next up previous
Next: Software/data resources Up: Linux cluster notes Previous: Introduction


Basic astronomy cluster usage

Establishing an account and setting the password

Before you can log in, you will need to get an account and a password. Note that an astronomy account is distinct from an NMSU account; the NMSU account is what you will use for registering for classes and university business. For best handling of email (see below), it is strongly preferred that your astronomy account name match the name of your NMSU account. The best way to insure this is to get your NMSU account set up first. This is done by going to, and using the Activate link to activate your account; it will ask for your desired account name, and will let you know if that name is already taken by someone else. Once you have your account name, then we can set up as astronomy account with the same name: ask Jon Holtzman. When we set up the account, we'll let you know along with a temporary password. You should immediately change this password using change your password, using the yppasswd command; you will need to enter your old password and your new one twice.

Logging in

Normally you will log into a machine directly from a keyboard and monitor directly attached to the machine. If you want to log in from a remote location like home or another institution, note that we only allow external logins via the SSH protocol. Most Unix machines have the ssh command installed, and SSH software is available for Windows and Mac machines as well.

By default, however, direct access to all of the Linux machines except the server, astronomy, is closed to all outside machines except for the Apache Point subnet; this is done to help to improve security. You can always log into astronomy from other locations, and if necessary, log into other cluster machines from astronomy. If you will be away for an extended period of time, and there are other machines that you would like direct access to, it is straightforward to set this up - just ask.

Logging in using SSH keys instead of passwords

You can always log in using your password. However, if you prefer, you can also log in using an SSH key, which has certain advantages. The SSH protocol allows the use of key pairs for login authentication. There are at least two flavors of SSH, which use different formats for the key pairs. Since mid-2004, we have been using OpenSSH. (As of late 2004, APO is using the other flavor).

To generate a key pair, use the command:

ssh-keygen -t dsa

which will create two files in the .ssh subdirectory under your home directory called id_dsa and; it will also prompt you to enter a passphrase to use with the key pair. The .pub file is public and can be copied to other machines; the other file should never be shared.

If you wish to log into another machine using the SSH key:

Once this is done, when you use ssh to connect to the machine, it will prompt you for the passphrase associated with your key.

If you want to come from a machine running the ohter flavor of SSH into our machines, you can convert that SSH format key into our format using ssh-keygen -i -f on the OpenSSH machine (ours), and then use the converted key.

You can register your passphrase for a session to avoid having to repeat typing it by using ssh-add, which will prompt you for your passphrase. If you are doing this from a remote log in session (as opposed to directly logged onto your workstation, you will need to first start a shell with the SSH agent using: ssh-agent shell (where shell is the name of your preferred shell, e.g., tcsh or bash). After you register your passphrase, any SSH operation which would have required you to enter your passphrase will automatically use the one you entered with ssh-add. This can make multiple uses of SSH very convenient, but beware, it also would allow someone else who happened to come into your office the capability of logging into some other machine without knowing a password!


As with all UNIX systems, you can customize your workspace. Linux provides several shells which you can work in. By default, all users are set up using tcsh as the login shell. You can start any other shell by typing its name (e.g., bash, sh). If you wish to change your default shell, use the chsh command.

You can define your own aliases, set environment variables using startup files which are read every time the shell is started up. tcsh uses the .cshrc file in your login directory, while bash uses the .bashrc file. Your account is setup up with a default .cshrc file; the default file has an entry which sources a systemwide file which resides in /home/local/etc/cshrc (or /home/local/etc/bashrc for bash). This file sets up a default path so that you will find all of the locally installed applications and packages, and sets some variables which may be used by these packages. You are welcome to modify your .cshrc file as you see fit, but if you want to use the locally installed packages, we strongly recommend you keep the source /home/local/etc/cshrc line in your .cshrc, and also, be careful about making modifications to the command path.

Window system/ managers

As will all X11 display system, Linux provide the possibility of using a varienty of different window managers which control the appearance of your display. The default window manager we are using is KDE, but gnome is also pre-installed, among others. If you want to use another window manager, you can set this up using menu entries on the main login screen. Different window managers can generally be customized to provide different resources in your menus, taskbars, mouse buttons, etc. For KDE, use the KDE control center application to customize things.


By default, outgoing mail has a return address If people reply to this, it gets received at the NMSU, not the department, level. It is possible to read your NMSU email using the web interface. You can also set up a personal machine as a mail client to the main NMSU mail server.

