Frequent Restarts — Troubleshooting Guide

Overview

A miner that continuously reboots is losing 100% of its potential hashrate during each restart cycle. Depending on the model, a full boot cycle takes 2–8 minutes, and if the miner restarts every 5–15 minutes, the effective uptime (and revenue) can drop to 50% or less.

Frequent restarts are always a symptom of an underlying protection mechanism triggering: over-temperature, overcurrent, watchdog timeout, or a hardware fault causing the control board to crash. Identifying which protection mechanism is responsible is the key to a fast resolution.

Symptoms

• Miner reboots every few minutes — uptime counter in the web UI never exceeds a consistent threshold
• Periodic hashrate drops to zero on the pool dashboard, followed by recovery
• Fans spin up, then everything powers off and restarts — the classic PSU overcurrent protection trip
• Web UI becomes unreachable periodically — the control board is rebooting
• LED indicators cycle through boot sequence repeatedly
• Antminer: Red LED blinks, then green, then red again in a loop
• Whatsminer: Status LED cycles through startup pattern, or "heartbeat" LED stops and restarts
• Avalon: AUC disconnects and reconnects in the management software

Quick Checks

# SSH into the miner quickly after a reboot
ssh root@<miner-ip>

# Check kernel log for shutdown reason
dmesg | tail -50

# Check miner log for error before shutdown
cat /tmp/log/bmminer.log | tail -50

# Check for temperature-related shutdowns
grep -i "temp\|overheat\|shutdown" /tmp/log/bmminer.log

# Check watchdog status
cat /proc/watchdog

ssh root@<miner-ip>

# Check miner log
tail -100 /tmp/btminerlog

# Look for fault codes
grep -i "fault\|error\|restart\|reboot" /tmp/btminerlog

Diagnostic Flowchart

Miner keeps rebooting
│
├─ Can you read error logs before reboot?
│  ├─ YES → What error appears last?
│  │  ├─ "Temp too high" / "overheat" → Temperature issue.
│  │  │  Go to Cause #1 below.
│  │  ├─ "Power supply fault" / PSU error → PSU issue.
│  │  │  Go to Cause #2 below.
│  │  ├─ "Watchdog" / no specific error → Software/firmware crash.
│  │  │  Go to Cause #3 below.
│  │  ├─ "Voltage error" / "chain error" → Hashboard fault.
│  │  │  Go to Cause #4 below.
│  │  └─ Kernel panic / segfault → Control board hardware issue.
│  │     Go to Cause #5 below.
│  └─ NO (reboots too fast to read logs) ↓
│
├─ Does the miner run longer with fewer hashboards installed?
│  ├─ YES → A hashboard is causing PSU overcurrent or
│  │        a board-level fault triggers the reboot.
│  │        Test each board individually.
│  └─ NO ↓
│
├─ Does the miner reboot even with NO hashboards installed?
│  ├─ YES → Control board or PSU issue (not hashboard-related).
│  │        Try a different PSU. If still rebooting,
│  │        control board fault.
│  └─ NO → Hashboard-related issue confirmed.
│          Install boards one at a time to identify the culprit.

Common Causes (Ordered by Probability)

1. Over-Temperature Shutdown and Restart

Probability: ~30%

The miner heats up under load, reaches the thermal shutdown threshold, powers off to protect the chips, cools down briefly, and then firmware initiates a restart. This cycle repeats indefinitely.

Indicators:

• Chip temperatures reach 85–95°C before each restart
• The miner runs longer in cooler ambient conditions
• Fans are running at 100% speed
• Restarts are more frequent during summer or midday

Root causes behind overheating:

• Ambient temperature too high (above 35–40°C)
• Dust-clogged heatsinks reducing airflow efficiency
• Failed fan(s) reducing total airflow
• Degraded thermal paste after 12+ months of operation
• Recirculating airflow (hot exhaust feeding back into intake)
• Miner positioned too close to a wall or other miners

Fix:

2. PSU Overload or Failure

Probability: ~25%

The PSU cannot sustain the miner's power demand. When the load exceeds the PSU's capacity, its overcurrent protection (OCP) trips and cuts power. The miner powers off completely, the PSU recovers, power returns, and the miner boots again — only to trip OCP again once hashing begins.

Indicators:

• Restarts happen 2–5 minutes after boot (once hashing at full speed begins)
• A "click" or "pop" sound from the PSU just before power loss
• All LEDs go dark simultaneously (versus just the control board rebooting)
• PSU is hot to the touch
• Problem worsens if you increase the miner's frequency/power target

How to diagnose:

1. Measure PSU output voltage under full load — if it dips below spec (e.g., below 11.5V for a 12V PSU), the PSU is failing
2. Try the miner with a known-good PSU
3. Try running with only 1 or 2 hashboards — if the miner is stable, the PSU cannot handle full load

Fix: Replace the PSU. Ensure the replacement is rated for at least 10% more than the miner's maximum power draw. For older PSUs, capacitor degradation is the most common cause of reduced capacity — capacitor replacement can restore the PSU if you have the skills, but PSU replacement is faster and more reliable.

3. Firmware Bug / Watchdog Reset

Probability: ~18%

The miner's firmware crashes or hangs, and the hardware watchdog timer resets the control board. This is a software-level issue — the hardware may be perfectly fine.

