Antminer S21 Pro Overheating — "ERROR_TEMP_TOO_HIGH: over max temp": Real Heat or a Bad Reading?

What the shutdown actually tells you

The Antminer S21 Pro is air-cooled and was released in 220, 234 and 245 TH/s variants, drawing roughly 3,300, 3,510 and 3,675 W (±5%) at about 15 J/TH. Almost all of that electrical power leaves the machine as heat, so the fans and the airflow through the heatsinks are doing real work all the time.

When a temperature the miner watches goes past its limit, the miner protects itself and stops. It writes a line such as ERROR_TEMP_TOO_HIGH: over max temp to the kernel log, normally with the board and chip temperature readings next to their limits. The limits differ between models and firmware versions, so always read them from your own log.

Wording can vary slightly between firmware versions. The important point is that the message tells you the miner believes something is too hot. It does not tell you whether that is true.

Step 1: is it real heat?

Work through these in order. Most over-temperature stops on a healthy board end here.

1. Step 1

   Save the log first

   Copy the kernel log before you power the miner off or restart it. A restart erases it, and the temperatures in the log are your best clue.

2. Step 2

   Compare the three boards

   Look at the board and chip temperatures on each of the three hashboards. All three hot points at the site or the fans. One board always hotter than the others points at that board.

3. Step 3

   Measure the air at the front grille

   Intake air that is too warm is the most common cause. Many S21-series guides treat about 35 °C at the intake as the practical ceiling for air-cooled units. Check that hot exhaust air is not being pulled back into the front.

4. Step 4

   Look at the airflow path

   Dust-packed heatsinks, blocked grilles, gaps that let hot air recirculate, or weak extraction in a room or container all raise chip temperatures.

5. Step 5

   Check every fan

   A fan that is slow, noisy or reading far below the others cuts the airflow through the heatsinks. If the log shows ERROR_FAN_LOST, start here.

When the heat is not real

A shutdown is triggered by the temperature the miner reads, not the temperature the board actually has. Every hashboard has temperature sensors, and the miner reads them through the board’s own electronics. If that reading is wrong, or cannot be read at all, the miner can protect itself when nothing is hot. Support threads for Bitmain miners include cases where read failures from a board came before an over-temperature stop.

You cannot clean your way out of this one. These are the signs:

Checks you can do yourself

1. Let it cool completely before you start it again. Bitmain’s guidance warns against repeatedly restarting a miner without removing the cause of the high temperature.
2. Save the kernel log and note which chain and which readings were over their limits.
3. Check the intake air and the airflow as described above, and clean the heatsinks and fans with dry compressed air.
4. Reseat the fan and board connectors after powering off, and look for heat marks on the connectors.
5. Swap the suspect board to another slot. Whether the fault follows the board or stays in the slot is the quickest way to split a board fault from a slot or cable fault.
6. Return any tuning to stock and run it for a while before drawing conclusions.

If the shutdown comes back with cool air, clean heatsinks and good fans, stop restarting it. A board that reports heat it does not have, or really does run hot with fresh paste and clear airflow, needs a bench.

How we handle S21 Pro temperature faults

The first job is to decide which of the two cases you have. On the bench we compare the board’s temperature readings against what it is really doing under load, using thermal imaging, and we check the fans and the board’s voltage domains. Board-level faults, including controller and EEPROM problems and failed chips, are repaired at chip level, and the board runs a full-load burn-in before it goes back.

A solo miner is typically repaired in 1–3 days, and small batches within 1–2 weeks. Farms can have the work done on site, without shipping the machines. If your unit is still under Bitmain’s warranty, use that first.

Different symptom on the same machine? See our guide to S21 Pro missing chips, our S21 Pro / S21+ / S21 XP repair page, or our S21 / T21 repair service. Power faults on the same units? We also repair APW17 power supplies.

Frequently asked questions

What does "ERROR_TEMP_TOO_HIGH: over max temp" mean on an S21 Pro?

A temperature the miner monitors went past its limit, so the miner stopped hashing to protect the hardware. The line is normally followed by the board and chip temperature readings next to their limits. The message alone does not say whether the heat is real or whether the miner is reading a bad value, so check both.

Is it safe to keep restarting an S21 Pro that shows over max temp?

No. Bitmain’s own guidance warns against repeatedly restarting a miner without first removing the cause of the high temperature. Let it cool, find out why it got hot, and only then start it again. Repeated restarts on a hot or faulty board can make the damage worse.

What intake air temperature is acceptable for an S21 Pro?

Keep it as low as you can. Many S21-series troubleshooting guides use about 35 °C at the front grille as the practical ceiling for air-cooled units, and warmer air is a very common cause of over-temperature stops. Hot air from the back of the machine being drawn into the intake is the classic trap.

Can a faulty sensor make an S21 Pro shut down for overheating?

Yes. A shutdown is triggered by the temperature the miner reads, not the temperature the board actually has. If a hashboard’s temperature reading is wrong or cannot be read, the miner can protect itself when nothing is really hot. Support threads show cases where read failures from a board came before an over-temperature stop. That is a board-level fault, and cleaning will not cure it.

How do I know if the board or the slot is at fault?

Move the suspect board to another slot. If the shutdown follows the board, the fault is on the board. If it stays with the slot whichever board is installed, look at the control board connection and cable instead.

How long does S21 Pro overheating repair take?

At Frontline Hash, a solo miner is typically repaired in 1–3 days and small batches within 1–2 weeks, depending on the fault and parts. On-site repair is also available for farms.