Alternatively, many people in the department chooses to have the NMSU email forwarded either to a personal email account or to the Astronomy cluster. This is accomplished by going to, go into the email preferences configuration, and have all of your email forwarded. To forward it to the astronomy cluster, use

You can also set forwarding to some alternate personal address if you prefer. If you receive your email somewhere other than the astronomy cluster, then you should also set things up so that any email that happens to be sent directly to the astronomy cluster gets forwarded to your preferred address. This can be accomplished by placing a .forward file in your home directory which contains a single line with the email address to which you wish to have your local mail forwarded (note that if you use a .forward file, both this file and the home directory must not allow world-write priviledge).

In any case, it is wise to set things up so that all of your email ends up in one place, so you will be sure not to miss anything!

If you set up your mail to be delivered to our server, mail can be seen from any of the Linux machines; you do not need to log onto the Linux server to read it.

Many mail clients are available; some of the most commonly used ones are netscape/mozilla/thunderbird (various versions of a similar package), pine (text-based), the MH suite (text and graphical interfaces), plus others. If you don't have a strong preference, we recommend you use netscape/mozilla/thunderbird. We also have a web interface for reading mail (

If you use a mail client, the server (astronomy) is set up to be either an IMAP or a POP server; we recommend using IMAP as it is more powerful (it allows you to manipulated mail directly on the server instead of just locally on your machine). The server name, both for incoming and outgoing mail, is

You do need to be careful about only using one tool to access mail at a given time; if two different tools are being run (e.g., webmail and a mail client), you may not see all of your messages on each.

Whatever mail tool you use, we prefer that you create folders for filing your mail; if you leave everything in your inbox, it clutters up our mail spooling disk.

Useful mail aliases

Current email lists for subsets of the Astronomy department, as well as the entire department are kept fairly up to date. You can use these by sending email to, where listname is one of the following:

The exact contents of the list can be viewed here. Although some effort is made to keep the lists up to date, please contact Jon if modifications/additions/deletions need to be made to the list.

Web pages

A Web server is run on the Linux server, astronomy. By default, users will have a Web page that is accessed at

We have set aside a specific partition, /home/httpd, to host individual Web pages. You will have a directory, /home/httpd/html/username, where you can place HTML files that can be seen on the WWW. This is where you would make a personal web page.

Department web pages

Individual person webpages

A set of research-oriented webpages for department members are available for your perusal on the server. To view them, simply load the webpages

and click on the appropriate sublinks in the top navigation bar, the left hand menu (under Directory), or on individual names in the primary window.

The contact information on the top of each page is created automatically from a text database, while the text on research interests, publications, and the like is stored in an include file on astronomy. You own this file for your webpage, and have the freedom to update its contents at any time.

You can modify the information on your own page by changing the file /home/httpd/html/dept/directory/[faculty,rstaff,grads]/username.include

(for faculty, research staff, and graduate students, respectively).

Due to the use of dynamic html, any changes that made to the include file are instantly reflected online.

The username.include file, and parallel versions for everyone else (named according to your login ID), are owned by each individual user. If for some reason you find that this is not the case and find it impossible to edit your own file, contact Jon to fix the permissions.

There are a few details about the format of the include files worth mentioning, if you decide to update them.

1. The extra line breaks (<br>) at the end of the file extend the length of the primary window so that the page footer lies below the bottom of the left hand menubar. If you have a short research entry and then add in a few paragraphs as your interests develop, some of the line breaks might be removed to compensate (but this depends upon the width of the viewer's browser window, and the browser in use, so it is always safer to have extra line breaks rather than too few).

2. If you have a personal webpage, feel free to place a link to it within the include file text (as has been done on Jon's page, for example). Please use the following format to construct the link.

<p> <a href= target="_public"><b>Home Page</b></a>

The purpose of the _public target item within the <a href=> flag is to insure that someone who is browsing through a set of our webpages keeps sending them to the same external page (to avoid cluttering up a viewer's screen with many spawned pages, if they are browsing through our entire directory tree).

3. If you publish a new paper, I hope that you will enjoy placing a link to the journal article within your include file. Please use the following format to construct the link.

<p> <a href= target ="_research">Optical Rotation Curves of Distant Field Galaxies: Sub-L* Systems</a> <br> Vogt, N.P., Phillips, A.C., Faber, S.M., Gallego, J., Gronwall, C., Guzman, R., Illingworth, G.D., Koo, D.C., & Lowenthal, J.D. 1997, ApJ, 479, L121

The purpose of the _research target item within the <a href=> flag is to send external journal paper pages to the same spawned page every time, again to avoid clutter for the viewer.

