Citations and acknowledgements help us demonstrate the importance of computational resources and support staff in research at our institutes, and we ask that you acknowledge use of this system in any publications, presentations, or talks that were made possible because of it. An acceptable example citation is below:
The authors acknowledge Research Computing at the James Hutton Institute for providing computational resources and technical support for the "UK's Crop Diversity Bioinformatics HPC" (BBSRC grants BB/S019669/1 and BB/X019683/1), use of which has contributed to the results reported within this paper.
If you cite or acknowledge us in your work, please :doc:`contact-us` to let us know and/or edit our :doc:`publications` list.
The cluster has 57 physical nodes, providing a total of 1,844 compute cores (3,688 threads) and 17,600 GB of memory. A 1.5 PB parallel storage array is complemented by a further petabyte of backup capacity. A full description is provided on the :doc:`system-overview` page.
Please visit :doc:`user-accounts` for details.
The cluster’s head node (where you can submit jobs from) is called gruffalo, and you'll need an SSH client to connect. One is built into Linux and macOS, but for Windows you may need to install a separate client (WSL, Cygwin, MobaXterm and PuTTY are all good choices).
You connect via SSH using:
$ ssh <username>@gruffalo.cropdiversity.ac.uk
making sure to replace <username> with the username you were allocated when requesting an account. More detailed connection instructions are available on the :doc:`ssh` page.
Please :doc:`contact-us` for help with passwords. We can't recover (or even see) your password, but we can reset it in order to allow you to log in again, at which point you'll be prompted to set a new password.
Eeep! Don’t do that, no.
You can use gruffalo for compiling and debugging code, installing software, editing and managing files, submitting jobs, or any other work that is not long-running or computationally intensive, but for everything else, you must submit a job using Slurm (see :doc:`slurm-overview`).
Note that gruffalo has a 6 GB memory limit for your shell to avoid processes using up the node's memory.
The cluster uses the Slurm Work Manager job scheduling system and all jobs should be submitted (from gruffalo) to Slurm, where they will be allocated resources on one of the underlying compute nodes. More detailed instructions can be found on the :doc:`slurm-overview` page.
In general, there is very little application software installed system-wide (check /mnt/shared/apps for details), as most tools can be managed and maintained individually using tools like :doc:`bioconda` and :doc:`singularity`.
If you get stuck installing applications though, don't hesitate to :doc:`contact-us`.
This is a complex question, and the answer depends on a variety of factors, not least the type of job you're running and the amount of data being processed. See Slurm - Queue Policies & Advice for more discussion about this.
Considerate data management is everyone's responsibility, and it's critical that you ensure you're only storing (and backing up) important project-related data while keeping temporary and/or intermediate working data to a minimum. This helps keep the system running smoothly for everyone and ideally means we don't need to start enforcing quotas.
You can find more information on how we expect you to manage your data on the :doc:`data-storage` page. A summary of your current disk usage is shown on login, with detailed tracking available via :doc:`monitoring`.
Yes. Access to the cluster via a username/password combination is available if you are connected via a :doc:`organizations` network address, but for other locations you must first enable your account for SSH public key authentication, described in more detail on the :doc:`ssh` page.
We do have training materials from past workshops that can be made available on demand. Please :doc:`contact-us` for more details. Our workshops and training sessions are run regularly and you should look out for emails advertising the next one.
There are also some basic guides covering :doc:`linux-basics` and :doc:`tips`, as well as more in-depth information for getting the most out of cluster computing in the various topics listed under High Performance Computing.
Additionally, it's worth joining our Slack workspace (https://cropdiversity-hpc.slack.com) where there are plenty of expects on hand to help answer your questions.
The BeeGFS storage system uses transparent compression to automatically compress every file it stores. The free space message looks at the current compression ratio across the system and uses that to estimate how much more data could be stored, if that same compression ratio were to apply. Obviously the final result will be different based on how compressible newly added files are, but it will be somewhere within the range shown.
The name gruffalo goes way back to the early days of HPC at the Scottish Crop Research Institute (that merged with the Macaulay Land Use Institute to become the James Hutton Institute in 2011). Our first cluster - circa 2004 - used this name, and we've carried it on ever since, upgrading and/or rebuilding it across a range of hardware and software (RHEL, Fedora, CentOS, Rocky) generations.