Draft. This manual is new and still growing. If something here does not match what you see in the software, the software is right — tell us and we will fix the page.
Run on a remote machine over SSH
Send the case to a Linux host, solve there, and get the results back.
Not part of the free tier. The machine can be configured and tested, but selecting it as the executor is reported as an error on the Problems tab until the licence allows remote execution. Settings → Licence lists what yours allows.
Run on offers three routes, not two: Bundled solver, SHD Cloud and
Remote SSH. This page is about the third — a host you own and administer. The
metered route is run on SHD Cloud, and it deliberately
does not reuse the SSH machine configuration.
Windows needs an OpenSSH client. Windows 10 and later include one — enable the OpenSSH Client optional feature if the Machines page reports it missing. Authentication is by key: the transport is non-interactive, so a password prompt would hang rather than ask.
Add the machine
- From the home screen, open Settings, then the Machines tab.
- Press ＋ and fill in the Add remote machine (SSH) dialog: Name, Host, User, Port, Key file, Remote dir, etc/bashrc — the path to OpenFOAM's on that host — and Cores (optional).
- Press Add. Without a Name it refuses and says so.
- Select the machine and press Test connection. The output appears below the button; Cancel stops the test.
Fill in Cores even though it is optional. This machine cannot discover a remote host's core count, so without it the Simulation control pane cannot check what you asked for against what is there, and says the check was skipped rather than saying nothing.
Use it for a case
- Select Simulation control.
- Under Executor, set Run on to
Remote SSH. - Choose the host in Machine. If none is configured, the pane says so and points back at Settings → Machines.
- Run as normal.
The whole pipeline runs on the remote host — mesh, initialisation and solver — with the log streamed back and the results fetched when it finishes. There is no separate "upload" step.
Check it worked
- Test connection succeeds before you commit a case to it.
- The Simulation runs status line names the executor.
- The Log tab fills as the remote solver runs; if it does not, the job never started on the far end.
- Post-processing finds results afterwards — that is the proof the fetch completed, not just the run.