Files
HODA/.env.agent.example
TMCandClaude Opus 5.5 c58cc329b0 Deploy with root docker-compose files, like the other repos
docker-compose.yml runs HODA on the main host, published on one address
(10.20.30.10:10040 in .env.example) for the Pangolin resource. The agent
settings are optional, so a single host needs nothing else.
docker-compose.agent.yml is for a second Docker host. Both follow FiMa's
layout: required values fail with a clear message, pull_policy always,
memory limits, capped logs, a read-only root filesystem.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-02 22:55:06 +01:00

25 lines
828 B
Bash

# HODA on a second Docker host (the agent). Copy to .env next to docker-compose.agent.yml.
# The main host then sets HODA_AGENT_URL=http://<HODA_AGENT_BIND_IP>:<HODA_AGENT_PORT>.
# Secret shared with the main host: the same value as its HODA_PEER_TOKEN.
HODA_PEER_TOKEN=
# This host's LAN IP. The agent's port is published on it only, e.g. 10.20.30.11
HODA_AGENT_BIND_IP=
# Port the main host reaches the agent on.
HODA_AGENT_PORT=8081
# Name of this host on tiles and in search.
HODA_NODE_NAME=host-2
# Image tag: latest, a release such as 1.0.0, or a build such as sha-9d6a921.
HODA_TAG=latest
# Bind-mount the icon cache to a host directory instead of a named volume
# (sudo chown -R 65532:65532 the directory first).
# HODA_DATA_DIR=/home/msemes/AppData/hoda-agent
# debug logs every icon lookup.
HODA_LOG_LEVEL=info