IceRiver Error 800/801/802 Firmware Update Failed

Stop retrying. Resist the urge to mash the upgrade button five more times. Each retry on a half-corrupt partition state increases the risk of corrupting both partition slots — recovery then jumps from a five-minute fresh-flash to a 30-minute SD-card reflash. Wait two minutes between attempts and read the rest of this list before retrying.

Hard-reboot at the breaker for 30 seconds. Not a soft reboot from the web UI — full mains kill. Clears wedged daemon state, flushes /tmp, resets the OTA daemon. After the miner comes fully back, verify `Status` reads your previous firmware version. If yes, the active partition is healthy and you can retry safely. If no, you're already at Tier 3.

Re-download the firmware fresh from https://www.iceriver.io/firmware-download/. Use a different browser if the first download produced an `801`. Compare downloaded byte size to the portal's listed size — mismatch = corrupt download, get it again. This is the single most common `801` cause we see, and it's a 60-second fix.

Verify the firmware filename matches your model exactly. `KS3` ≠ `KS3L` ≠ `KS3M`. `KS5` ≠ `KS5L` ≠ `KS5M`. IceRiver signs per-model and rejects cross-model images with `801` by design. Triple-check filename, file size, and IceRiver's published model-firmware mapping before flashing.

Move to a wired connection on the same subnet as the miner. Plug both miner and workstation into the same switch via Ethernet. Disable Wi-Fi on the workstation, close every non-essential browser tab, disable VPN. Eliminates the entire `802` transfer-layer drop class — the most common `8xx` cause overall.

Verify network stability before retrying. Run `ping -t <miner-IP>` from the workstation for two minutes; verify zero packet loss. Any packet loss on a wired short-haul ping means bad cable, flaky switch port, or NIC issue — fix that before continuing. A single dropped packet during upload can throw `802`.

Free up /tmp on the miner if you have SSH access. Some IceRiver firmware versions leave failed-upgrade artifacts in /tmp that fill the partition, blocking subsequent OTA attempts. SSH in, `df -h /tmp`, remove leftover .bin or .tar files, retry. Not all KS firmware enables SSH out-of-the-box; skip if yours doesn't.

Roll back to the previous published firmware version. If failure happens specifically on the latest IceRiver build, that build may have an issue on your hardware revision. Flash the previous version from the portal. If it succeeds, sit on the older build until IceRiver publishes a fix. Most 7xx/8xx post-upgrade codes clear with rollback to last-known-good.

Hold the reset button 20 seconds for factory reset. Write down network and pool config first — reset clears them. Red status LED flashes during the hold to confirm. After reset, the miner returns at DHCP. Re-enter config, then attempt the upgrade fresh on wired. Factory reset clears stale state in the OTA daemon and config database.

Verify AC line voltage at the wall under load. A marginal residential circuit sagging below 190 V AC during upgrade can fault the OTA daemon mid-write. Multimeter on AC, probe at the outlet during a hashing-load period. IceRiver PSUs accept 180-285 V AC. Sag below 190 V = move to a dedicated 240 V circuit before retrying.

SD-card reflash recovery — the canonical recovery path. Download the model-specific recovery .img from the IceRiver portal (not all models publish recovery images publicly — if yours doesn't, contact IceRiver support). Write to microSD with balenaEtcher (cross-platform) or `dd` (Linux/macOS). Insert into the miner's recovery slot (typically on the control board behind a chassis panel — varies per model), hold the boot/recovery button while powering on, release after 10 seconds. The miner reads the image off SD and re-flashes the eMMC. Soak-test for an hour at nameplate hashrate after recovery.

Do not cross-flash within a model family. Within the KS3 family (KS3, KS3L, KS3M), recovery images are similar but NOT drop-in compatible — flashing a KS3M image onto a KS3L will brick. Same applies across the KS5 family. Don't cross-flash unless IceRiver support has explicitly authorized it, and verify the file checksum even then.

Capture the eMMC partition table for support escalation. If you have SSH access on a partially-recovered miner, run `cat /proc/mtd` and `dmesg | grep -i mmc` and screenshot the output. Tells IceRiver / D-Central support whether the partition table is recoverable or needs a fresh image flash.

Re-attempt the upgrade only after a clean SD-reflash recovery. Once the miner is back at all-LEDs-green on a known-good firmware, wait 30 minutes of stable operation before attempting the next upgrade. Don't immediately retry the same image that failed twice — verify portal-listed file size, exact model match, and a wired connection first.

If anti-rollback lockout is suspected (post-xyys/tswift), ask IceRiver support directly. Third-party overclocking firmware can leave the eMMC in a state where official firmware refuses to install on signature grounds. Fix is typically a special recovery image from IceRiver support — open a ticket via the FAQ portal contact form, include model, serial, log output, and third-party firmware history. Don't try to brute-force the official upgrade.

Stop DIY and ship to a bench when: (a) SD-card reflash fails or recovery image isn't published for your model, (b) the control board shows visible damage from a power event during upgrade (burnt smell, scorched traces near the eMMC), (c) anti-rollback lockout is confirmed by IceRiver support and the recovery image they offer doesn't take, or (d) repeated 8xx cadences across multiple recovery attempts. Past those points DIY costs more in time and risk than the repair.

D-Central bench process: test fixture with programmable load and serial console access, eMMC re-imaging via JTAG / direct programmer if SD recovery fails, control-board replacement with eMMC content imaged across, post-repair 24-hour burn-in at nameplate hashrate before sign-off. Western retail repair authority for the KS line — only Western alternative is shipping to China (Zeus), and trust profile of that path is not great.

Ship safely. Pack the unit in anti-static bags, double-box with at least 5 cm foam on every side. Include the original PSU and AC cable so we can reproduce the AC environment. Critically, include a printed upgrade history: previous firmware version, attempted version, exact codes and substrings from the log, recovery steps already attempted, third-party firmware history. Saves us diagnostic time, which saves you money.