any protocol.
This option corresponds to the set csmaxtransfersize command within either of the htsh initialisation files. Note that the initialisation file command does not override this configuration file option - instead the initialisation file command can only be used to reduce the available protocols from the set specified by this configuration file option.
Example: allowedprotocols http
|
csmaxtransfersize nbytes |
Specify the largest number of bytes that will be transferred in a single block by the Client Shell to the HTTP Server. This is useful to prevent large outputs from choking the browser client. Otherwise, the HTTP Server/Client Browser could be inundated with large amounts of output and the user would be prevented from sending further input (for example a Control-C to abort the command generating the output) until all of the output has been completed.
This option corresponds to (and is overridden by) the set csmaxtransfersize command within either of the htsh initialisation files. Note however, that this configuration file option specifies an upper limit on the value which can be set via the corresponding initialisation file command.
Default = 10000 bytes.
Attempts to set this value less than 1000 bytes result in a setting of 1000 bytes.
Example: csmaxtransfersize 5000
|
csoutputtimeout nsecs |
Specify the time for which the Client Shell will wait for any further shell output (after receiving shell input) before informing the HTTP Server that output is complete in response to the most recent shell input. To avoid slow response times, this should be set to some small value (usually less than one second).
This option corresponds to (and is overridden by) the set csoutputtimeout command within either of the htsh initialisation files.
Default is 0.5 seconds
Example: csoutputtimeout 0.2
|
globalprofile pathname |
Identifies name of global initialisation file to be executed by all htsh logins. This can be used to set options and shortcuts defined for all users.
Default is to use (if present) /etc/htsh_profile
|
httpserver server-addr... |
Specify the (HTTP Server) hosts from which the htshd daemon will accept login requests and shell input messages.
This option may be specified repeatedly: the set of permitted servers is the union of all servers listed in all httpserver configuration file options and using the -H command line option to htshd.
The HTTP Server addresses may specified as symbolic host names or in Internet dot-address notation.
Default is to only allow connections from localhost (127.0.0.1), so that the HTTP Server and the htshd server daemon must reside on the same host.
Example: httpserver 192.168.1.110 localhost
|
logfiledirectory path |
Specify a directory into which script(1) type logs of each login session will be recorded.
Default is to create no log files.
Example: logfiledirectory /root/htsh
|
outputbufferlimit nbytes |
Set the upper limit (in bytes) for the size of the buffer used to record shell output. htsh maintains a separate output buffer for each login session.
Default is 100000 bytes.
This option corresponds to (and is overridden by) the set outputbufferlimit command within either of the htsh initialisation files. Note however, that this configuration file option specifies an upper limit on the value which can be set via the corresponding initialisation file command.
Example: outputbufferlimit 200000
|
port port-number |
Specifies the TCP port number (or TCP service name in /etc/services) to be used by htshd to listen for login requests.
The default is to use the port specified in /etc/services under the TCP service name htsh
Example: port 60001
|
runasdaemon {y|n} |
Specifies whether to run htshd as a background daemon. By default, htshd is run as a daemon. The only reason not to run htshd as a daemon is to for debugging purposes: if htshd is run as a foreground process, debugging output will be sent to standard error instead of the system error logger.
If started via inetd(8), then htshd is automatically made a daemon, and this option has no effect.
|
shelltimeout num-secs |
Specify (in seconds) the timeout to be used whenever the Client Shell waits for further shell input from the Client Browser.
If this timeout is exceeded then the Client Shell automatically terminates. This facility is required to deal with the possibility that the user shuts down their browser without doing a logout.
This option corresponds to (and is overridden by) the set shelltimeout command within either of the htsh initialisation files.
Default = 1800 seconds (30 minutes)
Example: shelltimout 600 (10 minutes)
|
user username... |
Specify users who is allowed to login via htsh. This option may be specified repeatedly: the set of permitted users is the union of all users listed in all user configuration file options and using the -u command line option to hthsd.
If this option is not given, then any user with a valid username/password on the Login Host is permitted to login.
Example: user mtk thk hat |
2.4.2 htshd command line options
The following command line options are permitted for the htshd command. Most of these options can also be set using corresponding configuration file commands.
-b nbytes |
Set the upper limit (in bytes) for the size of the buffer used to record shell output.
This corresponds to (and overrides) the outputbufferlimit configuration file option (and further information can be found under the description of that option).
Example: -b 100000
|
-c nbytes |
Specify the largest number of bytes that will be transferred in a single block by the Client Shell to the HTTP Server.
This corresponds to (and overrides) the csmaxtransfersize configuration file option (and further information can be found under the description of that option).
Example: -c 10000
|
-d octal-mask
|
Run in debugging mode. Voluminous debugging output (according to the given octal-mask value) is sent either to the system error logger (if running as a daemon) or standard error (if running as foreground process).
|
-D [y|n] |
Specifies whether to run htshd as a background daemon.
