Files
baiyue-firstleg/deploy/scripts/deploy-blue-green-service.sh
T

391 lines
14 KiB
Bash
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
#!/usr/bin/env sh
set -eu
if [ "$#" -ne 7 ]; then
echo "用法: $0 <环境> <镜像仓库> <镜像标签> <服务> <环境文件> <Compose文件> <项目名>" >&2
exit 2
fi
DEPLOY_ENV="$1"
IMAGE_REPOSITORY="$2"
IMAGE_TAG="$3"
SERVICE="$4"
ENV_FILE="$5"
COMPOSE_FILE="$6"
BASE_PROJECT="$7"
case "$DEPLOY_ENV" in
production) ;;
*) echo "蓝绿执行器仅用于 production,当前: $DEPLOY_ENV" >&2; exit 2 ;;
esac
for file in "$ENV_FILE" "$COMPOSE_FILE"; do
[ -f "$file" ] || { echo "必需文件不存在: $file" >&2; exit 2; }
done
case "$SERVICE" in
*[!A-Za-z0-9_-]*|'') echo "非法服务名: $SERVICE" >&2; exit 2 ;;
esac
env_value() {
sed -n "s/^$1=//p" "$ENV_FILE" | tail -n 1 | tr -d '\r'
}
service_contract() {
case "$SERVICE:$COMPOSE_FILE" in
gateway:*) echo "gateway|9000|9000|gateway|gateway-active" ;;
auth:*) echo "auth|9100|9100|auth-service|auth" ;;
iam:*) echo "iam|9201|9201|iam-service|iam" ;;
firstleg:*) echo "business|9202|9202|firstleg-service|firstleg" ;;
finance:*) echo "finance|9203|9203|finance-service|finance" ;;
order:*) echo "order|9204|9204|order-service|order" ;;
warehouse:*) echo "warehouse|9205|9205|warehouse-service|warehouse" ;;
file:*) echo "file|9206|9206|file-service|file" ;;
print:*) echo "print|9207|9207|print-service|print" ;;
integration-service:*) echo "integration|9210|9210|integration-service|integration-service" ;;
audit-service:*) echo "audit|9700|9700|audit-service|audit-service" ;;
gateway-overseas:*) echo "gateway-overseas|9300|9300|gateway-overseas|gateway-overseas-active" ;;
lastleg-service:*) echo "lastleg|9303|9303|lastleg-service|lastleg-service" ;;
warehouse-overseas-service:*) echo "warehouse-overseas|9305|9305|warehouse-overseas-service|warehouse-overseas-service" ;;
event-bridge-service:*overseas*) echo "event-bridge-overseas|9312|9313|event-bridge-overseas|event-bridge-overseas" ;;
event-bridge-service:*) echo "event-bridge|9312|9312|event-bridge-cn|event-bridge-cn" ;;
*) echo "不支持蓝绿部署的服务: $SERVICE" >&2; exit 2 ;;
esac
}
IFS='|' read -r CONTAINER_STEM INTERNAL_PORT BASE_PORT NACOS_SERVICE ROUTING_ALIAS <<EOF
$(service_contract)
EOF
STATE_ROOT="${BLUE_GREEN_STATE_ROOT:-/data/firstleg/deploy/blue-green-state}"
STATE_KEY="${BASE_PROJECT}-${SERVICE}"
STATE_FILE="$STATE_ROOT/$STATE_KEY.active"
ROUTING_NETWORK="${BLUE_GREEN_ROUTING_NETWORK:-firstleg-${DEPLOY_ENV}-routing}"
HEALTH_TIMEOUT="${DEPLOY_HEALTH_TIMEOUT_SECONDS:-240}"
HEALTH_POLL="${DEPLOY_HEALTH_POLL_SECONDS:-5}"
STABILIZATION_SECONDS="${BLUE_GREEN_STABILIZATION_SECONDS:-10}"
STOP_TIMEOUT="${BLUE_GREEN_STOP_TIMEOUT_SECONDS:-45}"
IMAGE="$IMAGE_REPOSITORY/$SERVICE:$IMAGE_TAG"
for value in "$HEALTH_TIMEOUT" "$HEALTH_POLL" "$STABILIZATION_SECONDS" "$STOP_TIMEOUT"; do
case "$value" in ''|*[!0-9]*) echo "蓝绿时间参数必须为非负整数: $value" >&2; exit 2 ;; esac
done
mkdir -p "$STATE_ROOT"
chmod 750 "$STATE_ROOT"
docker network inspect "$ROUTING_NETWORK" >/dev/null 2>&1 || docker network create "$ROUTING_NETWORK" >/dev/null
legacy_container="firstleg-${DEPLOY_ENV}-${CONTAINER_STEM}"
active_slot=""
active_container=""
active_port=""
active_worker=""
if [ -s "$STATE_FILE" ]; then
