Deploying and watching
A deployed tron is a real container. It has logs, a state, and a stop button.
The four states
running, sleeping, idle and error. Sleeping is woken automatically by the server; idle is not. They are not the same thing and nothing should treat them as one. The state has one source, and every page reads the same one. A sleeping tron is never stopped because time has passed.
Caps
Free: 1 tron, 1 awake, 3 hours awake a day, 3 agents per tron, 500 MB. Personal: 5 trons, 2 awake at once, 5 GB, the Tron API. Pro: 15 trons, 4 awake, 15 GB. Max: 30 trons, 8 awake, 50 GB. A company tron follows the company's caps (Enterprise), set for it. The caps apply to what you do next: nothing you already have is taken away. Beside the plan, the machine has a cap of trons awake at once; sleeping trons do not count.
Logs
Every tron keeps its log on the server, across redeploys and stops: the page reads the newest lines, older ones a page at a time, and filters by agent, kind and text on the server. Clear empties it for everyone, and leaves a line saying who and when. A tron on a runtime older than 0.20.0 shows only its container's recent lines, and the page says so.
Uptime
Only Free has a daily cap of time awake: 3 hours. Past it, deploying or waking the tron is refused until the next day (429 daily_uptime_exhausted).
Stopping
A stop takes the container and its network down and frees the tron's port and network. The workspace, the memories and the logs stay; the next deploy starts from the saved canvas.