This corresponds to (and overrides) the runasdaemon configuration file option (and further information can be found under the description of that option).
Example: -D y
|
-f config-file |
Specifies an alternate name for the file which contains configuration options for htshd. (This file represents an alternative/and or complementary way of setting options that are set on the htshd command)
Default is to use (if present) /etc/htshd.conf
|
-F |
Do not execute the default configuration file.
htshd will give an error, and fail to start, if you specify both the -f and -F options.
|
-g global-rc-file |
Identifies name of global initialisation file to be executed by all htsh logins. This can be used to set options and shortcuts defined for all users.
This corresponds to (and overrides) the globalprofile configuration file option (and further information can be found under the description of that option).
Default is to use (if present) /etc/htsh_profile
|
-h |
Show usage message, including default values used for configuration settings
|
-H http-server-addr |
Specify the (HTTP Server) hosts from which the daemon will accept login requests and shell input messages. This option may be specified repeatedly, permitting connections to be made from multiple HTTP Servers.
This corresponds to the httpserver configuration file option (and further information can be found under the description of that option).
Example: -H 192.168.1.10 -H localhost
|
-L path |
Specify a directory in which script(8) type logs of each login session will be recorded.
This corresponds to the logfiledirectory configuration file option (and further information can be found under the description of that option).
Example: -L /root/htsh
|
-o output-resp-delay |
Specify the time for which the Client Shell will wait for any further shell output (after receiving shell input) before informing the HTTP Server that output is complete in response to the most recent shell input.
This corresponds to (and overrides) the csoutputtimeout configuration file option (and further information can be found under the description of that option).
Example: -o 0.2
|
-p port-num |
Specifies the TCP port number (or TCP service name in /etc/services) to be used by htshd to listen for login requests.
This corresponds to (and overrides) the port configuration file option (and further information can be found under the description of that option).
Example: -p 60000
|
-P 'protocol-name...' |
Specify the list of protocols which may be used to login to htsh. If more than one protocol is specified, then the protocol names should be separated by spaces, and enclosed in quotes (to prevent word splitting by the shell). Permitted protocol names are (lowercase) "wap" and "http".
This corresponds to (and overrides) the allowedprotocols configuration file option (and further information can be found under the description of that option).
Example: -P 'wap http'
|
-t numsecs
|
Specify (in seconds) the timeout to be used whenever the Client Shell waits for further shell input from the Client Browser.
This corresponds to (and overrides) the shelltimeout configuration file option (and further information can be found under the description of that option).
Example: -t 600 (10 minutes)
|
-u username |
Specify a user who is allowed to login via htsh. This option may be specified repeatedly, thus allowing a set of users to be specified. If this option is not given, then any user with a valid username/password on the Login Host is permitted to login.
This corresponds to the user configuration file option (and further information can be found under the description of that option).
Example: -u mtk -u thk -u ht
|
2.5 Global user initialisation file
When each user logins, htsh processes two files which specify how the htsh session should be specified. These are:
- The global initialisation file executed by all users
- The user-specific initialisation file (~/.htshrc).
The first of these files (by default the file /etc/htsh_profile) can be used by the system administrator to create shortcuts and set options which apply to all htsh logins. For details on the format of this file, see the htsh User Guide.
System administrators should take special note of the set +o allowuserinit initialisation file command which disables the processing of user-specific login initialisation files. This command can be used to create "captive" htsh logins whose shortcuts and other settings are defined by the administrator (in the htsh global initialisation file) and cannot be changed by users.
2.6 htshd exit status
htshd may exit with the following status values
0
|
Success (only occurs after displaying usage message in response to the -h option to htshd) |
1
|
Error during startup |
2 |
Error in configuration file |
3. Security Features of htsh
The following points regarding security are noteworthy:
- To allow secure transmission of data across the Internet, an SSL certificate should be installed on the HTTP Server.
- It is possible for the htsh server daemon (htshd) on the Login Host to handle login requests from multiple HTTP Servers which may be on separate hosts from the Login Host. By default, connections are only accepted from an HTTP Server on the same computer as the htsh daemon. When starting the htsh daemon it is possible to specify a series of HTTP Servers from which the daemon will accept login requests.
- To prevent "spoofing" of shell input from malicious processes on an HTTP Server, at login, htsh generates a random authorisation key which is passed back to the HTTP Server. This authorisation must be transmitted by the HTTP Server with every subsequent shell input request. The Client Shell will reject any input which does not provide the correct authorisation key.
- For high security environments making use of WAP access, it is possible to install a WAP gateway on the same host as the HTTP Server, eliminating the need to transmit data across the Internet.
|