romm/docker/init_scripts/init

#!/usr/bin/env bash

set -o errexit           # treat errors as fatal
set -o nounset           # treat unset variables as an error
set -o pipefail          # treat errors in pipes as fatal
shopt -s inherit_errexit # inherit errexit

LOGLEVEL="${LOGLEVEL:="info"}"

# make it possible to disable the inotify watcher process
ENABLE_RESCAN_ON_FILESYSTEM_CHANGE="${ENABLE_RESCAN_ON_FILESYSTEM_CHANGE:="false"}"
ENABLE_SCHEDULED_RESCAN="${ENABLE_SCHEDULED_RESCAN:="false"}"
ENABLE_SCHEDULED_UPDATE_SWITCH_TITLEDB="${ENABLE_SCHEDULED_UPDATE_SWITCH_TITLEDB:="false"}"

# if REDIS_HOST is set, we assume that an external redis is used
REDIS_HOST="${REDIS_HOST:=""}"

# set DEFAULT_WEB_CONCURRENCY to 1 if not set by docker env to reduce resource usage
# (since backend is almost 100% async this won't block anything)
DEFAULT_WEB_CONCURRENCY=1

# logger colors
RED='\033[0;31m'
LIGHTMAGENTA='\033[0;95m'
GREEN='\033[0;32m'
YELLOW='\033[0;33m'
BLUE='\033[0;34m'
CYAN='\033[0;36m'
RESET='\033[0;00m'

# print debug log output if enabled
debug_log() {
	if [[ ${LOGLEVEL} == "debug" ]]; then
		echo -e "${LIGHTMAGENTA}DEBUG:    ${BLUE}[RomM]${LIGHTMAGENTA}[init]${CYAN}[$(date +"%Y-%m-%d %T")]${RESET}" "${@}" || true
	fi
}

info_log() {
	echo -e "${GREEN}INFO:     ${BLUE}[RomM]${LIGHTMAGENTA}[init]${CYAN}[$(date +"%Y-%m-%d %T")]${RESET}" "${@}" || true
}

warn_log() {
	echo -e "${YELLOW}WARNING:  ${BLUE}[RomM]${LIGHTMAGENTA}[init]${CYAN}[$(date +"%Y-%m-%d %T")]${RESET}" "${@}" || true
}

error_log() {
	echo -e "${RED}ERROR:    ${BLUE}[RomM]${LIGHTMAGENTA}[init]${CYAN}[$(date +"%Y-%m-%d %T")]${RESET}" "${@}" || true
	exit 1
}

wait_for_gunicorn_socket() {
	debug_log "Waiting for gunicorn socket file..."
	local retries=60
	while [[ ! -S /tmp/gunicorn.sock && retries -gt 0 ]]; do
		sleep 0.5
		((retries--))
	done

	if [[ -S /tmp/gunicorn.sock ]]; then
		debug_log "Gunicorn socket file found"
	else
		error_log "Gunicorn socket file not found after waiting 30s"
	fi
}

# function that runs or main process and creates a corresponding PID file,
start_bin_gunicorn() {
	# cleanup potentially leftover socket
	rm /tmp/gunicorn.sock -f

	# commands to start our main application and store its PID to check for crashes
	info_log "Starting backend"

	# TODO: Remove support for GUNICORN_WORKERS in future version.
	if [[ -n ${GUNICORN_WORKERS-} ]]; then
		warn_log "GUNICORN_WORKERS variable is deprecated, use WEB_CONCURRENCY instead"
		: "${WEB_CONCURRENCY:=${GUNICORN_WORKERS}}"
	fi

	gunicorn \
		--error-logfile /tmp/gunicorn_access.log \
		--error-logfile /tmp/gunicorn_error.log \
		--worker-class uvicorn.workers.UvicornWorker \
		--bind=0.0.0.0:5000 \
		--bind=unix:/tmp/gunicorn.sock \
		--pid=/tmp/gunicorn.pid \
		--forwarded-allow-ips="*" \
		--workers "${WEB_CONCURRENCY:-${DEFAULT_WEB_CONCURRENCY:-1}}" \
		main:app &
}

# Commands to start nginx (handling PID creation internally)
start_bin_nginx() {
	wait_for_gunicorn_socket

	info_log "Starting nginx"
	if [[ ${EUID} -ne 0 ]]; then
		nginx
	else
		# if container runs as root, drop permissions
		nginx -g 'user romm;'
	fi
	info_log "🚀 RomM is now available at http://[::1]:8080"
}