Indicators:

• The restart interval is fairly consistent (e.g., always around 10–15 minutes)
• Kernel log shows "watchdog" or "wdt" messages before the reset
• The problem appeared after a firmware update
• No temperature or power anomalies in the logs before the reset
• The miner runs stable on a different firmware version

Fix:

4. Hashboard Short Causing PSU Protection Trip

Probability: ~12%

A short circuit on one hashboard can draw enough excess current to trip the PSU's overcurrent protection, shutting down all boards. This looks like a PSU issue but is actually caused by the hashboard.

Indicators:

• Restarts happen under load (same timing as PSU overload)
• Removing one specific board makes the problem disappear
• The suspect board has visible damage or burns
• PSU output voltage is fine when tested without the faulty board

How to diagnose:

1. Remove all hashboards
2. Install them one at a time, testing stability with each
3. The board that causes restarts when installed is the culprit
4. Measure resistance across the power input of the suspect board (should be 2–50 ohms; less than 0.5 ohms indicates a short)

Fix: Locate and repair the short on the hashboard. See Abnormal Chip Voltage Troubleshooting for techniques to find shorted components on a hashboard.

5. Control Board Memory or Hardware Issue

Probability: ~8%

The control board itself has a hardware fault — failing NAND flash, bad RAM, or a degraded processor — causing crashes.

Indicators:

• Restarts happen even with no hashboards installed
• Kernel log shows memory errors, NAND errors, or filesystem corruption messages
• The problem persists across different firmware versions
• The miner was subjected to a power surge or lightning strike

Fix: Replace the control board. Control board-level repair (NAND replacement, RAM replacement) is possible but requires specialized equipment and is typically not cost-effective compared to a replacement board.

6. Electrical Issues (Input Power)

Probability: ~7%

Unstable input power to the PSU — voltage sags, brownouts, or poor wiring — can cause the PSU to lose input and restart.

Indicators:

• Other devices on the same circuit also experience issues
• Restarts correlate with times of high electrical demand (air conditioning starting, other equipment turning on)
• Restarts are more frequent at certain times of day
• The miner is on a long extension cord or undersized wiring

Fix: Ensure the miner is on a dedicated circuit with appropriate amperage. Check all connections from the wall outlet to the PSU input. For mining farms, install a UPS or power conditioner to filter input power quality.

How to Read Kernel Logs via SSH

The kernel log is your best friend for diagnosing restart causes. Here is how to capture it effectively:

# Quick connection after reboot
ssh root@<miner-ip>

# Save the kernel log immediately
dmesg > /tmp/saved_dmesg.txt

# Save the miner application log
cp /tmp/log/bmminer.log /tmp/saved_bmminer.log 2>/dev/null

# On the miner (Antminer example)
echo "*.* @<syslog-server-ip>:514" >> /etc/syslog.conf
/etc/init.d/syslog restart

When to Seek Professional Help

Consider professional assistance if:

• The miner reboots even with no hashboards and a known-good PSU — this strongly indicates a control board failure
• Kernel logs show persistent NAND or memory errors — the control board storage is failing
• You suspect electrical wiring issues at the facility level — have an electrician inspect the circuit
• The miner went through a power surge or lightning event — multiple components may be damaged simultaneously

Frequently Asked Questions

How do I tell the difference between a PSU restart and a control board restart?

A PSU restart causes a complete power loss — all LEDs go dark, fans stop, and everything powers up from scratch. A control board restart keeps the PSU running (fans may continue spinning briefly) and only the control board reboots. Watch the fans: if they stop completely during each restart, it is a PSU issue. If they briefly stutter but the PSU stays on, it is a control board issue.

My miner runs fine for hours then restarts once and runs fine again. Should I worry?

An occasional single restart (once per day or less) is often caused by a transient event — a brief power fluctuation, a pool connection timeout triggering a firmware restart, or a single temperature spike. If it is truly rare and the miner recovers and runs normally, it is not urgent. Monitor the frequency — if restarts increase over time, investigate.

Can firmware corruption cause a restart loop?

Yes. If the firmware image is partially corrupted, the miner may boot partially, encounter the corrupt section, crash, and reboot. This creates a consistent restart loop where the miner never fully starts. The fix is to reflash firmware via SD card (Antminer), USB recovery (Whatsminer), or TFTP (Avalon).

Why does my miner restart more in summer?

Higher ambient temperature causes faster thermal rise, reaching the shutdown threshold sooner. A miner that barely stays within thermal limits at 25°C ambient will thermally throttle and restart when ambient reaches 35°C+. Solutions: improve ventilation, add intake cooling, reduce power target during hot months, or switch to immersion cooling.

Is it safe to disable the watchdog timer?

No. The watchdog timer exists to recover from firmware crashes. Disabling it means that if the firmware hangs, the miner will stay hung indefinitely with zero hashrate instead of rebooting and recovering. The watchdog is a safety net, not the problem — fix the root cause of the crash instead.

Related Guides

• High Temperature Troubleshooting
• Power Issues Troubleshooting
• Fan Error Troubleshooting
• Abnormal Chip Voltage
• Hashboard Not Detected
• Antminer Error Codes
• Whatsminer Fault Codes