IFS='|' read -r active_slot active_container active_port active_image active_worker < "$STATE_FILE"
elif docker inspect "$legacy_container" >/dev/null 2>&1; then
active_slot="legacy"
active_container="$legacy_container"
active_port="$BASE_PORT"
fi
if [ "$active_slot" = "blue" ]; then
target_slot="green"
else
target_slot="blue"
[ "$active_slot" = "legacy" ] && target_slot="green"
fi
if [ "$target_slot" = "blue" ]; then
target_port="$BASE_PORT"
else
target_port=$((BASE_PORT + 10000))
fi
candidate="firstleg-${DEPLOY_ENV}-${CONTAINER_STEM}-${target_slot}"
worker_candidate="${candidate}-worker"
if docker inspect "$candidate" >/dev/null 2>&1; then
if [ "$candidate" = "$active_container" ]; then
echo "活动槽状态异常,候选容器与活动容器相同: $candidate" >&2
exit 1
fi
docker rm -f "$candidate" >/dev/null
fi
compose() {
docker compose --project-name "$BASE_PROJECT" --env-file "$ENV_FILE" -f "$COMPOSE_FILE" "$@"
}
container_health() {
container="$1"
docker inspect --format '{{if .State.Health}}{{.State.Health.Status}}{{else}}{{.State.Status}}{{end}}' \
"$container" 2>/dev/null || echo missing
}
cleanup_candidate() {
docker rm -f "$candidate" >/dev/null 2>&1 || true
}
wait_for_container() {
container="$1"
label="$2"
elapsed=0
while [ "$elapsed" -lt "$HEALTH_TIMEOUT" ]; do
state="$(container_health "$container")"
echo "$label 健康状态: $state(已等待 ${elapsed}s)"
case "$state" in
healthy) return 0 ;;
unhealthy|exited|dead|missing) return 1 ;;
esac
sleep "$HEALTH_POLL"
elapsed=$((elapsed + HEALTH_POLL))
done
return 1
}
needs_worker() {
case "$SERVICE" in
iam|firstleg|finance|order|event-bridge-service|lastleg-service) return 0 ;;
file)
[ "$(env_value FILE_EXPORT_CONSUMER_ENABLED)" = true ] \
|| [ "$(env_value FILE_EXPORT_PUBLISHER_ENABLED)" = true ] \
|| [ "$(env_value FILE_EXPORT_RECONCILER_ENABLED)" = true ] \
|| [ "$(env_value AUDIT_OUTBOX_PUBLISHER_ENABLED)" = true ]
;;
audit-service)
[ "$(env_value AUDIT_MQ_CONSUMER_ENABLED)" = true ] \
|| [ "$(env_value AUDIT_DLQ_CONSUMER_ENABLED)" = true ] \
|| [ "$(env_value AUDIT_DLQ_COMPENSATION_ENABLED)" = true ]
;;
auth|warehouse|warehouse-overseas-service)
[ "$(env_value AUDIT_OUTBOX_PUBLISHER_ENABLED)" = true ]
;;
*) return 1 ;;
esac
}
start_worker_candidate() {
docker rm -f "$worker_candidate" >/dev/null 2>&1 || true
echo "启动单例 Worker: $worker_candidate"
if ! compose run -d --no-deps --name "$worker_candidate" \
-e "NACOS_REGISTER_ENABLED=false" \
-e "BAIYUE_RUNTIME_SCHEDULING_ENABLED=true" \
-e "BAIYUE_DEPLOY_SLOT=${target_slot}-worker" \
-e "BAIYUE_RELEASE_ID=$IMAGE_TAG" "$SERVICE" >/dev/null; then
return 1
fi
docker network connect "$ROUTING_NETWORK" "$worker_candidate"
if ! wait_for_container "$worker_candidate" "候选 Worker"; then
docker logs --tail 200 "$worker_candidate" 2>&1 || true
docker rm -f "$worker_candidate" >/dev/null 2>&1 || true
return 1
fi
worker_image="$(docker inspect --format '{{.Config.Image}}' "$worker_candidate")"
[ "$worker_image" = "$IMAGE" ] || return 1
}
connect_routing_aliases() {
container="$1"
shift
docker network inspect "$ROUTING_NETWORK" --format '{{json .Containers}}' | grep -q "\"$container\"" \
&& return 0
set -- docker network connect $(for alias_name in "$@"; do printf '%s\n' --alias "$alias_name"; done)
"$@" "$ROUTING_NETWORK" "$container"
