Stratum Endpoint Migration Runbook for ASIC Fleets

Stratum endpoint migration runbook is a focused operating decision rather than another general introduction to mining. This guide concentrates on whether an ASIC fleet can move to a new pool endpoint without losing worker identity, payout control or useful work. It uses official endpoints, DNS and latency tests, worker formats, payout ownership and matched accepted-share exports as the evidence boundary, applies a staged endpoint change with independently verified destination and immediate rollback as the control, and measures the result through accepted availability, reject reasons, latency and credited payout across the migration window. That separation prevents the article from competing with the broader guide already established on The Mining Shop UK.

Define the exact decision and boundary

Reassess Stratum endpoint migration runbook whenever network conditions, firmware, tariffs or official guidance changes.

The purpose of Stratum endpoint migration runbook is to decide whether an ASIC fleet can move to a new pool endpoint without losing worker identity, payout control or useful work. It does not need to repeat every feature of ASIC mining, hardware selection or profitability. Write the question, owner, equipment, location, start time and intended outcome before opening a dashboard or changing a configuration.

State what remains outside the decision. Hardware purchase price, electricity, pool method, tax, warranty and site safety may influence the answer, but they should not be silently mixed into one favourable number. Separate boundaries make the record understandable months later.

The working boundary uses official endpoints, DNS and latency tests, worker formats, payout ownership and matched accepted-share exports. Record whether each item is measured, contractual, manufacturer supplied, observed from a pool, or assumed. An assumption should never be presented with the authority of a meter, signed term or accepted-share export.

Name the exact model, serial, firmware, operating mode and time zone where hardware is involved. Similar family names can hide different control boards, power supplies, cooling systems or software requirements. A precise identity prevents advice for one variant being applied to another.

Collect evidence that can be reproduced

Start with official endpoints, DNS and latency tests, worker formats, payout ownership and matched accepted-share exports. Save the source, date, version, units and any exclusions. A screenshot can support the record but should not replace the underlying export, manual, contract, invoice or meter interval when that evidence is available.

Prefer primary technical, regulatory and contractual sources. Community reports can reveal a possible issue, but they do not override the current manufacturer instructions, protocol rules, utility terms or applicable law. Record uncertainty where primary evidence does not answer the question.

Match time periods before comparing results. Pool work, wall energy, temperatures, incidents, reward and cost must cover the same interval. Unmatched windows can create a convincing but false explanation for a change in performance.

Ask a second person to reproduce the conclusion from the saved evidence. If they cannot confirm accepted availability, reject reasons, latency and credited payout across the migration window without a verbal explanation from the original operator, the evidence pack is incomplete.

Apply controlled changes and approvals

The principal control is a staged endpoint change with independently verified destination and immediate rollback. Assign one person to authorise the change and another to review high-impact settings such as firmware, pool endpoints, payout destinations, electrical modes or customer allocations.

Establish a stable baseline before changing anything. Export the configuration, photograph labels and connections, preserve official recovery material and record pool accepted work. Change one variable at a time so the result has a defensible cause.

Keep miners on a separated trusted network with no unnecessary public management access. Use unique credentials and multi-factor authentication for pool, cloud and administrator accounts. A technical improvement is not acceptable if it weakens account or network security.

Set a rollback and safe-stop rule in advance. Excess temperature, unsafe electrical behaviour, unexplained destination changes, repeated restarts, increased rejects or lost recovery should stop the test rather than becoming an invitation to keep experimenting.

Measure the result on a complete boundary

The primary measure is accepted availability, reject reasons, latency and credited payout across the migration window. Record numerator, denominator, units, period and source. Avoid percentages without a base and averages that hide downtime, warm periods, failed starts or excluded fees.

Use complete wall energy rather than catalogue power for an operating conclusion. Use pool accepted work rather than local hashrate when measuring useful mining output. Keep gross reward, variable contribution, total operating profit and capital return as separate layers.

Run long enough to expose heat soak, ordinary share variance, reconnects and intermittent faults. A short successful screenshot proves that the interface opened; it does not establish stable performance or a reliable commercial process.

