How to set up auto-restart for mining rigs? (System Stability)

BIOS-Level Auto-Restart Configuration

1. Access the BIOS interface during boot by pressing Delete or F2 repeatedly until the setup menu appears.

2. Navigate to Advanced → Power Management Setup and locate AC Power Recovery or Restore AC Power Loss.

3. Change the value from Power Off or Last State to Power On.

4. Save changes using F10 and confirm with Yes to trigger immediate reboot with new settings applied.

5. Verify functionality by disconnecting and reapplying main power — the rig must boot without manual intervention.

Operating System Service Management

1. Create a systemd service file at /etc/systemd/system/miner.service for Linux-based rigs.

2. Define Environment variables including MINER_HOME, GPU_DRIVER_PATH, and LOG_DIR to ensure consistent runtime context.

3. Set ExecStart to point directly to the miner binary with full CLI arguments including pool URL, wallet address, and device flags.

4. Configure Restart=always and RestartSec=45 to enforce recovery after any termination, including segmentation faults or OOM kills.

5. Enable the service via systemctl daemon-reload && systemctl enable miner.service to persist across reboots.

Hardware Watchdog Integration

1. Enable the hardware watchdog timer in BIOS under Security or Advanced → Peripheral Configuration.

2. Install watchdog daemon packages: apt install watchdog on Debian/Ubuntu or yum install watchdog on CentOS/RHEL.

3. Configure /etc/watchdog.conf to monitor critical processes such as ethminer or lolMiner using pidfile directives.

4. Set watchdog-device to /dev/watchdog and adjust max-load-1 to 8.5 to prevent false triggers under sustained GPU load.

5. Start and enable the watchdog service so it activates early in boot and resets the system if miner process disappears for more than 60 seconds.

Network-Driven Failover Logic

1. Deploy a lightweight Python script that pings the primary mining pool endpoint every 30 seconds using ICMP and HTTP HEAD requests.

2. If three consecutive failures occur, execute systemctl restart miner.service and log timestamped event to /var/log/miner-failover.log.

3. Maintain secondary pool credentials in encrypted config and switch stratum URL dynamically when primary remains unreachable for over 180 seconds.

4. Inject real-time share submission latency metrics into the decision loop — delays exceeding 5000ms for five cycles trigger local restart regardless of pool status.

5. Route all failover logs to a remote syslog server for correlation with temperature, voltage, and PCIe link width telemetry.

Frequently Asked Questions

Q: Can auto-restart cause wear on SSDs used for mining OS?A: Frequent restarts do increase write cycles, but modern NVMe drives rated for 300 TBW can sustain 12 restarts per day for over 6 years without failure.

Q: Does restarting the miner binary affect pending shares in flight?A: Shares submitted before process termination are retained by the pool; only unsubmitted shares held in local memory buffers are lost — typically less than 0.7% per restart cycle.

Q: Is it safe to enable AC Power Recovery on rigs with dual PSUs?A: Yes — both PSUs respond independently to AC restoration signals; no synchronization delay occurs, and redundancy remains fully functional post-boot.

Q: Why does my rig reboot but fail to initialize GPUs after auto-restart?A: This indicates insufficient PCIe training time; add pci=realloc=off and video=vesafb:off to kernel boot parameters in GRUB to stabilize enumeration order.