If you have a problem that can't be resolved by reading the FAQ or the documentation then you are very welcome to contact us.
How to contact us
In order to reach the SCITAS team please send an e-mail to firstname.lastname@example.org starting the subject with HPC - e.g. HPC segmentation fault with Intel MPI
The HPC is needed to route your message directly to us. If you omit it your e-mail will have to be manually assigned by the helpdesk and this will take a bit longer.
You will then receive an e-mail saying "You have reached the Scientific IT and Application Support Team (SCITAS) with your message" along with a reference number begining with INC. This means that your problem is being tracked and that things are working.
What we need to know
If you need general advice then please give as much information as possible - we don't know every code out there nor what they all do.
If you have a problem running a job then, as a bare minimum, we need to know the job ID and cluster along with the following information
- The script you used to submit the job or the command line options to sbatch / salloc
- Your environment - "module list" and "env"
- The location of the executable in question
- The detailed error messages and location of the output files
If it's a compilation problem then obviously there won't be a job ID but the rest of the points remain valid.
You are also welcome to provide other background information that may help us such as
- Do your colleagues have the same problem?
- Does the problem only occur on one cluster?
- Did you compile the code yourself and if so how?
- Has this problem suddenly appeared? If so when.
As an example of how not to and how to do things here are a few examples:
BAD: My job doesn't run but it works fine at my friend's institute. Please help.
BETTER: My job number 5467879 on Aries fails with the error "No remaining cheese - please milk the cows". The output files and executable are in /scratch/bob/mycode
GOOD: My job number 5467879 on Aries fails with the error "No remaining cheese - please milk the cows". I submitted this with the script at /home/bob/runmyjob and the output files and executable (mooveit.x) are in /scratch/bob/mycode
This code was running fine until Wednesday the 13th when I started to get this error. It doesn't seem to happen all the time as some jobs (such as 5467312) run to completion.
What happens next
The SCITAS team members on support duty will start investigating within half a working day and, should more information be needed, you will be contacted. Depending on the complexity of the problem we may invite you to visit us to discuss the problem in more depth.
Providing computing resources, training and expertise to the EPFL community.
- General purpose and specialized computing platforms
- Application support
Tel: +41 (0) 21 693 14 05