Performance issues on JASMIN

JASMIN is currently under a heavy load, in part because of overuse* and inappropriate use of the scientific analysis servers. This has regularly resulted in overloading of the scientific servers causing unresponsive and slow processes/commands over the past few weeks. 

We are policing inappropriate use where we find it, and will be rolling out more science servers at some point. However, meanwhile, please move large tasks to LOTUS, and use the data transfer servers as appropriate. 

The general guidance is that if your task takes a long time (greater than tens of minutes), you repeat them often, or use many processes for anything but testing, you should move your work to LOTUS. If you are still unsure about where/how to run your tasks, ask for help. Whether that is by asking your colleagues/peers, searching our help documentation - or finally, contacting our helpdesk (please ensure you have investigated other avenues before this). 

If you are seeing performance problems, by all means let us know, but be specific: “exactly what is causing, what problem on what machine, and when did/does that happen?”. The type of necessary information we need is detailed here.

Best wishes, 
JASMIN team

* The high volume of data transfers from ARCHER/RDF (RDF is available until 29th February) to JASMIN overloaded the JASMIN network which is shared by JASMIN scientific, login and transfer servers.