diff --git a/ai/hermes-workspace/Dockerfile b/ai/hermes-workspace/Dockerfile index 2f40c9a..9530f0a 100644 --- a/ai/hermes-workspace/Dockerfile +++ b/ai/hermes-workspace/Dockerfile @@ -29,10 +29,12 @@ RUN pnpm install --frozen-lockfile && pnpm build # ---------- Stage 2: Hermes Agent + Workspace runtime ---------- FROM nousresearch/hermes-agent:latest -# ---------- Install Node.js + tmux for Workspace + Swarm ---------- +# ---------- Install tmux for Swarm workers + curl for health checks ---------- +# Note: Node.js is already shipped with the base hermets-agent image; apt's nodejs +# would be older. Only add what's missing. USER root RUN apt-get update && apt-get install -y --no-install-recommends \ - nodejs tmux ca-certificates curl \ + tmux curl \ && rm -rf /var/lib/apt/lists/* # ---------- Overlay our forked Hermes source ---------- @@ -115,6 +117,8 @@ USER hermes ENV HERMES_HOME=/opt/data ENV PATH="/opt/data/.local/bin:${PATH}" ENV CHROME_EXECUTABLE=/opt/hermes/.playwright/chromium/chrome-linux/chrome +ENV HOST=0.0.0.0 +ENV NODE_ENV=production RUN chown -R hermes:hermes /opt/hermes/tools /opt/hermes/toolsets.py diff --git a/ai/hermes-workspace/entrypoint-combined.sh b/ai/hermes-workspace/entrypoint-combined.sh old mode 100644 new mode 100755 index 614b535..b6be66b --- a/ai/hermes-workspace/entrypoint-combined.sh +++ b/ai/hermes-workspace/entrypoint-combined.sh @@ -1,33 +1,154 @@ #!/bin/bash -set -e +# ── Hermes Workspace Combined Entrypoint ── +# Waits for the Hermes gateway container (hermes:8642) to become healthy, +# then starts the Hermes Workspace web UI in the foreground. +# Supports graceful shutdown via SIGTERM/SIGINT. +# ────────────────────────────────────────── -# ── Hermes Workspace + Swarm Worker Entrypoint ── -# Starts Hermes Workspace web UI (port 3000) and makes -# hermes CLI + tmux available for Swarm workers. -# The Hermes gateway runs in a separate container (hermes:8642). -# Swarm workers spawned here connect to the gateway via HTTP. -# ────────────────────────────────────────────────────────── +set -euo pipefail -# Install custom tools from persistent volume -if [ -f /opt/data/hermes-tools/install.sh ]; then - bash /opt/data/hermes-tools/install.sh || true -fi +# ── Configuration ────────────────────────────────────────────── +GATEWAY_HOST="${GATEWAY_HOST:-hermes}" +GATEWAY_PORT="${GATEWAY_PORT:-8642}" +GATEWAY_URL="http://${GATEWAY_HOST}:${GATEWAY_PORT}" -# Wait for Hermes gateway to be healthy before starting workspace -if [ -n "${HERMES_API_URL:-}" ]; then - echo "Waiting for Hermes gateway..." - for i in $(seq 1 30); do - if curl -fsS "${HERMES_API_URL}/health" >/dev/null 2>&1; then - echo "Gateway healthy after ${i}s" - break +HEALTH_ENDPOINT="${HEALTH_ENDPOINT:-/health}" +MAX_RETRIES="${HEALTH_MAX_RETRIES:-60}" +RETRY_INTERVAL="${HEALTH_RETRY_INTERVAL:-2}" + +WORKSPACE_DIR="${WORKSPACE_DIR:-/workspace}" +WORKSPACE_ENTRY="${WORKSPACE_ENTRY:-server-entry.js}" + +PID_FILE="${PID_FILE:-/tmp/workspace.pid}" + +# ── Logging ──────────────────────────────────────────────────── +log_info() { echo "[$(date '+%Y-%m-%d %H:%M:%S')] [INFO] $*"; } +log_warn() { echo "[$(date '+%Y-%m-%d %H:%M:%S')] [WARN] $*"; } +log_error() { echo "[$(date '+%Y-%m-%d %H:%M:%S')] [ERROR] $*"; } + +# ── Graceful Shutdown ────────────────────────────────────────── +_workspace_pid="" +_shutting_down=false + +cleanup() { + if [ "$_shutting_down" = true ]; then + return fi - if [ "$i" -eq 