Productier/scripts/ops/deploy-prod.sh

#!/usr/bin/env bash
set -euo pipefail

ROOT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")/../.." && pwd)"
ENV_FILE="${1:-$ROOT_DIR/.env.production}"
COMPOSE_FILE="${2:-$ROOT_DIR/infra/docker-compose.prod.yml}"
HEALTH_TIMEOUT_SECONDS="${DEPLOY_HEALTH_TIMEOUT_SECONDS:-240}"
HEALTH_POLL_SECONDS="${DEPLOY_HEALTH_POLL_SECONDS:-2}"
DEPLOY_BUILD="${DEPLOY_BUILD:-1}"
DEPLOY_PULL="${DEPLOY_PULL:-1}"
DEPLOY_RUN_SMOKE="${DEPLOY_RUN_SMOKE:-1}"
DEPLOY_REMOVE_ORPHANS="${DEPLOY_REMOVE_ORPHANS:-1}"
DEPLOY_PRINT_LOGS_ON_FAILURE="${DEPLOY_PRINT_LOGS_ON_FAILURE:-1}"
DEPLOY_LOG_TAIL_LINES="${DEPLOY_LOG_TAIL_LINES:-200}"

if [[ "${1:-}" == "--help" || "${1:-}" == "-h" ]]; then
  cat <<'USAGE'
Usage:
  scripts/ops/deploy-prod.sh [env-file] [compose-file]

Examples:
  scripts/ops/deploy-prod.sh
  scripts/ops/deploy-prod.sh .env.production infra/docker-compose.prod.yml

Behavior:
  - runs production preflight checks
  - optionally pulls external images
  - performs compose deployment (with optional build)
  - waits for core services to become healthy
  - runs post-deploy smoke checks

Optional env toggles:
  DEPLOY_PULL=1|0
  DEPLOY_BUILD=1|0
  DEPLOY_RUN_SMOKE=1|0
  DEPLOY_REMOVE_ORPHANS=1|0
  DEPLOY_HEALTH_TIMEOUT_SECONDS=240
  DEPLOY_HEALTH_POLL_SECONDS=2
  DEPLOY_PRINT_LOGS_ON_FAILURE=1|0
  DEPLOY_LOG_TAIL_LINES=200
USAGE
  exit 0
fi

is_true() {
  local raw="${1:-}"
  case "${raw,,}" in
    1|true|yes|on) return 0 ;;
    *) return 1 ;;
  esac
}

print_failure_logs() {
  if ! is_true "$DEPLOY_PRINT_LOGS_ON_FAILURE"; then
    return
  fi
  if ! command -v docker >/dev/null 2>&1; then
    return
  fi
  echo "[info] last ${DEPLOY_LOG_TAIL_LINES} log lines from compose services:"
  docker compose --env-file "$ENV_FILE" -f "$COMPOSE_FILE" logs --tail "$DEPLOY_LOG_TAIL_LINES" || true
}

on_error() {
  local exit_code="$1"
  echo "[error] deployment failed (exit code ${exit_code})" >&2
  print_failure_logs
}

trap 'on_error $?' ERR

if ! [[ "$HEALTH_TIMEOUT_SECONDS" =~ ^[0-9]+$ ]] || [[ "$HEALTH_TIMEOUT_SECONDS" -lt 5 ]]; then
  echo "[error] DEPLOY_HEALTH_TIMEOUT_SECONDS must be an integer >= 5" >&2
  exit 1
fi

if ! [[ "$HEALTH_POLL_SECONDS" =~ ^[0-9]+$ ]] || [[ "$HEALTH_POLL_SECONDS" -lt 1 ]]; then
  echo "[error] DEPLOY_HEALTH_POLL_SECONDS must be an integer >= 1" >&2
  exit 1
fi

bash "$ROOT_DIR/scripts/ops/preflight-prod.sh" "$ENV_FILE" "$COMPOSE_FILE"

if is_true "$DEPLOY_PULL"; then
  echo "[step] pulling external base images..."
  docker compose --env-file "$ENV_FILE" -f "$COMPOSE_FILE" pull gateway postgres rustfs rustfs-init || \
    echo "[warn] one or more image pulls failed; continuing with local cache"
fi

echo "[step] deploying compose stack..."
up_args=(--env-file "$ENV_FILE" -f "$COMPOSE_FILE" up -d)
if is_true "$DEPLOY_BUILD"; then
  up_args+=(--build)
fi
if is_true "$DEPLOY_REMOVE_ORPHANS"; then
  up_args+=(--remove-orphans)
fi

docker compose "${up_args[@]}"

wait_for_service() {
  local service="$1"
  local deadline=$((SECONDS + HEALTH_TIMEOUT_SECONDS))

  while (( SECONDS < deadline )); do
    local container_id
    container_id="$(docker compose --env-file "$ENV_FILE" -f "$COMPOSE_FILE" ps -q "$service" || true)"

    if [[ -z "$container_id" ]]; then
      sleep "$HEALTH_POLL_SECONDS"
      continue
    fi

    local status
    status="$(docker inspect -f '{{if .State.Health}}{{.State.Health.Status}}{{else}}{{.State.Status}}{{end}}' "$container_id" 2>/dev/null || true)"

    case "$status" in
      healthy|running)
        echo "[ok] service healthy: $service"
        return 0
        ;;
      unhealthy|dead|exited)
        echo "[error] service entered terminal unhealthy state: $service ($status)" >&2
        return 1
        ;;
      *)
        sleep "$HEALTH_POLL_SECONDS"
        ;;
    esac
  done

  echo "[error] timed out waiting for service health: $service" >&2
  return 1
}

echo "[step] waiting for service health checks..."
runtime_services=(postgres rustfs auth api frontend gateway)
for service in "${runtime_services[@]}"; do
  wait_for_service "$service"
done

if is_true "$DEPLOY_RUN_SMOKE"; then
  echo "[step] running smoke checks..."
  bash "$ROOT_DIR/scripts/ops/smoke-prod.sh" "$ENV_FILE"
fi

echo "[ok] production deployment completed successfully"