archy/reticulum-daemon
ssmithx beff5dd577 fix(mesh): stop the flash-triggered device boot-loop, fix stuck/wedged jobs
Three real incidents from live-testing the flash feature on archy-x250-dev,
each traced through logs on the node and fixed at the root:

- A failed flash left the mesh listener auto-resuming into a reconnect loop
  whose backoff reset to its 5s minimum on any momentary connection, even
  mid-boot-loop — every retry's open() toggles DTR/RTS, which resets many
  ESP32 boards, so the retries were themselves sustaining the loop. Backoff
  now only resets after a session runs stably for 20s+, and the flash job
  no longer auto-resumes the listener after a failure (only on success,
  after a settle delay).

- The HTTP client's blanket 30s request timeout covered entire downloads
  (Meshtastic's zip is ~170MB), killing large transfers mid-stream; fixed
  with a per-chunk stall timeout instead of a fixed total-transfer cap. But
  the download's initial request had no timeout at all, so a slow-to-start
  server hung it forever — wedging the single-flash-job guard permanently
  ("already in progress" on every subsequent attempt). Fixed with a bounded
  wait for the response to start, plus an absolute ceiling around the whole
  job as a last-resort safety net.

- archy-rnodeconf's PyInstaller freeze broke its own internal esptool
  invocation (sys.executable pointed at the frozen binary itself instead of
  a real interpreter) — fixed via a new runtime hook. esptool's own
  auto-reset-into-bootloader handshake is separately known to be flaky on
  some CP2102/CH340 boards; it now retries once at a conservative baud
  rate instead of failing outright, and failure logs show the full error
  chain instead of just the outer context message.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
2026-07-23 02:18:55 +00:00
..

reticulum-daemon

Host-supervised Reticulum (RNS) + LXMF bridge for Archipelago's Mesh tab. This is the Python side of the Reticulum transport plan: archipelago spawns one of these per active Reticulum (RNode) radio, it owns the serial port, and the Rust mesh subsystem drives it over a Unix-socket JSON-RPC.

Why a daemon (not the Rust reticulum-rs crate): the canonical Python rns/lxmf guarantees interop with Sideband / NomadNet / MeshChat, and lets us derive the RNS identity from the existing Archy key (proven in spike_identity.py).

Layout

  • archy_rns_identity.py — derive a deterministic RNS Identity from the 32-byte Archy Ed25519 seed (identity_dir/node_key) via domain-separated HKDF. The node's LXMF destination hash is a stable function of the Archy identity.
  • spike_identity.pyPhase-0 gate #1 (no radio): proves that determinism.
  • reticulum_daemon.py — the daemon: RNS bring-up, LXMF router, announce handler, and the Unix-socket RPC. See its module docstring for the wire protocol.
  • requirements.txt — pinned rns==1.3.5, lxmf==1.0.1 (validated on Python 3.13).

Dev setup

python3 -m venv .venv
.venv/bin/pip install -r requirements.txt

Run the spike / smoke tests (no hardware)

.venv/bin/python spike_identity.py                     # gate #1: identity determinism
.venv/bin/python reticulum_daemon.py --check \
    --identity-key /path/to/node_key                   # print this node's dest hash
.venv/bin/python reticulum_daemon.py --selftest \
    --identity-key /path/to/node_key                   # bring up RNS+LXMF, no radio

Run against a real RNode (Phase-0 hardware gate, on .116 / .228)

.venv/bin/python reticulum_daemon.py \
    --identity-key /var/lib/archipelago/identity/node_key \
    --serial-port /dev/reticulum-radio \
    --socket /run/archy/reticulum.sock \
    --display-name "archy-228"

Then verify a two-node LXMF DM over LoRa and interop with a stock Sideband/MeshChat client (Phase-0 gates #2 and #3).

Packaging (Phase 1)

Ship as a PyInstaller single binary in the OTA next to /usr/local/bin/archipelago (no provision-time pip install). archipelago supervises it: start on RNode detect, kill on unplug/disable. The RPC socket and RNS config dir are archipelago-owned, 0600.

./build.sh   # → dist/archy-reticulum-daemon (~16M, fully standalone)

-d noarchive is required, not optional — see the comment in build.sh: RNS computes RNS.Interfaces.__all__ via a glob() against its own __file__ directory at import time, which only works when PyInstaller keeps modules as loose files instead of zipping them into the binary.

Status

Phase-0 gate #1 (identity determinism) passes, verified in both the dev venv and the packaged binary (same dest hash). The signed-identity announce (ARCHY:2:{ed}:{x25519} in _announce_app_data, via --archy-ed-pubkey-hex/--archy-x25519-pubkey-hex) is wired and the Rust side (reticulum.rs) already passes the node's real keys through. Packaging is done and verified standalone. What's left is entirely hardware-dependent: the live LoRa message path (Phase-0 gates #2/#3) needs a real RNode-flashed board.