30 ]; then - echo "WARNING: Gateway not healthy after 30s, starting workspace anyway" - fi - sleep 1 - done -fi + _shutting_down=true -# Start Hermes Workspace in foreground -cd /workspace -exec node --max-old-space-size=2048 server-entry.js + log_info "Shutdown signal received, cleaning up..." + + # Stop workspace process if running + if [ -n "$_workspace_pid" ] && kill -0 "$_workspace_pid" 2>/dev/null; then + log_info "Stopping workspace (PID: $_workspace_pid)..." + kill -TERM "$_workspace_pid" 2>/dev/null || true + + # Give it time to shut down gracefully + local wait_sec=10 + while kill -0 "$_workspace_pid" 2>/dev/null && [ "$wait_sec" -gt 0 ]; do + sleep 1 + wait_sec=$((wait_sec - 1)) + done + + # Force kill if still running + if kill -0 "$_workspace_pid" 2>/dev/null; then + log_warn "Workspace did not shut down gracefully, force killing..." + kill -KILL "$_workspace_pid" 2>/dev/null || true + fi + fi + + # Clean up PID file + [ -f "$PID_FILE" ] && rm -f "$PID_FILE" + + log_info "Shutdown complete." + exit 0 +} + +# Trap termination signals for graceful shutdown +trap cleanup SIGTERM SIGINT + +# ── Gateway Health Check ─────────────────────────────────────── +wait_for_gateway() { + local url="${GATEWAY_URL}${HEALTH_ENDPOINT}" + local retries="$MAX_RETRIES" + local interval="$RETRY_INTERVAL" + local attempt=0 + + log_info "Waiting for Hermes gateway at ${GATEWAY_URL}..." + log_info "Max retries: ${retries}, interval: ${interval}s" + + while [ "$attempt" -lt "$retries" ]; do + attempt=$((attempt + 1)) + + if curl -fsS "${url}" >/dev/null 2>&1; then + log_info "Gateway is healthy after ${attempt} attempt(s) (${GATEWAY_URL})" + return 0 + fi + + if [ "$attempt" -lt "$retries" ]; then + log_info "Gateway not ready yet (attempt ${attempt}/${retries}), retrying in ${interval}s..." + sleep "$interval" + fi + done + + log_error "Gateway did not become healthy after ${retries} attempts ($((retries * interval))s)" + return 1 +} + +# ── Workspace Startup ────────────────────────────────────────── +start_workspace() { + local entry="${WORKSPACE_DIR}/${WORKSPACE_ENTRY}" + + if [ ! -d "$WORKSPACE_DIR" ]; then + log_error "Workspace directory not found: ${WORKSPACE_DIR}" + return 1 + fi + + if [ ! -f "$entry" ]; then + log_error "Workspace entry point not found: ${entry}" + return 1 + fi + + log_info "Starting Hermes Workspace web UI..." + log_info " Directory: ${WORKSPACE_DIR}" + log_info " Entry: ${entry}" + + cd "$WORKSPACE_DIR" + + # Start workspace in background so we can trap signals + node --max-old-space-size=2048 "${entry}" & + _workspace_pid=$! + echo "$_workspace_pid" > "$PID_FILE" + + log_info "Workspace started (PID: ${_workspace_pid})" + + # Wait for workspace process + wait "$_workspace_pid" + local exit_code=$? + + log_info "Workspace exited with code ${exit_code}" + return "$exit_code" +} + +# ── Main ─────────────────────────────────────────────────────── +main() { + log_info "=== Hermes Workspace Combined Entrypoint ===" + log_info "Gateway: ${GATEWAY_URL}" + log_info "Workspace: ${WORKSPACE_DIR}/${WORKSPACE_ENTRY}" + log_info "PID file: ${PID_FILE}" + + # Wait for gateway to be healthy + if ! wait_for_gateway; then + log_warn "Proceeding without confirmed gateway health..." + fi + + # Start the workspace + start_workspace + local exit_code=$? + + log_info "Entrypoint exiting with code ${exit_code}" + return "$exit_code" +} + +# Run main; exit with its return code +main "$@"