When you use ssh to connect to the TACC machines, if you have not placed your public key on that system (see the ssh manual page for details), the login processor will prompt you for your password, for example:

login@lonestar's password:

Enter your login password and press <RETURN>. The system will match the password you supply with the encrypted password stored in the password file to validate your access. If you have placed your public key file on the supercomputer, then sshd will prompt you for your passphrase, as shown below:

Enter passphrase for RSA key 'login@your-workstation':

Type in the passphrase that you used when you generated your ssh key. Once your access has been validated, the system will present a banner that looks similar to this:

Assuming your authentication was successful, the login processor will set some default environment variables (like $HOME, the path to your home directory as given in the password file) and will spawn a command shell for you using the name of your command shell as given in your password file entry. When your shell launches, it will look for its start-up files in your $HOME directory and, if these are found, execute the commands therein, allowing you to customize your setup as you wish. When the shell's start-up process completes, it will prompt you for a command and you may begin working.

When a new project is awarded time on a TACC supercomputer, we will create a login for the principal investigator (PI) and inform him or her of the password when we send notice of the award. Once the PI's login has been initialized (see below), he or she can use the amgr utility to create additional logins for co-workers & other individuals who will work on the project. When a login is created, the accounting system makes an entry into the system's password file (/etc/passwd) and we seed that entry with a special shell called the Welcome! shell (wsh). The first time you login to the system, wsh will walk you through a short introductory dialogue so that you can do a number of tasks that people routinely do when they start working on a new UNIX system, including:

• changing your initial password (a good idea)
• setting your name in the system password file (it's up to you)
• changing your login name (takes about 24 hours to take effect)
• selecting a command shell (if you're not sure which one to pick, start with /bin/csh; you can always change it later.)

Earlier, we mentioned that when a command shell begins executing, it looks for its start-up files in your $HOME directory. The Welcome! shell completes its work by ensuring that your $HOME directory exists and that copies of the system's default shell and session start-up files are placed into it so that your command shell will be properly initialized whenever you login interactively or submit an NQS batch job, as shown below:

Start-up file	System master	Function
.cshrc	/usr/local/etc/cshrc	C shell run commands that are executed each time a C-style (e.g., csh or tcsh) shell is spawned
.login	/usr/local/etc/login	C shell session initialization commands that are executed once at the start of an interactive session or batch job
.profile	/usr/local/etc/profile	Bourne shell run commands that are executed each time a Bourne-style (e.g., sh, ksh or bash) shell is spawned

The system-supplied versions of the shell start-up files are nothing more than stubs whose purpose is to source the master copies from /usr/local/etc. However, it is important that you not change the contents of the system-supplied .cshrc, .login and .profile files because we set system-wide default values for things like your shell's command search path ($PATH), the path to the online programmer's manual pages ($MANPATH), the path to your $ARCHIVE directory, and so forth in the master versions. If you modify the system-supplied start-up files, you run the risk of changing things so that the start-up procedure won't execute properly. If you wish to customize your login setup, use your favorite text editor to create the following companion start-up files with the exact names given below in your $HOME directory:

Start-up file	Companion file	Function
.cshrc	.cshrc_user	Place C shell commands that you would normally put into .cshrc into your .cshrc_user file.
.login	.login_user	Put your C shell session start-up commands into your .login_user file.
.profile	.profile_user	Place your Bourne shell start-up commands into your .profile_user file.

If you accidentally delete the default start-up files from your $HOME directory, it is simple matter to reseed them. Login to the system as normal (but remember that not much initialization will have been done for you) and enter the following commands:

/bin/cp /usr/local/etc/cshrc $HOME/.cshrc
/bin/cp /usr/local/etc/login $HOME/.login
/bin/cp /usr/local/etc/profile $HOME/.profile

Once the files have been copied, logout and then reconnect to the system. When you get the command prompt in the new session, check that your environment is configured properly with the env(1) command, viz.:

lonestar% env

The UNICOS and UNICOS/mk operating systems on the supercomputers at TACC support two standard command shells:

/bin/csh - the C shell

csh is normally used for interactive work. It supports a full-featured shell programming language and has optimized $PATH lookup, command history & extensive job control features. csh is stable and mature, and has been a staple of UNIX timesharing for many years. Most TACC users use csh as their shell for interactive sessions.

Start-up files:	$HOME/.cshrc, executed whenever csh is spawned.
	$HOME/.login, executed after .cshrc has been processed whenever a C shell login session is initiated.

/bin/ksh - the Korn shell

ksh was originally a product of Bell Laboratories and was initially released with System V UNIX. The Korn shell is also known as the POSIX shell because it conforms to the POSIX 1003.2 specification for command interpreters. ksh is a superset of the older Bourne shell (/bin/sh), offering a full-featured shell programming language, command history and command-line editing with excellent functionality for scripting and subprocessing. On UNICOS and UNICOS/mk systems, ksh and sh are one and the same and most TACC users use ksh as the command processor for their NQS jobs.

Start-up files:

$HOME/.profile, executed whenever ksh is spawned.

In addition to the standard shells discussed above, we provide two additional command shells that you can use if you so choose. Be aware that, if you encounter bugs or other problems using them, we may not be able to address the problem other than by reporting it to the authors of the software.

/usr/local/bin/tcsh - an enchanced C shell

tcsh is an enhanced but completely compatible version of the Berkeley C shell with command-line editing, programmable word completion including filename completion, spelling correction and improved job history and control mechanisms over csh.

Start-up files:	$HOME/.cshrc, executed whenever tcsh is spawned.
	$HOME/.login, executed after .cshrc has been processed whenever a tcsh login session is initiated.

/usr/local/gnu/bin/bash - the Bourne-again shell

bash is an sh-compatible command interpreter from the Free Software Foundation that incorporates useful features from the Korn and C shells. It supports full-featured scripting and shell programming, job history and control as well as command-line editing and programmable word completion. bash is ultimately intended to be a faithful implementation of the POSIX standard for command interpreters.

Start-up files:	$HOME/.bashrc, executed whenever bash is spawned.
	$HOME/.bash_profile, executed after .bashrc has been processed whenever bash login session is initiated.

Complete documentation on each of these command processors is available via the man command on each system, e.g.,

lonestar% man tcsh