Grace is a compute cluster designed for extensively parallel, multi-node batch-processing jobs, having high-bandwidth connections between each individual node.
Grace end of life
The Grace cluster is being retired. Notable dates are:
- 1st February 2021: Job submission will be switched off. Jobs still in the queue will run. Access to the login nodes will remain for three months so you can recover your data.
- 3rd May 2021: Access to the login nodes will be removed and all data will be deleted.
No new user accounts will be created on Grace.
Kathleen is the replacement.
Grace accounts can be applied for via the Research Computing sign up process.
As Grace is intended for multi-node jobs, users who specify that they will need to use multi-node jobs (e.g. with MPI) will be given access to Grace.
Please use your UCL username and password to connect to Grace with an SSH client.
If using PuTTY, put
grace.rc.ucl.ac.uk as the hostname and your
seven-character username (with no @ after) as the username when logging
uccaxxx. When entering your password in PuTTY no characters or
bulletpoints will show on screen - this is normal.
If you are outside the UCL firewall you will need to follow the instructions for Logging in from outside the UCL firewall.
Logging in to a specific node§
You can access a specific Grace login node by using their dedicated addresses instead of the main
grace.rc.ucl.ac.uk address, for example:
ssh email@example.com ssh firstname.lastname@example.org
The main address will unpredictably direct you to either one of these (to balance load), so if you need multiple sessions on one, this lets you do that.
Copying data onto and off Grace§
You will need to use an SCP or SFTP client to copy data onto Grace. Please refer to the page on How do I transfer data onto the system?
You can connect directly in both directions between Grace and Kathleen.
If you find you cannot connect directly from one cluster to another, this is generally because of firewalls in between and so you need to use tunnelling with the scp command.
This section has not been filled in.
Job sizes and durations§
For interactive jobs:
For batch jobs:
If you have a workload that requires longer jobs than this, you may be able to apply to our governance group for access to a longer queue. Applications will be expected to demonstrate that their work cannot be run using techniques like checkpointing that would allow their workload to be broken up into smaller parts. Please see the section on Additional Resource Requests for more details.
Grace's compute capability comprises roughly 650 compute nodes each with two 8-core Intel Xeon E5-2630v3 2.4GHz processors, 64 gigabytes of 2133MHz DDR4 RAM, 120GB hard drives, and an Intel TrueScale network.
Two nodes identical to these serve as the login nodes.