ITS Research FAQ regarding use of services during COVID-19 pandemic¶
As a result of the large-scale shift to remote working due to the COVID-19 pandemic, we have been asked various questions relating to computational research, which we'll try to address below. We've seen an increase in the number of new account requests for the HPC service, and we realise there will be quite a few users wishing to run workloads on the cluster for the first time. Fortunately, the design of the HPC cluster service means that for many of you, your workflow may remain the same as when you were based on-campus.
How you can help us (and yourself)¶
The cluster has been very busy lately, but jobs are getting run in a timely fashion. However, if you are performing EPSRC-remit research, then you are eligible to use Tier2 (and higher) services for multi-node parallel job and large jobs.
Please be patient for GPU jobs as we've noticed a sudden surge in demand for these. We added a new sbg3 node last week which contains 4 x Nvidia V100 cards with 32GB RAM. Please also be diligent to check that your jobs are using the GPUs correctly, so as not to waste this resource.
If your jobs run for strictly less than 1 hour, using h_rt=1:0:0
will allow
your job to run on an expanded range of nodes that may otherwise be idle,
reducing your queueing time, and allowing longer jobs to run on the main nodes.
Otherwise for any jobs lasting more than 1 hour, a value of h_rt=240:0:0
is
recommended.
Additionally, we still see users who are restricting their jobs to run on nxv
nodes only, using the node_type
option. This will result in long queueing
times. If you are trying to avoid the dn nodes for compatibility reasons, then
using -l avx2
is much less restrictive, and includes the new sdx nodes.
Can I use services provided by ITS Research from other locations?¶
The Apocrita HPC Cluster can be accessed from any location and does not require a VPN to access.
Our Tier 2 partners (Athena, Thomas and Jade) also do not require a VPN to access. Athena is restricted to QMUL IP addresses though, so you will need to login to Apocrita first, and subsequently connect to Athena.
The QMUL Github Enterprise, and Globus services are also accessible from any location.
I have a desktop at QMUL running simulations. Can I access it remotely?¶
It is unlikely. A couple of schools have local solutions to this, though there is no centrally supported solution at present. We recommend checking if your work is suitable to run on the HPC cluster.
Can I run (insert application name) on my desktop or laptop at home?¶
We are discussing this with a number of vendors now. IT Services are maintaining a list of which applications are available and from which service. The VDI service is currently at capacity and being ring-fenced for critical business functions, but is in the process of being scaled up. A managed laptop would be advised as this will have the applications setup already.
We are working with 3ds and Ansys to see if we can make these available.
MATLAB can be installed on work/home machines as part of the campus licence by visiting this link Please note that while the ITS Research team support the use of MATLAB on the cluster, use on other devices is supported by IT Services at helpdesk@qmul.ac.uk.
Can I run (insert application name) on the HPC cluster?¶
Apocrita is a Linux cluster designed for command-line access via SSH and submission of computational tasks using the job scheduler. You can request an account and suitable applications from our website.
We support a large number of research applications, typically designed to be queued up to run in batch mode (non-interactively). We don't have much capacity for interactive or graphical applications.
Can I use Apocrita as a backdoor into the QMUL network?¶
The Apocrita network is not considered "inside" the QMUL network, so will be treated the same as an external connection.
How can I get support for ITSR services during the social distancing period?¶
Please continue to raise support requests via the usual means at its-research-support@qmul.ac.uk or visit our Slack chat channel at https://qmul-research-it.slack.com for more informal discussion and help. For instances where we would typically have face-to-face meetings to discuss issues and requirements, we can arrange a video call instead.
Please remember that we support the cluster and related research IT services such as GitHub Enterprise and Globus, but you will need to contact the main IT helpdesk for general IT issues.
What resources do you have for learning about the cluster?¶
We have some learning resources at
- https://docs.hpc.qmul.ac.uk - documentation
- https://learn.hpc.qmul.ac.uk - tutorials
- https://blog.hpc.qmul.ac.uk - articles, news and tutorials
We are looking at ways to expand this with additional relevant content in the coming weeks.
I am working on COVID-19 related research. What can I do?¶
Please contact Simon Butcher/Tom King directly to discuss access to computational resources and software engineering assistance.
How can I reset my password for the HPC cluster/GitHub Enterprise?¶
Previously we restricted the password reset manager to campus machines only, but we are working on a solution for that and will update this page when we have a better solution. In the meantime please raise a support ticket with us to reset forgotten passwords. Please note that if you enter your cluster account password incorrectly 5 times within 10 minutes, you will be automatically blocked for 30 minutes, but after this time you may attempt to connect again.
This page will continue to be updated with additional questions and answers as we receive them.