Notes for users of the Royal Society computers at UH


These notes are for users of the Royal Society computer systems at UH. These machines, so called because the key systems are owned by the RS under the terms of Martin's grant, are separate from the Starlink system. Although accounts are available on them for any UH Starlink user on request, you cannot log in to them without arranging with Martin for an account to be set up. If you do not know that you are authorised to use these machines then you are not, and should not try.

What the machines are and what they are for

The seven machines in the regular cluster are Numbers here list the number of cores, not the number of CPUs!

hydra is the machine that should be used by general users by default. hydra is also supposed to be the main data-crunching machine. However, in principle all the machines can be used by anyone with an account: please be considerate if you are taking all the CPU on someone else's desktop machine. nice should be used if you are planning to run long-term CPU-intensive jobs on any machine.

Other machines include

These are not fully integrated into the cluster and you may need to ask for an account to be set up before you can use them.

Operating systems and software

The machines run Debian 7.0 (aka wheezy). This is slightly different from the various Fedora versions used by the main UH machines. The differences are unlikely to cause anyone any difficulty, but if you aren't sure how to accomplish some particular task, talk to Martin.

As well as the usual Starlink software and IRAF, the machines have installations of AIPS, the Chandra software package CIAO and the XMM reduction software SAS, and the X-ray packages XANADU, FTOOLS and Funtools. Various software tools, including gcc and g77 and the Intel C and Fortran compilers, are also installed.

The default shell is tcsh, as is usual for Starlink systems.

Use of Starlink software

The recommended method is to add the following lines to your .login file:
setenv STARLINK_DIR /usr/local/star-nanahope
if (-e $STARLINK_DIR/etc) then
    source $STARLINK_DIR/etc/login
and these lines to your .cshrc file:
setenv STARLINK_DIR /usr/local/star-nanahope
if (-e $STARLINK_DIR/etc) then
    source $STARLINK_DIR/etc/cshrc
Then all Starlink software should be available to you after you log out and in again.

Use of AIPS

You need to be in the aipsusers group -- talk to Martin. Then run
You will need a user number, and although you can probably just pick your favourite number, you would probably be best advised to talk to Martin (or you can run /aips/aipsnumbers to see what's already in use).

AIPS is kept up to date using a 'midnight job'.

Use of CIAO

source /usr/local/ciao4/bin/ciao.csh
The CIAO commands should then be available. CIAO is kept up to date as new versions are released.


setenv HEADAS /usr/local/heasoft-6.6.3/x86_64-unknown-linux-gnu-libc2.7
source $HEADAS/headas-init.csh
setenv CALDB /usr/local/hea-caldb
source /usr/local/hea-caldb/software/tools/caldbinit.csh

Use of SAS

HEASOFT must be installed before SAS can be started (see above). Then, at a minimum, run
setenv SAS_DIR /usr/local/xmmsas_20110223_1801/
source $SAS_DIR/sas-setup.csh
setenv SAS_CCFPATH /usr/local/XMM/ccf
You will probably want other variables set for a given analysis:
setenv SAS_ODF /path/to/your/ODF/file
setenv SAS_CCF $SAS_ODF/ccf.cif

Use of funtools

The command line tools are installed in /usr/local and should be on your path by default. The library is in /usr/local/lib .

Use of compilers

I use the following
setenv INTEL /opt/intel/Compiler/11.0/083/
setenv LD_LIBRARY_PATH /usr/local/lib:/usr/local/pgplot:${INTEL}lib/intel64
setenv PATH /sbin:/usr/sbin:/home/mjh/bin.linux:${PATH}:/usr/local/utilities:${INTEL}bin/intel64
The commands icc and ifort should then be available to you.

Use of MPI

See the separate page for details on this.

Discs and backups

There is a large /home area. Feel free to use this for anything that is likely to change in the short term (papers, e-mail, programs...) but not for data.

Large datasets and static data generally should be stored on one of the data discs. In general each machine has local hard-drive space named /data/(machine name)(number), e.g. /data/pictor1, /data/vulpecula2. In principle you can use any of these (make a directory named after yourself in the root directory and start filling it up). In practice it would probably be sensible to talk to the main user of the machine in question to make sure that you won't be inconveniencing him/her (except for hydra which is for general use, including /data/raid1 disc).

As usual, it is sensible to make sure that the machine you are storing data on is the same as the machine that will be processing it. The discs are named as they are for a reason. The machines are linked by a Gigabit switch, but even so it is more efficient to make sure that the data are where you want them to be in the first place.

The /home disc is backed up by nightly incremental backup (on-site) and by fortnightly off-site backup. Some of the other data discs (see table below) are backed-up off-site on a monthly basis. Thus in principle all data stored on these discs is fairly secure against a site-wide disaster. Discs protected by RAID are spread across more than one physical hard drive in such a way that a single drive failure will not destroy data, so these are safe against local hard drive failure but NOT necessarily against fire, flood etc. Discs that are neither protected by RAID nor backed up off-site might fail at any time and will not be recoverable if they do. Do not rely on these discs! You should of course take your own backups of anything that is important to you. virgo, vulpecula and taurus all have DVD writers, and virgo also has tape drives which take DAT tapes up to DAT-72. Do not rely on the system-wide backups to preserve your thesis, papers or other masterwork. It would be very easy to back such things up onto the Starlink machines, for example. However, please do not back up from one internal disc to another (e.g. vulpecula1 to vulpecula2). This gains you very little, especially if one of the discs is already protected by RAID, and wastes a lot of disc space. The backup situation is summarized below.

Disc nameComputerCapacity (Tb)Incremental backupOffsite backupRAID?


It is possible to arrange to receive your UH e-mail directly on the RS machines (and benefit from superior spam filtering to what the university provides). If you want to do this, talk to Martin. All the machines will send e-mail fine by default. pine can be configured to talk to the UH mail server in the same way as on the Starlink cluster: so can the recommended command-line mailer mutt. Alternatively, it's always possible to use the University's web mail service.

E-mail accounts can be set up either to deliver to the local machines or to forward elsewhere.


hydra is a web server and you can set up personal pages on it simply by making a public_html directory and putting stuff in it: this will then be available to the world as If you need a non-personal web area, that can be arranged. Domains inside can be provided and served on request.


hydra is the printer server for all the other machines. The printers are Each system has its own default printer. You probably as a user want to select the printer that's nearest you, irrespective of what system you're on -- in this case do lpoptions -d [printername] to set a persistent default.


All the laptops can be booted into Windows, but none of the desktops is now set up as dual-boot. All machines have the option of using a virtual windows desktop via VirtualBox. This is the preferred option for running Windows on the desktop machines. Talk to Martin for help setting this up.


For setting up accounts, or any further information, talk to Martin.