}
bootstrap_legacy_routes() {
for pair in \
gateway-active,gateway:gateway auth,auth-service:auth iam,iam-service:iam \
firstleg,firstleg-service:business finance,finance-service:finance order,order-service:order \
warehouse,warehouse-service:warehouse file,file-service:file print,print-service:print \
integration-service:integration audit-service:audit \
gateway-overseas-active,gateway-overseas:gateway-overseas \
lastleg-service:lastleg warehouse-overseas-service:warehouse-overseas \
event-bridge-cn:event-bridge event-bridge-overseas:event-bridge-overseas
do
aliases="${pair%%:*}"
stem="${pair#*:}"
container="firstleg-${DEPLOY_ENV}-${stem}"
if docker inspect "$container" >/dev/null 2>&1; then
connect_routing_aliases "$container" $(printf '%s' "$aliases" | tr ',' ' ') || true
fi
done
for container in firstleg-prod-admin-web firstleg-prod-client-web; do
if docker inspect "$container" >/dev/null 2>&1; then
connect_routing_aliases "$container" "${container##*-}" || true
fi
done
}
switch_gateway_proxy() {
[ "$SERVICE" = "gateway" ] || [ "$SERVICE" = "gateway-overseas" ] || return 0
if [ "$SERVICE" = "gateway" ]; then required_alias=gateway-active; else required_alias=gateway-overseas-active; fi
for frontend in firstleg-prod-admin-web-blue firstleg-prod-admin-web-green \
firstleg-prod-client-web-blue firstleg-prod-client-web-green; do
if docker inspect "$frontend" >/dev/null 2>&1 \
&& ! docker exec "$frontend" nginx -T 2>&1 | grep -q "$required_alias"; then
echo "$frontend 尚未使用稳定上游 $required_alias,拒绝切换 Gateway" >&2
return 1
fi
done
if [ "$SERVICE" = "gateway" ]; then
config="$(env_value GATEWAY_BLUE_GREEN_NGINX_CONFIG)"
config="${config:-/www/server/panel/vhost/nginx/api-cn.buggoglobal.com.conf}"
else
config="$(env_value GATEWAY_OVERSEAS_BLUE_GREEN_NGINX_CONFIG)"
config="${config:-/www/server/panel/vhost/nginx/api-global.buggoglobal.com.conf}"
fi
if [ -z "$config" ] || [ ! -f "$config" ]; then
echo "$SERVICE 蓝绿切换要求环境文件配置有效的 Nginx 配置路径" >&2
return 1
fi
backup="$config.blue-green-backup"
cp -f "$config" "$backup"
if ! sed -i "s#127\.0\.0\.1:${active_port:-$BASE_PORT}#127.0.0.1:$target_port#g" "$config" \
|| cmp -s "$config" "$backup"; then
cp -f "$backup" "$config"
echo "未能在 Nginx 配置中切换 $SERVICE 上游端口" >&2
return 1
fi
if ! nginx -t || ! nginx -s reload; then
cp -f "$backup" "$config"
nginx -t && nginx -s reload || true
return 1
fi
rm -f "$backup"
}
reload_frontend_proxies() {
[ "$SERVICE" = "gateway" ] || [ "$SERVICE" = "gateway-overseas" ] || return 0
for frontend in firstleg-prod-admin-web-blue firstleg-prod-admin-web-green \
firstleg-prod-client-web-blue firstleg-prod-client-web-green; do
if docker inspect "$frontend" >/dev/null 2>&1; then
docker exec "$frontend" nginx -t
docker exec "$frontend" nginx -s reload
fi
done
}
rollback_gateway_proxy() {
[ "$SERVICE" = "gateway" ] || [ "$SERVICE" = "gateway-overseas" ] || return 0
if [ "$SERVICE" = "gateway" ]; then
config="$(env_value GATEWAY_BLUE_GREEN_NGINX_CONFIG)"
config="${config:-/www/server/panel/vhost/nginx/api-cn.buggoglobal.com.conf}"
else
config="$(env_value GATEWAY_OVERSEAS_BLUE_GREEN_NGINX_CONFIG)"
config="${config:-/www/server/panel/vhost/nginx/api-global.buggoglobal.com.conf}"
fi
