Run DCENT_OS in 15 Minutes — No Miner Required

Before you start

You do not need: a miner, a pool account, a wallet, or a soldering iron.

The run

1. Get the source

   git clone https://github.com/DCentralTech/DCENT_OS.git
   cd DCENT_OS

   You should see

   A working tree containing DCENT_OS_Antminer/, DCENT_OS_ESP/, and TESTING.md — the file this tutorial follows.

   Proves the published tree is public and complete. Undo delete the folder.

2. Install the pinned toolchain and type-check everything

   cd DCENT_OS_Antminer/dcentrald
   rustup toolchain install 1.90.0
   cargo +1.90.0 check --workspace --locked

   You should see

   Cargo compiles every crate in the workspace and exits without errors. --locked means you built exactly the dependency set we ship — no silent substitutions.

   Proves the workspace compiles from a cold clone. Undo nothing to undo.

3. Run the test suite — the same one our CI runs

   cargo +1.90.0 test --workspace --locked

   You should see

      Compiling dcentrald v…
       Finished `test` profile [unoptimized + debuginfo] target(s) in …
        Running unittests …
   
   test result: ok. … passed; 0 failed; … ignored

   Exact test counts change from release to release, so we don't print one — the contract is the ok and the 0 failed on every crate's summary line. The suite exercises the Stratum V1/V2 clients against in-process mock pools, share validation, BIP 320 version-rolling reconstruction, ASIC frame codecs, autotuner logic, and the API contracts — all without hardware.

   Proves the mining logic passes CI on your machine — not just ours. Undo nothing to undo.

   If the build or tests fail here

   A failure in this step on a clean checkout with the pinned 1.90.0 toolchain is exactly the kind of result the project wants reported — open a GitHub issue with your OS, toolchain version, and the tail of the output. Most first-run failures are environment, not code: confirm rustup show lists 1.90.0, and on WSL make sure you cloned inside the Linux filesystem, not /mnt/c (a general WSL performance/locking caveat, not a DCENT_OS-specific one).

4. Build the dashboard and start it on mock telemetry

   cd ../dashboard
   npm ci
   npm run build
   python3 scripts/preview-with-mocks.py --port 4173

   You should see

   A local server on port 4173 backed by ~70 mocked API endpoints with realistic mining data. Your terminal stays attached to the server — leave it running.

   Proves the shipped UI is real, not renders. Undo Ctrl-C stops the server.

5. Drive the product

   Open http://127.0.0.1:4173 and switch between the three modes: Space Heater / Mining / Hacker.

   You should see

   The full product UI — the autotuner panel, the shares table with pool-target vs achieved difficulty, the Companion chat, the Hacker-mode tools — rendered from mock data, no miner required. The telemetry is mocked; the interface is real. On real hardware, the difference is that the numbers are yours.

   Proves what you'd actually operate day-to-day. Undo close the tab, Ctrl-C the server.

6. Optional: cross-compile the real firmware binary

   cd ../dcentrald
   cargo build --release --target armv7-unknown-linux-musleabihf    # Zynq (S9/S17/S19)

   You should see

   An ARMv7 release binary under target/armv7-unknown-linux-musleabihf/release/ — the artifact class that runs on a Zynq control board. "A successful build proves the published tree is complete and standalone." (If cargo reports the target is missing, rustup target add armv7-unknown-linux-musleabihf installs it — standard rustup usage, not a DCENT_OS-specific step.)

   Proves you could produce the miner-side binary yourself. Undo nothing to undo.

7. Stop

   Press Ctrl+C in the terminal running the preview server. That's the whole teardown: nothing was installed system-wide, nothing is listening anymore, and the checkout is just a directory you can delete.

   Proves the exit is as small as the entry. Undo already done.

What you just proved — and what you didn't

Where this ladder goes next

Deeper: build from source · architecture · security model.