Compare the measured result with the written acceptance limit and the baseline. If accepted availability, reject reasons, latency and credited payout across the migration window worsens, investigate before accepting the change. Record an inconclusive result honestly rather than selecting only the favourable part of the interval.

Control the failure modes

The topic-specific failure to control is work silently fails over to an old account or unverified destination. Define its earliest observable signal and the person authorised to act. Waiting for a complete outage, rejected warranty or lost payout is not a monitoring strategy.

Rank consequences separately from likelihood. A rare event involving unsafe power, wallet authority or irrecoverable firmware deserves a strong preventive control even when it has not occurred on the current unit.

Retain failed tests. They show the limits of a model, site or process and prevent another technician from repeating the same change without knowing the result.

Run a representative acceptance test

The acceptance exercise is to move one worker, force primary failure, confirm backup behaviour and reconcile the first payout. Begin with one unit, one customer allocation or the smallest practical batch. Record the baseline, exact change, start and finish time, environmental condition and person responsible.

Observe board detection, controller logs, fan or pump behaviour, temperatures, wall power, network connection, pool accepted work and payout ownership. The exact list changes by topic, but the evidence should connect physical operation with the intended commercial result.

Test one controlled failure such as endpoint loss, restart, sensor alarm, rejected approval or rollback. Confirm that the system enters the documented safe state and that the audit trail remains available afterwards.

Close the test only when another reviewer can reproduce accepted availability, reject reasons, latency and credited payout across the migration window, all exceptions are resolved or accepted in writing, and the rollback path remains usable. Expansion before acceptance multiplies uncertainty rather than proving scale.

Final implementation checklist

• Write the exact Stratum endpoint migration runbook decision, owner and scope.
• Record model, serial, firmware, mode, account and time zone.
• Collect official endpoints, DNS and latency tests, worker formats, payout ownership and matched accepted-share exports from current primary sources.
• Preserve the stock configuration and official recovery route.
• Apply a staged endpoint change with independently verified destination and immediate rollback with recorded approval.
• Change one variable and retain the complete baseline.
• Measure accepted availability, reject reasons, latency and credited payout across the migration window across a representative matched period.
• Test for work silently fails over to an old account or unverified destination and verify the safe response.
• Reconcile technical, safety and commercial evidence separately.
• Approve wider deployment only after independent review.

The checklist is designed to create an auditable decision, not a perfect-looking score. When the evidence conflicts, keep the conflict visible and obtain the missing measurement or advice. A postponed decision is safer than an irreversible change based on an attractive dashboard.

Frequently asked questions

What is the purpose of Stratum endpoint migration runbook?

It provides a narrow method for whether an ASIC fleet can move to a new pool endpoint without losing worker identity, payout control or useful work, using evidence and acceptance limits rather than repeating a general mining guide.

What evidence should be saved?

Save official endpoints, DNS and latency tests, worker formats, payout ownership and matched accepted-share exports, with dates, units, versions and the exact hardware or account identity.

Should several settings be changed together?

No. Establish a stable baseline and change one controlled variable so the result and rollback remain understandable.

Which result matters most?

Use accepted availability, reject reasons, latency and credited payout across the migration window on a matched period, alongside safety, security, warranty and contractual limits.

When should the work stop?

Stop when there is evidence of work silently fails over to an old account or unverified destination, unsafe operation, lost recovery, destination uncertainty or a breached acceptance limit.

Can this process guarantee mining profit?

No. Reward, difficulty, price, fees, energy, uptime, repair and hardware value can all change. The process improves evidence, not certainty.

Conclusion

Stratum endpoint migration runbook is useful when it owns one clear decision: whether an ASIC fleet can move to a new pool endpoint without losing worker identity, payout control or useful work. Build the record from official endpoints, DNS and latency tests, worker formats, payout ownership and matched accepted-share exports, apply a staged endpoint change with independently verified destination and immediate rollback, and judge the outcome through accepted availability, reject reasons, latency and credited payout across the migration window. Preserve failure evidence and rollback. That discipline creates a distinct, defensible guide without competing with the site’s broader established article.

Next steps

Use The Mining Shop UK tools, policies and support pages to check the exact hardware, site and commercial boundary before making the change or approving wider deployment.