Run your first job
A guided walkthrough using a real case. By the end you will have run an OpenFOAM airfoil simulation on Rescale and downloaded the results.
What you're running
simpleFoam (a steady-state solver for incompressible turbulent flow) on a 2D airfoil mesh. It's a small case, chosen because it finishes in a couple of minutes and produces output worth looking at. Engineers often use a steady-state run like this to generate starting conditions for a transient one.
Create the job and attach input files
Go to Jobs → Create New Job.
- Name it something you'll recognize later:
Basic Job: Airfoil2D. - Attach
airfoil2D.zip. Rescale extracts.zip,.tar,.tar.gz,.tgz, and.gzarchives into the working directory before the run starts. - Leave Job Type as Basic. Other types run the same setup many times over: see Workflows & DOE.
It's faster to transfer, cheaper to store, and the archive stays in your files for reuse. Just remember the contents are extracted into the working directory, so paths in your command are relative to what's inside the archive, not to the archive itself. That's the most common first-job mistake.
Choose software and write the command
Search for OpenFOAM and select it. Pick a version, this tutorial uses an Intel MPI build.
Then enter the command Rescale should run:
cd airfoil2D
./Allrun
That's the whole thing. cd airfoil2D enters the directory the archive extracted to; ./Allrun is OpenFOAM's own run script.
Not needed here, OpenFOAM is open source. For commercial solvers you can check Use Existing License and enter your license server details, which Rescale remembers for future jobs. See create and submit a job.
Choose hardware
Three fields for a basic job:
One core is plenty for this mesh. Walltime is a ceiling, not a reservation. The job stops if it exceeds it, but you're only billed for the time actually used, so an hour is a safe over-estimate.
Skip the post-processing panel; there's nothing to filter on a run this small.
Review and submit
Check the summary, then click Submit.
A required field is missing. Work back through the panels. The usual culprits are no software version selected, or an empty command. See the greyed-out submit button.
Watch it run
Open the Status tab. The job moves through five stages: Queued → Validating Input → Starting Cluster → Running Job → Stopping Cluster, each getting a green check as it passes.
Starting Cluster takes a few minutes. That's normal, Rescale is provisioning hardware for you. See the job lifecycle for what each stage is doing.
Once it reaches Running Job, use Live Tailing to watch output as the solver writes it. Two files are worth opening:
Residuals dropping steadily means the solve is converging.
Get your results
When the job completes, open the Results page. You'll see logs, solution data, and any plots the case produced.
- Download Job takes everything as one zip — inputs and outputs.
- Per-file actions let you view (under 1 MB in-browser), download, or delete individual files.
- Make selected files available as input files promotes an output into an input for a later job. This is how you chain a run: use one job's result as the next job's starting point.
Check the end of process_output.log for the exit code. 0 means your command exited cleanly, though it doesn't prove the results are good.
What you did
You configured a job, submitted it, watched a solver converge in real time, and pulled the results back. That's the same loop for a 4-core case and a 4,000-core one: only the hardware line changes.
- Run it again with a change — duplicate the job and edit one parameter.
- Sweep a parameter instead of guessing — DOE jobs run the same case across a range of values.
- Do it repeatedly — save this setup as a template.
- Post-process interactively — open the results on a workstation instead of downloading them.
Related
- Create and submit a job — the reference version of these steps
- Monitoring & status — everything about watching a run
- Download your results — bulk and scripted downloads
- Troubleshooting — if the job failed