Node Operator SLA
The contract between your node and the EZLINK platform. Read before going live — non-compliance affects your trust score, payouts, and dispute outcomes.
What your agent firmware MUST do
Heartbeat every ≤ 60 seconds
POST to /api/public/heartbeat/<nodeId> at least once a minute while the node is powered on, whether or not anyone is connected. We mark a node offline after 5 minutes of silence.
Accounting tick every ≤ 30 seconds per active session
For each connected client, POST byte deltas to /api/public/accounting-tick/<nodeId>. Acceptable cadence is 10–30 seconds. Slower than 60 seconds will be treated as a missing-tick incident.
Send cumulative deltas since the last tick, not absolute counters. If the agent restarts, start the next delta from zero.
Honor disconnect responses
When the tick response returns should_disconnect: true for a session, the client must be removed from the network within 10 seconds. Ignoring this lets users consume data they no longer have, and the loss is charged to the operator.
Why the cadence matters (for you)
Slow ticks = over-credits charged to you
If your agent ticks every 5 minutes instead of every 30 seconds and the node then drops, EZLINK auto-credits the user based on whatever balance was last reported. Their actual balance was lower — the gap is data they consumed on your bandwidth that you can no longer bill for.
We mitigate this with an interpolation guard (we estimate unbilled usage from the session's average rate while the agent was ticking, capped at 30 minutes), but the guard's accuracy depends on having had at least 60 seconds of ticks before silence. A node that goes silent in the first minute gets no interpolation and credits the full remaining balance.
Silent agent = treated as offline
Router still powered but agent process dead is the worst case: clients keep consuming data while we believe the node is offline. After 5 minutes of agent silence we mark the node offline and start issuing service credit to every active user. Your earnings for the silent window are forfeit.
Trust score impact
Trusted (85+)
Standard payout schedule, dispute decisions default to your favor when evidence is split, priority placement in the user map.
Standard (60–84)
Standard payout schedule. No placement boost.
Probation (40–59)
Payouts held 7 days, dispute decisions default against you when evidence is split, capped daily revenue until score recovers.
Restricted (< 40)
Node hidden from the user map. Existing sessions allowed to finish, no new sessions. Payouts held until manual review.
Score = 100 − kicks (×2, cap 30) − disputes lost (×5, cap 30) − refunds against your node (×3, cap 20), recomputed daily over a rolling 30-day window.
Consequences for SLA breach
First incident: auto-credit charged to your earnings, trust score adjusted, no other action.
Repeated incidents (3+ in 30 days): node moves to probation, payouts held 7 days, you receive a compliance review request.
Persistent breach (10+ in 30 days): node moves to restricted. You can dispute in Comp pass deductions with packet captures, ISP outage notices, or power-utility reports.