NerdMiner – solo.ckpool.org DNS Resolution Failure

Cold power-cycle the Nerdminer for 30 full seconds — pull the USB cable, wait 30 seconds for capacitors to drain and the ESP32 LWIP DNS cache to fully clear, then reconnect. Five seconds is not enough — LWIP DNS table state can persist in SRAM longer than expected and a soft reboot does not always wipe it. This single step fixes the 'DNS cache wedged after a router reboot' pattern that bites operators who restart network gear without restarting the miner. If the worker registers within 60 seconds of power-up, you are done.

Verify the pool hostname spelling and TLD letter-by-letter. Open the Nerdminer's WiFi config portal (5-second BOOT button hold puts most boards into AP mode for re-config). Inspect the pool URL field. Compare letter-by-letter to `solo.ckpool.org`. Common typos: `solockpool.org`, `solo.ckpool.com` (wrong TLD), `solo.ckpool.org/` (trailing slash breaks parsing on some firmware builds). A single missing or extra character produces NXDOMAIN every time, and the config portal does not validate the hostname before saving.

Confirm the port is `3333`, not `21496` or `4334`. `solo.ckpool.org` listens on port `3333` for stratum. `21496` is `public-pool.io`. `4334` is some braiins variants. Pasting the wrong port from a guide written for a different pool is the second most common operator error at the Nerdminer setup screen — DNS resolves fine but the TCP socket connects to nothing on the wrong port and the firmware logs a generic 'connection failed' that looks identical to a DNS issue.

Switch to a known-good public pool as a sanity check. Temporarily change the pool URL to `public-pool.io` with port `21496` and a worker name `<your-bc1q-segwit-address>.nerdminer`. Public-Pool is permissive about workers and well-known clean DNS-wise. Cold-cycle the miner. If the worker registers at Public-Pool but not at `solo.ckpool.org`, the issue is specifically DNS resolution of `solo.ckpool.org` from your network — and the fix is router DNS or hardcoded IP.

Check the worker / address format. `solo.ckpool.org` requires your own Bitcoin address as the worker name (anything you mine to is paid directly to that address — no account, no signup). Use a `bc1q…` SegWit address. Do NOT use a `bc1p…` Taproot address — CKPool's payout path historically rejects bech32m / Taproot. Do NOT use a Lightning address. Wrong address format produces an immediate disconnect that can look like DNS failure if you were not watching the serial log carefully.

Change router DNS to Cloudflare (`1.1.1.1` and `1.0.0.1`) or Quad9 (`9.9.9.9` and `149.112.112.112`). Log into your router admin (typically `http://192.168.1.1`). Find DNS settings under WAN, Internet, or Network. Replace ISP defaults with Cloudflare or Quad9 primary + secondary. Save. Reboot the router, then cold-cycle the Nerdminer for 30 seconds so it picks up the new DNS via DHCP. ISP DNS is the single most common cause of intermittent Nerdminer DNS failures — Cloudflare and Quad9 are free, fast, and do not hijack `NXDOMAIN`.

Disable DNS-rebinding protection on the router or allow-list `solo.ckpool.org`. Many consumer routers (UniFi, ASUS, Netgear, eero, OPNsense, pfSense) enable DNS-rebinding protection by default — silently filtering DNS responses where a public hostname resolves to an RFC1918 IP. Find the setting under Security or Firewall. Either disable globally or add an exception for `solo.ckpool.org` and `*.ckpool.org`. Pi-hole / AdGuard Home / NextDNS users: open the query log, search `solo.ckpool.org`, confirm no 'crypto,' 'mining,' or 'P2P' category is blocking it. Allow-list explicitly.

Hardcode the CKPool IPv4 into the NerdMiner_v2 pool config. From a desktop on the same LAN, run `nslookup solo.ckpool.org 1.1.1.1` to get the current authoritative IPv4. In the Nerdminer's config portal, replace `solo.ckpool.org` with the bare IPv4. Save and reboot. The worker should register within 60 seconds. CKPool's IP set is stable for weeks at a time — refresh the hardcoded IP every month or two via `nslookup` and update the firmware config. Bulletproof way to remove DNS entirely from the Nerdminer's path.

Set a DHCP reservation for the Nerdminer with explicit DNS pushed. In your router's DHCP table, find the Nerdminer's MAC (look for hostname `nerdminer-xxxx`, `T-Display-S3`, or an Espressif OUI). Reserve a fixed IP. In advanced DHCP options, push specific DNS servers (Cloudflare, Quad9) directly to that reservation. Bypasses any ISP-gateway-level DNS hijacking and ensures the Nerdminer always lands on the same IP — handy for monitoring scripts and for telling at a glance which board is which.

Verify with `nslookup` from the same LAN before and after the router DNS change. Run `nslookup solo.ckpool.org 192.168.1.1` (querying your router) and compare against `nslookup solo.ckpool.org 1.1.1.1` (querying Cloudflare directly). If the router returns a different IP than Cloudflare, the router is rewriting DNS — fix the router config or bypass it with the DHCP-reservation trick in step 9. If the router returns the same IP as Cloudflare and the Nerdminer still fails, the issue is firmware-side (Tier 3) not network-side.