4. If you put in a link to a close collaborator's page, or some other research link, please use the _research target flag again. We are trying to give all spawned webpages one of these two flags throughout the entire set of department webpages, to distinguish broadly between research information and public (which includes academic) information.

5. Please keep the </font> flag at the end of the file, so that the default fonts for the menu bars can come into effect after your prose.

The individual photos are stored in a parallel directory, at


If you would like to replace your current photo, please let us know. If you have a digitized version of a photo of your own, just send it to Jon.

If the information at the top of your webpage needs updating (for example, if you change offices or you receive a M.S. degree), please let Jon know and we will update your database entry. Your shtml file is overwritten automatically every time the files are created, so do not try to put updates in there (they will disappear).

If you would like to make your research-oriented webpage your public homepage on astronomy, with an easy path for people to load, then place the following redirect file inside your owned directory at /home/httpd/html/username. The redirect file should be called index.html, and should contain the following text (replacing faculty.nicole with rstaff.sallam or grads.carrande, etc.).

<html> <meta http-equiv="Refresh" content="0; URL="> <body></body> </html>

For example, Nicole placed this file within /home/httpd/html/nicole and now if you load

into your browser, you will be redirected to the page at

Research group webpages

We also have a set of research group web pages, at:

The content for these research group webpages can be found in files: /home/httpd/html/dept/include/research.[extragalactic,galactic,planetary,solar,stellar].inc

These files are associated with the faculty group, of which all faculty are members, so any faculty can modify the content in these.

File transfer

The preferred method for file transfer is to use encryped file transfer via the scp or sftp commands.

Please note that standard FTP service (not sftp) to individual accounts is turned off for all of the Linux machines. This is because username and password information used when accessing FTP service is passed over the net unsecured, and can be stolen. Please do not attempt to ftp into your account from another machine; it will not succeed, but can still compromise security!

The only FTP access that is allowed is anonymous FTP access to the Linux server, astronomy. The anonymous ftp area that is accessible can be seen directly from any of the Linux machines, in /home/ftp. Upon account creation, there will be a directory /home/ftp/pub/username created; files placed here will be available to external users accessing astronomy via anonymous ftp. If you wish to allow external users to put files onto the FTP server, external users will need to cd to incoming/username after logging into the ftp server. You will be able to see files they have downloaded in /home/ftp/incoming/username from any of the Linux machines. Files uploaded into the /home/ftp/incoming/ area will be owned by the user ftp, but can be deleted by anyone; copy these files into your area so that you will own them, and then please delete them from the incoming/ftp area. ALL LARGE FILES IN THE INCOMING AREA OLDER THAN 6 MONTHS MAY BE AUTOMATICALLY DELETED TO KEEP THE AREA FROM FILLING UP.

Network usage

Please be aware of the distinction between CPU usage/location and disk usage/location. When you log into a machine, you are using its CPU. However, each CPU can access either local disks or disks over the network. The locations /home/users, /home/httpd, and /home/ftp are all located on the astronomy server. Thus, if you log into your machine but access files in one of these directories, you are creating network traffic. This is not a problem so long as the volume is relatively small, but if you were to start using or writing programs that do a lot of transfer to and from disk, it can cause poorer performance.

In addition to being able to access the server disks, each Linux workstation has local disk space. It is important to remember that performance of applications which access disk files depends on whether the file access is local or over the network. In addition, if you are accessing things over the network, you affect the performance of other computer systems in the department as well.

By default, when you get an account, your home directory will be located on the server disk, in /home/users/username. Once you start to use significant I/O, you should request a directory on your local machine.

It is important to try to work on local disks whenever possible. Responsible network usage will help everybody in the department.

Computing resources

There are two department servers, astronomy and astrodisk, that provide basic services. In general, these machines should not be used for any significant processing, as this could affect the cluster performance for all users.

Your desktop is meant to provide basic computer services like editing of files, web access, email access, image display, plotting, and the capability to compile and run basic programs. Many basic programs are installed either on each individual system, or centrally on the server disks, and thus accessible to all machines. However, the desktop computers are probably not optimal for significant computing.

If significant computing resources are needed for your work (e.g. more powerful/faster processing, more memory, etc.), there are several centrally located computers that are available via remote login:
machine CPUs Memory (Gby) Dedicated usage
praesepe 48 132 Large jobs
hyades 48 32 Large jobs, better for parallel code?
virgo 64 64 Large jobs
antigona 16 16 Klypin group
seismo 32 64 Jackiewicz group
solarstorm 48 132 McAteer group
milkyway 64 64 Holtzman group

next up previous
Next: Software/data resources Up: Linux cluster notes Previous: Introduction
root 2015-08-12