- Use of login nodes for running code against anything other then small test datasets, especially where this results in significant numbers of processes, threads, or I/O operations.
- Any form of multi-threaded or parallel internet download, as this can easily saturate our public internet connection. Data must be downloaded to our storage in series before being further processed.
- Running interactive Slurm sessions on anything other than the queues specified in slurm-basics(7).
- Repeadedly running Slurm jobs (or interactive sessions) which request excessive amounts of CPU or memory which is then not used. It is an expectation that jobs will only request the resources that they need and automatic profiling(7) tools are provided to make this process as easy as possible.
- Attempting to circumvent any system restriction, such as CPU/Memory limits or the Slurm queues.
- Using SSH port forwarding over a port with the same numerical ID as another user's login. See ssh-tunnel(7) for a detailed explanation.
Any usage which we consider to be unaccepable may be terminated without warning, and we reserve the right to suspend login accounts if necessary.