Run a Python script online, day and night
A script that only runs when your laptop is open is not really running. Put it on a panel and it runs in the cloud on Python 3.11, either continuously or on a schedule, with dependencies installed for you.
Continuous or on a schedule
A worker panel runs forever and is restarted on failure. It suits a bot, a queue consumer, or a watcher with its own sleep loop, and you control the timing inside your own code.
A scheduled panel is started for you hourly, every six hours, or daily, and your script is expected to finish and exit. It suits a scraper, a report, a backup, or a daily digest, and since each run is a fresh process, state belongs in a file or a database.
Getting a script running
- Create the panel. Worker for something always on, scheduled for a timed job. Set the runtime to Python.
- Add your files and requirements.txt. Dependencies install when the panel starts, so there is no pip command to run and no virtualenv to manage.
- Point the start file at your script. main.py, app.py, whatever you called it. Secrets go in environment variables rather than in the file.
- Start it and read the log. Output streams live while it runs and is kept after it exits, with the exit code, so a failed run is still diagnosable tomorrow.
The things that catch people out
- Buffered output. Use print with flush=True, or run python with -u, or the logs lag behind reality.
- Times are UTC. Do the conversion yourself if a job should fire in your local morning.
- A scheduled run must exit. If it never returns, the next run will not start cleanly.
- Pin your dependencies in requirements.txt so a fresh install does not pull a breaking release.
- Write state to a file or a database, not to memory, if the process restarts or is scheduled.
- Keep an eye on RAM if you load a large file or a dataframe, since the panel limit is a hard ceiling.
Let ChatGPT or Claude set it up
Add Redon3 as a connector and your assistant can write the script, write requirements.txt, set the runtime and start file, store the API keys as encrypted environment variables, start the panel, and read the traceback if it fails.
It cannot delete the panel, spend money, or read a stored secret back. So handing it the deploy is a smaller decision than it sounds.
Common questions
- How do I run a Python script continuously online?
- Create a worker panel with the Python runtime, point the start file at your script, and press start. It runs until you stop it and is restarted on failure up to five times.
- Can I run it on a schedule instead?
- Yes. A scheduled panel runs your script hourly, every six hours, or daily, and the script is expected to finish and exit each time.
- Which Python version is it?
- Python 3.11, with dependencies installed from requirements.txt when the panel starts.
- Why is my log empty when the script is clearly running?
- Buffered stdout, almost always. Add flush=True to your prints, or set the start command to run python with -u.
- Can I use a database?
- Yes, any database you can reach over the network. Put the connection string in an environment variable. There is also panel storage on disk if a file is enough.
- Is there a free trial?
- Seven days on a free trial panel, no card, which is plenty to get a script running and watch a few cycles.