Code of Conduct
On this page we list some recommended practice that is crucial for maintaining a good and professional working relationship between the user and the system administrators.
Most of those contents are quite self-explanatory while others help to reduce the amount of support time.
Be always aware that you are working on a shared system where your behavior could have a negative impact on the workflow of other users.
General
- We assume that you are familiar with some basic knowledge about Linux command line (
shell
) navigation and shell scripting. If you never worked on the command line, consider some Linux tutorials on the subject first. - We expect you to exploit this valuable documentation before asking for help. All that is needed to get some simple jobs done is documented here.
Mailing List
- We communicate upcoming events (e.g. maintenance downtimes) on our mailing list. Make sure that you are subscribed to this list, otherwise you will miss important announcements ( you're automatically subscribed at account creation).
Security
- Do not share your account
- If using public key authentication, do not share your private key
Communication with the Cluster Administrators
- Use our ticketing system email svpmeso@univ-fcomte.fr
- Do not use the personal email address of a cluster administrator
- For each new problem start a new conversation with a new subject
- Always give an exact as possible description of the problem. Provide your username, error messages, the path to the job script, the id of the job, and other hints that make the problem-solving process as economic as possible
Housekeeping
- Clean up your HOME and WORK directories frequently, in particular before asking for an increase of your quota limit
- Do not save thousands of files in a single directory. Distribute the files to sub-directories
Cluster Performance
- DO NOT run resource-intensive computations directly on the login node AKA submit node. This will have a negative impact on the performance of the whole cluster. Instead, generate a job script and submit this job script to the cluster using
SGE
. - DO NOT Stress the File Systems
- LIMIT Input/Output (I/O) Activity