How to use EERIE's facilities at JASMIN

JASMIN is the UK's data analysis facility for environmental science. JASMIN is operated by UK Research and Innovation who are EERIE partners.

EERIE users have access to the following facilities at JASMIN:

• 250 TB of group workspace
• 50 TB of object store
• interactive analysis servers
• LOTUS batch cluster - 19,000 core batch computing system
• Orchid GPU cluster - 14 node GPU cluster availble through the SLURM scheduler and one interactive GPU node
• Jupyter Notebook service
• transfer servers
• scientific software
• CEDA Archive
• Elastic Tape system
• Met Office MASS tape system client

There is extensive help documentation available for all JASMIN systems at https://help.jasmin.ac.uk/

Getting a JASMIN account

Please follow the instructions at http://help.jasmin.ac.uk/article/189-get-started-with-jasmin to get access to the JASMIN facilities:

• At step 4, applying for the jasmin-login role, you should include the email addresses of Malcolm Roberts and Jon Seddon who will provide you with a reference to support your application.
• At step 5 you should apply for access to the "eerie" Group Workspace and "eerie-o" object store.
• To access data that has already been published to the ESGF, for example the EERIE simulations or CMIP6-HighResMIP data, you will also need to register for a CEDA account and link this to your JASMIN account.

Accessing your JASMIN account

Please follow the instructions at https://help.jasmin.ac.uk/article/187-login to login to your JASMIN account using SSH.

Using JASMIN storage

The JASMIN storage is summarised in the JASMIN help system. EERIE users will have access to the following forms of storage:

• 100 GB of home directory at /home/users/<username>, which is the only storage that is backed up. The home directory has snapshots enabled to allow for the quick recovery of deleted files. Important results should be stored in your home directory.

• temporary scratch storage for the storage of large files that are being actively analysed. Data on the scratch storage that is older than 28 is automatically deleted.

• there is a single 250 TB allocation of group workspace (GWS) for the EERIE project at /gws/nopw/j04/eerie. This GWS will be used to hold the model output data that is required for analysis. Data will be written to it using the Data Management Tool, a web based catalogue of EERIE model data. If this GWS becomes full then all EERIE users will be unable to work and so we must be very careful what data is written to the GWS.

• files in the /gws/nopw/j04/eerie/public directory of the EERIE GWS are available via password protected HTTPS at https://gws-access.jasmin.ac.uk/public/eerie/. All model output will be stored in this area to make it available by HTTPS. Please contact Jon Seddon or the other WP3 leads for the username and password. Please see https://help.jasmin.ac.uk/article/202-share-gws-data-via-http for more details. The performance of this HTTPS access is limited and so it is recommended that the data is accessed from the Jupyter Notebook service, the interactive analysis servers or the LOTUS batch cluster for best performance.

• there is 100 GB of fast solid state disk (SSD) storage that is available at /gws/smf/j04/eerie. This SSD is ideal for storing software on. Please contact Jon Seddon if you would like some software to be installed here.

• the CEDA archive is available at /badc. The archive contains a range of observation and reanalysis datasets along with CMIP6 data that is available at CEDA's ESGF node. EERIE data that has been published to, or replicated to, CEDA's ESGF node will be available at /badc/cmip6.

• EERIE has 50 TB of object storage. The object store is only available internally at JASMIN and is not available externally to JASMIN.

Analysis services

• for interactive analysis the scientific analysis servers are a great way to get started at JASMIN, or to develop and test analysis software or scripts that will be run on LOTUS.

• the LOTUS batch cluster is designed for the analysis of large volumes of data. LOTUS contains the same software stacks and storage as the analysis servers and jobs can be submitted to the cluster using the SLURM scheduler.

• the Jupyter Notebook service provides access to a Python environment through a web browser. The Notebook service has read-only access to the EERIE GWS and the CEDA archive, but does not have access to the scratch storage. The Notebook service may be particularly useful during hackathons but unlike the equivalent service at other institutes, it has more limited compute resources available to it for data analysis. Therefore, the Notebook service is useful for the interactive viewing of data, but LOTUS or the analysis servers should be used for the analysis of large volumes of data.

Data Transfer

JASMIN has fast network access to the UK's Janet network and from there to the GÉANT network. JASMIN supports many different methods for fast and efficient data transfer.

Getting help

Initially, please check the existing documentation at https://help.jasmin.ac.uk/

If the documentation does not help then please either ask for informal help from another member of the EERIE project (Jon Seddon has much previous experience of using JASMIN) or contact JASMIN support by clicking on the big blue question mark "?" button at the bottom of https://help.jasmin.ac.uk/.