Reflash the latest stable NerdMiner_v2 via the official web-flasher at `https://bitmaker-hub.github.io/NerdMiner_v2/`. Connect a USB **data** cable (charge-only cables will not enumerate) to a Chrome or Edge browser. Select the correct board variant from the dropdown — `T-Display-S3`, `T-Dongle-S3`, `ESP32-2432S028R` (CYD), `ESP32-WROOM-32`, M5StickC, etc. Wrong board variant = wrong pinout = bricked TFT or no boot. Click 'Erase Flash' before reflashing — wipes NVS, old WiFi credentials, old pool config, and any wedged LWIP DNS state. Then flash the latest stable `.bin` for your board.

Capture a serial boot log at `115200 baud`. USB data cable, serial terminal (PuTTY on Windows, `screen /dev/ttyACM0 115200` on Linux/macOS, Arduino Serial Monitor cross-platform). Reboot the Nerdminer. Save the first 60 seconds of output. Search for these lines: `WiFi connected`, `[NTP]`, `lwip_getaddrinfo:`, `[POOL]`. The error code on `lwip_getaddrinfo` tells you the failure class: `1` = no such host (NXDOMAIN), `11` = would-block / retry timeout (DNS server too slow), `4` = name truncated (typo or buffer issue). Cross-reference the ESP-IDF LWIP API docs.

Run a stratum proxy on your LAN to take public DNS out of the Nerdminer's path entirely. Options: `cgminer --proxy`, `stratum-mining-proxy`, or a Raspberry Pi running a small Node.js or Go relay. Point the Nerdminer at the proxy's bare IPv4 and a local port (e.g. `192.168.1.50:3333`). The proxy resolves `solo.ckpool.org` using the host OS's resolver (full IPv6 support, larger cache, proper retry logic), then forwards stratum to the Nerdminer over a bare-IP socket. If shares get through via the proxy but not direct, the issue is 100% the Nerdminer's LWIP DNS resolver — not CKPool, not credentials. Right architecture for 3+ Nerdminers.

Test for IPv6-only `AAAA` records and force IPv4 at the resolver. Run `dig solo.ckpool.org A @1.1.1.1` and compare to `dig solo.ckpool.org AAAA @1.1.1.1`. CKPool publishes both `A` and `AAAA` records — if your specific resolver path is only returning `AAAA`, that is the trap. Two fixes: (a) hardcode the IPv4 found via Cloudflare's resolver into the Nerdminer firmware, (b) change router DNS to a resolver that returns `A` records preferentially (Cloudflare and Quad9 both do). If both `A` and `AAAA` come back from Cloudflare but you still see IPv6-only behavior, your router is filtering responses.

Audit your DNS-blocking layer for over-blocking. Pi-hole admin → Query Log; AdGuard Home → Filtering → Logs; NextDNS → Logs. Search for `ckpool.org`. If it shows blocked (red), check which list flagged it. Common offenders: NextDNS's 'crypto' category, certain hagezi or OISD blocklist subcategories on Pi-hole / AdGuard. Allow-list `*.ckpool.org` explicitly. Also check for wildcard rules (`*.io`, `*.org`) in custom block lists that might catch legitimate pools — over-broad regex is a common cause of 'I added a blocklist last week and now my Nerdminer broke' tickets.

Stop DIY and contact D-Central support when (a) you have reflashed the latest stable NerdMiner_v2 twice with 'Erase Flash' between attempts, hardcoded the CKPool IPv4, switched router DNS to Cloudflare, and allow-listed `solo.ckpool.org` on every DNS-blocking layer — and DNS still fails; (b) the serial log shows `lwip_getaddrinfo` errors even when the Nerdminer config contains a hardcoded IPv4 (impossible mode pointing at corrupted flash, partition-table damage, or LWIP regression); (c) multiple Nerdminer units on the same network all fail DNS simultaneously while desktops resolve cleanly. The D-Central Discord and `BitMaker-hub/NerdMiner_v2` issue tracker are both right escalation channels.

What D-Central does at this point: bench-side reproduction with identical Nerdminer board variant (T-Display-S3, T-Dongle-S3, CYD on the bench), identical NerdMiner_v2 firmware build, controlled network with known-good DNS, full serial trace including `CONFIG_LWIP_DEBUG` output if needed (custom build with debug flags). If we reproduce the failure, we file an issue against `BitMaker-hub/NerdMiner_v2` with the trace and a reproducer. D-Central stocks the full Nerdminer family and contributes back to the open-source pleb-mining ecosystem — maintainers triage hardware-grade reports faster than vague tickets.

Ship to D-Central if you want bench-grade diagnosis. Anti-static bag, padded mailer or small box. Include a note: Nerdminer board variant (T-Display-S3 / T-Dongle-S3 / CYD / M5StickC / other), firmware version, pool URL exactly as configured, photo of the config portal, your router make/model, and what you have already tried. We test on a controlled bench network, isolate firmware-vs-network-vs-hardware, return with a full report. Most 'Nerdminer DNS fail' tickets close without hardware repair because the issue is upstream — but bench reproduction proves it definitively. Turnaround typically 3–7 business days for diagnostic-only tickets.