ASIC Miner Uptime: Measure It and Improve It

ASIC miner uptime should mean more than a web interface that answers. Separate scheduled time, powered availability, local hashing and productive pool-accepted time. A miner can be online while one board is missing. At the same time, shares are rejected or while a collector displays stale data.

Create a reason code for every lost interval, reconcile local and pool timestamps, and calculate availability and productive utilisation separately. Improvement then follows the largest controllable loss, whether planned maintenance, electrical trips, heat, network, pool configuration, firmware restarts or repair delay.

Define scheduled, available and productive time

Choose the denominator. A miner deliberately curtailed for uneconomic electricity may be unavailable but correctly controlled; report it separately from an unplanned fault.

Define a productive threshold by exact model and mode, such as minimum accepted hashrate with all required boards. Avoid one universal percentage across different miners.

Align site, dashboard and pool clocks. Record data freshness so a collector outage is not mistaken for a complete mining outage.

Write the intended outcome before looking at a headline hashrate. A learning device, a useful room heater, a quiet home miner and a commercially productive machine are different purchases. The correct comparison changes when the available circuit, sound limit, heat demand, pool route or expected ownership period changes.

Use a dated decision sheet and keep manufacturer claims separate from measured results. Record the exact model, variant, power supply, firmware and operating mode. Similar product names do not make accessories, voltage, firmware or thermal limits interchangeable.

Build trustworthy outage and asset evidence

Collect local board and hashrate state, power or PDU status, pool last share and accepted work, network reachability, temperature protection and maintenance tickets.

Use stable asset and worker identities. A control-board replacement or worker rename should not reset the machine’s operating history.

Create reason codes for planned maintenance, economic curtailment, utility, PDU, miner hardware, heat, network, pool, firmware, configuration and unknown loss. Keep unknown as a visible problem rather than guessing.

Prefer the manufacturer specification, manual and firmware portal for identity and limits. But treat them as the starting point rather than a promise of site performance. Keep a copy of the pages and files used because support pages, downloads and product revisions can change.

Ask the seller for a serial photograph, condition statement, included accessories and a recent operating record for the actual unit. A generic product image cannot prove board revision, power supply condition, repair history or whether the miner reaches stable accepted work.

Prepare alerts, spares and recovery routes

Set alerts for sustained offline or low-accepted-work conditions and route them to an owner with a runbook. Avoid immediate alerts for pool noise that create fatigue.

Hold model-specific spares and safe isolation procedures based on failure history. Faster diagnosis is not useful when the site cannot safely remove and replace a failed unit.

Provide resilient network and approved backup pool endpoints, while avoiding uncontrolled automatic changes to wallets or customer workers.

A competent person should confirm the electrical route for the real continuous load. Check voltage, protective device, earthing, cable, connector, socket, isolation and ventilation together. Do not assume that a plug physically fitting a socket proves that the circuit is suitable for sustained operation.

Place the miner on a trusted network segment with no unnecessary inbound exposure. Change supplied credentials, use a documented wallet and pool account, set approved backup endpoints and confirm that every endpoint belongs to the intended operator before power is applied.

Calculate availability and accepted-work loss

Availability equals available scheduled hours divided by scheduled hours. Productive utilisation can use accepted terahash-hours divided by expected terahash-hours for the approved mode. Report both.

Weight incidents by lost capacity and duration. Ten minutes on a large miner may matter more than an hour on a display device, and a whole-site collector failure needs separate treatment.

Review mean time to detect, acknowledge, repair and verify. Closure should require restored accepted work, not just a reboot command or a green local status.

Measure power at the wall and compare local hashrate with accepted pool work over a representative period. Local display figures can look healthy while stale shares, invalid work, reconnects or a wrong payout address reduce useful output.

Calculate revenue and cost over a range, not one favourable day. Include electricity, pool fees, auxiliary cooling, maintenance, downtime, conversion costs and hardware value. For a heat-use case, credit only heat that replaces a cost the owner would otherwise incur.

Control misleading uptime records

Rank each risk by consequence and by the practical ability to detect it before purchase. A low-priced machine with uncertain firmware, exhausted cooling or a weak algorithm market can require more working capital and attention than a newer unit with a higher invoice price.

Set written stop conditions. Examples include an unsafe supply, unavailable official firmware, rejected work above the approved limit, repeated thermal shutdown, no lawful payout route or an energy break-even price below the contracted rate. A stop condition prevents sunk cost from becoming the reason to continue.

Improve one measured cause at a time

Audit one month for one site, reconcile miner and pool intervals, and classify the largest losses. Sample tickets against logs to find missing or inconsistent reason codes.

Implement one targeted control, then compare the same metric over another representative period. Do not claim improvement from a cooler month without accounting for season and scheduled curtailment.

Begin with one unit or the smallest sensible batch. Photograph labels and connections, export the original configuration, note ambient conditions and record the start time. Watch the kernel or system log, board detection, fan behaviour, temperatures, local hashrate, pool connection and accepted work.

Do not declare acceptance from a short dashboard snapshot. Run long enough to expose heat soak, intermittent network faults and pool variance. Retain the test record with the invoice, serial number, firmware file and any seller correspondence so a later repair or warranty question has a clear baseline.

ASIC uptime reporting checklist

• Confirm the exact model, variant, condition and included power equipment.
• Verify official specifications, instructions and the correct firmware route.
• Approve the continuous electrical load, airflow, heat and sound plan.
• Test network isolation, credentials, pool endpoints and payout ownership.
• Compare wall power with accepted work over a representative run.
• Model downside revenue, electricity, downtime, maintenance and resale.
• Record acceptance limits and a safe stop or return route.
• Reassess whenever firmware, network economics or site conditions change.

The checklist is deliberately evidence based. Marketing language such as home friendly, efficient or profitable has no fixed meaning without a measured operating mode and a real site boundary. The record should make it possible for another competent person to reproduce the decision.

Key points to remember

Useful uptime measures delivered mining work, not just an online page. Define the schedule and productive threshold, reconcile pool and local evidence, classify every loss and close incidents only after accepted work returns. Improvement comes from removing the largest repeated cause and proving the change over comparable conditions.

Next steps

Use The Mining Shop UK tools and support pages to compare the exact hardware against your real electricity, installation, pool and operating constraints before ordering or commissioning it.