|HP-UX Reference > R
HP-UX 11i Version 3: February 2007
remsh, rexec — execute from a remote shell
remsh connects to a specified host and executes a specified command. The host name can be either the official name or an alias as understood by gethostbyname() (see gethostent(3N) and hosts(4)). remsh copies its standard input (stdin) to the remote command, the standard output of the remote command to its standard output (stdout), and the standard error of the remote command to its standard error (stderr). Hangup, interrupt, quit, terminate, and broken pipe signals are propagated to the remote command. remsh exits when the sockets associated with stdout and stderr of the remote command are closed. This means that remsh normally terminates when the remote command does (see remshd(1M)).
By default, remsh uses the following path when executing the specified command:
remsh uses the default remote login shell with the -c option to execute the remote command. If the default remote shell is csh, csh sources the remote .cshrc file before the command. remsh cannot be used to run commands that require a terminal interface (such as vi) or commands that read their standard error (such as more). In such cases, use rlogin or telnet instead (see rlogin(1) and telnet(1)).
The remote account name used is the same as your local account name, unless you specify a different remote name with the -l option. This remote account name must be equivalent to the originating account. In addition, the remote host account name must also conform to other rules, which differ depending upon whether the remote host is operating in a Kerberos V5 Network Authentication, i.e., secure environment, or not.
In a non-secure, or traditional environment, the remote account name must be equivalent to the originating account; no provision is made for specifying a password with a command. For more details about equivalent hosts and how to specify them, see hosts.equiv(4). The files inspected by remshd on the remote host are /etc/hosts.equiv and $HOME/.rhosts (see remshd(1M)).
In a Kerberos V5 Network Authentication environment, the local host must be successfully authenticated before the remote account name is checked for proper authorization. The authorization mechanism is dependent on the command line options used to invoke remshd on the remote host (i.e., -K, -R, -r, or -k). For more information on Kerberos authentication and authorization see the Secure Internet Services man page, sis(5) and remshd(1M).
Although Kerberos authentication and authorization may apply, the Kerberos mechanism is not applied to the command or to its response. All the information that is transferred between the local and remote host is still sent in cleartext over the network.
The default Kerberos options for the applications are set in the krb5.conf configuration file. Refer to the appdefaults Section in the krb5.conf(4) manpage for more information. The options -f, and -F described in the subsequent paragraphs, can be set in the krb5.conf file with the tag names forward, and forwardable respectively. Refer to the krb5.conf(4) manpage for more information on the appdefaults Section.
The fallback option can be set in the krb5.conf file within the appdefaults Section. If fallback is set to true and the Kerberos authentication fails, remsh will use the non-secure mode of authentication.
Note: Command line options override the configuration file options.
In a secure or Kerberos V5-based environment, the following command line options are available:
If command is not specified, instead of executing a single command, you will be logged in on the remote host using rlogin (see rlogin(1)). Any rlogin options typed in on the command line are transmitted to rlogin. If no command and the option -P are specified, rlogin will be invoked with -P to indicate that Kerberos authentication (or secure access) is not necessary. This means that if a password is requested, the password will be sent in cleartext. If command is specified, options specific to rlogin are ignored by remsh.
By default, remsh reads its standard input and sends it to the remote command because remsh has no way to determine whether the remote command requires input. The -n option redirects standard input to remsh from /dev/null. This is useful when running a shell script containing a remsh command, since otherwise remsh may use input not intended for it. The -n option is also useful when running remsh in the background from a job control shell, /usr/bin/csh or /usr/bin/ksh. Otherwise, remsh stops and waits for input from the terminal keyboard for the remote command. /usr/bin/sh automatically redirects its input from /dev/null when jobs are run in the background.
Host names for remote hosts can also be commands (linked to remsh) in the directory /usr/hosts. If this directory is specified in the $PATH environment variable, you can omit remsh. For example, if remotehost is the name of a remote host, /usr/hosts/remotehost is linked to remsh, and if /usr/hosts is in your search path, the command
executes command on remotehost, and the command
is equivalent to
The rexec command works in the same way as remsh except that it uses the rexec() library routine and rexecd for command execution (see rexec(3N) and rexecd(1M)) and does not support Kerberos authentication. rexec prompts for a password before executing the command instead of using hosts.equiv for authentication. It should be used in instances where a password to a remote account is known but there are insufficient permissions for remsh.
Shell metacharacters that are not quoted are interpreted on the local host; quoted metacharacters are interpreted on the remote host. Thus the command line:
remsh otherhost cat remotefile >> localfile
appends the remote file remotefile to the local file localfile, while the command line
remsh otherhost cat remotefile ">>" otherremotefile
appends remotefile to the remote file otherremotefile.
If the remote shell is /usr/bin/sh, the following command line sets up the environment for the remote command before executing the remote command:
remsh otherhost . .profile 2>&- \; command
The 2>&- throws away error messages generated by executing .profile when stdin and stdout are not a terminal.
The following command line runs remsh in the background on the local system, and the output of the remote command comes to your terminal asynchronously:
remsh otherhost -n command &
The background remsh completes when the remote command does.
The following command line causes remsh to return immediately without waiting for the remote command to complete:
remsh otherhost -n "command 1>&- 2>&- &"
remsh otherhost -n "sh -c \"command 1>&- 2>&- &\""
If remsh fails to set up the secondary socket connection, it returns 2. If it fails in some other way, it returns 1. If it fully succeeds in setting up a connection with remshd, it returns 0 once the remote command has completed. Note that the return value of remsh bears no relation to the return value of the remote command.
Besides the errors listed below, errors can also be generated by the library functions rcmd() and rresvport(). In the case of IPv6 systems, the library functions rcmd() and rresvport() are replaced by rcmd_af() and rresvport_af(); respectively, and can generate errors (see rcmd(3N) and rcmd_af(3N)). These errors are preceded by the name of the library function that generated them. remsh can produce the following diagnostic messages:
Kerberos-specific errors are listed in sis(5).
For security reasons, the /etc/hosts.equiv and .rhosts files should exist, even if empty, and they should be readable and writable only by the owner. Note also that all information, including any passwords asked for, is passed unencrypted between the two hosts.
If remsh is run with an interactive command, it hangs.
remsh is the same service as rsh on BSD systems. The name was changed due to a conflict with the existing System V command rsh (restricted shell).