Self-healing: claims that recover on their own

When a machine drops out mid-storm

Static routes break at the worst moments: a tractor breaks down, a driver doesn't show, a phone dies, or GPS drops out in a parking structure. With a fixed route, every property behind that machine is now stuck, and a dispatcher has to notice and re-assign it by hand.

Snowmass heals this automatically.

How self-healing works

Every claim carries a lifetime, and a background job runs continuously (about every 60 seconds) to check the fleet's health:

• Silence release — if a tractor has gone quiet for 5+ minutes and the claim's time-to-live has expired, the engine releases that claim. The property drops back into the open pool and the swarm re-scores it for a healthy machine.
• Wedge healing — if a glitch leaves a deployment holding two on-property claims at once (a "missed-exit" wedge), the engine detects and untangles it so no property is stuck behind a stale claim.

Because claiming is atomic, this recovery is safe: a released property can be re-claimed by exactly one healthy tractor, with no risk of double-booking.

Dispatcher recall

A dispatcher doesn't have to wait for the timeout. If a machine breaks down, they can recall it explicitly (recall reason: breakdown), and its claimed work flows straight back to the swarm — instantly re-dispatched to the machines still running.

Why it matters

• Breakdowns and no-shows self-absorb — the exact moments manual routes fall apart are handled without a radio scramble.
• No orphaned properties — work never gets stuck behind a machine that's gone dark.
• Less dispatcher firefighting — the fleet re-balances itself instead of waiting for a human to notice.