What an active rental commits you to
Keep the accepted hardware, storage, accelerator access, connectivity, and advertised environment available until the renter ends the rental or the contract reaches its end date. Do not mine, benchmark, render, or run provider workloads on capacity assigned exclusively to a renter. Do not inspect tenant files, processes, or network traffic except where strictly required for an authorized security incident.
The provider view shows active and historical rental counts and cumulative provider revenue for each listing. Cluster Activity Monitor events offer a second operational timeline.
Uptime and marketplace evidence
The marketplace displays uptime windows for approximately 1 hour, 1 day, 1 week, and 1 month. Short outages can therefore remain visible after service recovers. Treat these windows as an operational signal rather than a vanity score. Stable service improves renter confidence and reduces interrupted work.
Monitor power, temperatures, ECC or GPU errors, disks, RAM pressure, upstream connectivity, WireGuard handshakes, k3s readiness, and available capacity. A ready badge alone does not replace host monitoring.
Plan maintenance
- 01
Close future demand
Delist the affected node or cluster so no new contract can start. - 02
Drain contracts, not just workloads
Wait for every starting and active rental to end. A quiet GPU does not mean the renter has released it. - 03
Perform and test the change
Reboot, replace hardware, update firmware or drivers, then run local validation and a renter-style test. - 04
Refresh evidence
Synchronize nodes and wait for current inventory and benchmarks before relisting.
Respond to an outage
First preserve the host and platform error, the time, affected node, cluster, and rentals. Restore power or connectivity without changing hardware identity. Verify WireGuard, k3s-agent, GPU health, storage, and time synchronization. The platform can pause billing while a node outage is detected and resume after readiness returns.
When preparing a replacement operating system, consult the recommended Linux releases and recheck the GPU driver before joining.
If repair requires a disruptive driver or hardware change, contact support and avoid repeated blind reinstalls. Record what changed. Once recovered, verify the renter connection path and the activity timeline.
Earnings and automatic payouts
Hosts set the earnings price of their resources. ComputeCubed adds a listing-specific markup to calculate the complete renter price. Settlement credits your agreed host earnings to Earnings Balance without a ComputeCubed commission. Top-up processing costs are paid by ComputeCubed and never charged to hosts.
Connect an eligible Stripe account for automatic payouts. Periods close on the 1st UTC, and transfers are attempted on the 15th. A first partial month joins the next full month. There is no ComputeCubed listing fee or withdrawal fee. ComputeCubed pays platform-billed Stripe costs; Stripe may charge your connected account directly under its agreement. Transfer completion does not establish the exact bank receipt. Statements and transaction details show earnings, included rentals, Stripe references and balance changes. See billing and automatic payouts.
Reviews
After an eligible completed rental, the renter can review the relevant workspace, node, or cluster. Ratings aggregate across related entities where the Console indicates. Providers can reply to reviews. Keep replies factual, avoid exposing tenant information, and use support channels—not public review text—for security or billing disputes.