# Commands to start valkey-server (handling PID creation internally)
start_bin_valkey-server() {
	info_log "Starting internal valkey"

	if [[ -f /usr/local/etc/valkey/valkey.conf ]]; then
		if [[ ${LOGLEVEL} == "debug" ]]; then
			valkey-server /usr/local/etc/valkey/valkey.conf &
		else
			valkey-server /usr/local/etc/valkey/valkey.conf >/dev/null 2>&1 &
		fi
	else
		if [[ ${LOGLEVEL} == "debug" ]]; then
			valkey-server --dir /redis-data &
		else
			valkey-server --dir /redis-data >/dev/null 2>&1 &
		fi
	fi

	VALKEY_PID=$!
	echo "${VALKEY_PID}" >/tmp/valkey-server.pid

	local host="127.0.0.1"
	local port="6379"
	local max_retries=120
	local retry=0

	debug_log "Waiting for internal valkey to be ready..."

	# Temporarily disable errexit for this part of the script
	set +o errexit

	while ((retry < max_retries)); do
		# Attempt to check if valkey TCP port is open
		if (echo >/dev/tcp/"${host}"/"${port}") 2>/dev/null; then
			debug_log "Internal valkey is ready and accepting connections"
			set -o errexit # Re-enable errexit after success
			return 0
		fi

		sleep 0.5
		((retry++))
	done

	error_log "Internal valkey did not become ready after $((max_retries * 500))ms"
}

# function that runs our independent python scripts and creates corresponding PID files,
start_python() {
	SCRIPT="${1}"
	info_log "Starting ${SCRIPT}"
	python3 "${SCRIPT}.py" &
	WATCHER_PID=$!
	echo "${WATCHER_PID}" >"/tmp/${SCRIPT}.pid"
}

watchdog_process_pid() {
	TYPE=$1
	PROCESS=$2
	if [[ -f "/tmp/${PROCESS}.pid" ]]; then
		# check if the pid we last wrote to our state file is actually active
		PID=$(cat "/tmp/${PROCESS}.pid") || true
		if [[ ! -d "/proc/${PID}" ]]; then
			if [[ ${TYPE} == "bin" ]]; then
				start_bin_"${PROCESS}"
			elif [[ ${TYPE} == "python" ]]; then
				start_python "${PROCESS}"
			fi
		fi
	else
		# start process if we dont have a corresponding PID file
		if [[ ${TYPE} == "bin" ]]; then
			start_bin_"${PROCESS}"
		elif [[ ${TYPE} == "python" ]]; then
			start_python "${PROCESS}"
		fi
	fi
}

stop_process_pid() {
	PROCESS=$1
	if [[ -f "/tmp/${PROCESS}.pid" ]]; then
		PID=$(cat "/tmp/${PROCESS}.pid") || true
		if [[ -d "/proc/${PID}" ]]; then
			info_log "Stopping ${PROCESS}"
			kill "${PID}" || true
			# wait for process exit
			while [[ -e "/proc/${PID}" ]]; do sleep 0.1; done
		fi
	fi
}

shutdown() {
	# shutdown in reverse order
	stop_process_pid scheduler
	stop_process_pid worker
	stop_process_pid watcher
	stop_process_pid nginx
	stop_process_pid gunicorn
	stop_process_pid valkey-server
}

# switch to backend directory
cd /backend || { error_log "/backend directory doesn't seem to exist"; }

# setup trap handler
exited=0
trap 'exited=1 && shutdown' SIGINT SIGTERM EXIT

# clear any leftover PID files
rm /tmp/*.pid -f

# Start Valkey server if REDIS_HOST is not set (which would mean user is using an external Redis/Valkey)
if [[ -z ${REDIS_HOST} ]]; then
	watchdog_process_pid bin valkey-server
else
	warn_log "REDIS_HOST is set, not starting internal valkey-server"
fi

# Run needed database migrations once at startup
info_log "Running database migrations"
if alembic upgrade head; then
	info_log "Database migrations succeeded"
else
	error_log "Failed to run database migrations"
fi

# main loop
while ! ((exited)); do
	watchdog_process_pid bin gunicorn

	watchdog_process_pid python worker

	# only start the scheduler if enabled
	if [[ ${ENABLE_SCHEDULED_RESCAN} == "true" || ${ENABLE_SCHEDULED_UPDATE_SWITCH_TITLEDB} == "true" ]]; then
		watchdog_process_pid python scheduler
	fi

	# only start the watcher if enabled
	if [[ ${ENABLE_RESCAN_ON_FILESYSTEM_CHANGE} == "true" ]]; then
		watchdog_process_pid python watcher
	fi

	watchdog_process_pid bin nginx

	# check for died processes every 5 seconds
	sleep 5
done