Scientific analysis servers
This article describes the scientific analysis servers. It covers:
- Scientific Analysis (sci) servers
- Software installed
- Access to storage
Once you have logged on to one of the login servers you can then
ssh to the shared analysis servers that you have been given access to.
Scientific analysis servers
Access to the scientific analysis (sci) servers is restricted to ssh login from hosts inside the RAL firewall, so access from outside needs to be achieved via one of the login servers.
|Server name||Virtual/Physical (OperatingSystem)||Processor model||CPU Cores||RAM (GB)||/tmp (GB)|
||V (CentOS7)||Xeon(R) Silver 4114 CPU @ 2.20GHz||8||16|| NA
||V (CentOS7)||Xeon(R) E5-2630 @ 2.20GHz||8||16|| NA
|sci3.jasmin.ac.uk||P (CentOS7)||AMD EPYC 7402||48||1000|| -
|sci4.jasmin.ac.uk||V (CentOS7)||Xeon E5-2660 @ 2.20GHz||8||32||NA|
|sci5.jasmin.ac.uk||V (CentOS7)||Intel(R) Xeon(R) CPU E5-2630 v4 @ 2.20GHz||8||32||NA|
|sci6.jasmin.ac.uk||P (CentOS7)||AMD EPYC 7402||48||1000||-|
|sci8.jasmin.ac.uk||P (CentOS7)|| Intel(R) Xeon(R) Gold 5118 CPU @ 2.30GHz
Note 1: sci[3,6,8].jasmin.ac.uk are physical servers with IP addresses on a private internal network, so outbound internet access (via NAT) is only for https. This means that outbound SSH will not work (to hosts outside of JASMIN) on these particular machines.
Note 2: The local /tmp of the virtual VMs is Not Available (NA) for users as this is used by the VM system
Note 3: A new monitoring utility "Arbiter" was implemented across all Sci machines to control CPU and memory usage. This utility "Arbiter" records the activity on the node, automatically sets limits on the resources available to each user, and notifies JASMIN admin when resources quota are changed. From the user's perspective, his/her process will be capped from using any extra resources and the process can be slowed or run out of memory.
For more information on using the JASMIN Analysis Platform can be found here.
Scientific analysis servers are designed for interactive and ad-hoc analysis of data in group workspaces, the CEDA archives, and users' home directories. For long-running and resource-intensive jobs, users are required to use the LOTUS cluster which offers better I/O performance, parallelism, and scheduling.
The following guide should be considered when using the scientific analysis servers:
- Be aware of resources available at the time before your process starts and choose the scientific server that is suitable ( check the average load on the scientific server from the message of the day on login server for example or using any of Linux monitoring command:
- Execution/processing time should be less than 1 hour;
- Serial jobs only
- High memory jobs should be executed on the high memory scientific servers with 1000 GB of RAM sci[3,6].jasmin.ac.uk (see above Table )
- Monitor your process on a scientific server using
- Report if there is a user's process affecting the performance of a scientific server
Each analysis server has the following features:
- Default Redhat Enterprise Linux development environment.
- The software packages that make up the JASMIN Analysis Platform are all installed - providing commonly-used open-source analysis tools. These packages include NCO, CDO, Python2.7 (with netCDF4, matplotlib, numpy etc.,) and R.
- Access to proprietary tools, e.g. IDL and Intel Fortran, through the Module System.
- Ability to run X11/X-Windows applications (use the NX graphical desktop service to view X11 graphics output efficiently on your remote machine).
Access to storage
Each analysis server has:
- Access the LOTUS /work/scratch-pw and /work/scratch-nopw area.
- Read access to the CEDA Archives according to user permissions.
- Read/Write access to Group Workspaces according to user permissions.