old_port="${active_port:-$BASE_PORT}"
sed -i "s#127\.0\.0\.1:$target_port#127.0.0.1:$old_port#g" "$config"
nginx -t && nginx -s reload
}
bootstrap_legacy_routes
export DEPLOY_ENV IMAGE_REPOSITORY IMAGE_TAG
echo "启动候选槽: service=$SERVICE slot=$target_slot port=$target_port image=$IMAGE"
if ! compose run -d --no-deps --name "$candidate" \
-e "BAIYUE_DEPLOY_SLOT=$target_slot" -e "BAIYUE_RELEASE_ID=$IMAGE_TAG" \
-e "BAIYUE_RUNTIME_SCHEDULING_ENABLED=false" \
-e "AUDIT_OUTBOX_PUBLISHER_ENABLED=false" \
-e "NOTIFICATION_SCHEDULER_ENABLED=false" \
-e "NOTIFICATION_MQ_CONSUMER_ENABLED=false" \
-e "FILE_EXPORT_CONSUMER_ENABLED=false" \
-e "FILE_EXPORT_PUBLISHER_ENABLED=false" \
-e "FILE_EXPORT_RECONCILER_ENABLED=false" \
-e "AUDIT_MQ_CONSUMER_ENABLED=false" \
-e "AUDIT_DLQ_CONSUMER_ENABLED=false" \
-e "AUDIT_DLQ_COMPENSATION_ENABLED=false" \
-p "${PUBLISHED_BIND_ADDRESS:-0.0.0.0}:$target_port:$INTERNAL_PORT" "$SERVICE" >/dev/null; then
cleanup_candidate
exit 1
fi
if ! wait_for_container "$candidate" "候选 Web 槽 $target_slot"; then
echo "候选槽未通过健康检查,保留旧实例" >&2
docker logs --tail 200 "$candidate" 2>&1 || true
cleanup_candidate
exit 1
fi
running_image="$(docker inspect --format '{{.Config.Image}}' "$candidate")"
if [ "$running_image" != "$IMAGE" ]; then
echo "候选镜像不一致: expected=$IMAGE actual=$running_image" >&2
cleanup_candidate
exit 1
fi
if [ "$NACOS_SERVICE" = "$ROUTING_ALIAS" ]; then
connect_routing_aliases "$candidate" "$ROUTING_ALIAS"
else
connect_routing_aliases "$candidate" "$ROUTING_ALIAS" "$NACOS_SERVICE"
fi
if [ "$STABILIZATION_SECONDS" -gt 0 ]; then
echo "候选槽稳定观察 ${STABILIZATION_SECONDS}s"
sleep "$STABILIZATION_SECONDS"
[ "$(container_health "$candidate")" = "healthy" ] || { cleanup_candidate; exit 1; }
fi
if ! switch_gateway_proxy; then
cleanup_candidate
exit 1
fi
worker_required=false
if needs_worker; then worker_required=true; fi
if [ "$worker_required" = true ] && [ "$active_slot" != legacy ]; then
if [ -n "$active_worker" ] && docker inspect "$active_worker" >/dev/null 2>&1; then
echo "暂停旧 Worker,避免新旧任务并行: $active_worker"
docker stop --time "$STOP_TIMEOUT" "$active_worker" >/dev/null
fi
if ! start_worker_candidate; then
[ -z "$active_worker" ] || docker start "$active_worker" >/dev/null 2>&1 || true
rollback_gateway_proxy || true
cleanup_candidate
exit 1
fi
fi
old_container_stopped=false
if [ -n "$active_container" ] && docker inspect "$active_container" >/dev/null 2>&1; then
echo "优雅停止旧 Web 实例: $active_container"
docker stop --time "$STOP_TIMEOUT" "$active_container" >/dev/null
old_container_stopped=true
fi
if [ "$worker_required" = true ] && [ "$active_slot" = legacy ]; then
if ! start_worker_candidate; then
[ "$old_container_stopped" = false ] || docker start "$active_container" >/dev/null 2>&1 || true
rollback_gateway_proxy || true
cleanup_candidate
exit 1
fi
fi
if ! reload_frontend_proxies; then
[ "$old_container_stopped" = false ] || docker start "$active_container" >/dev/null 2>&1 || true
if [ "$worker_required" = true ]; then
docker rm -f "$worker_candidate" >/dev/null 2>&1 || true
[ -z "$active_worker" ] || docker start "$active_worker" >/dev/null 2>&1 || true
fi
rollback_gateway_proxy || true
cleanup_candidate
reload_frontend_proxies || true
exit 1
fi
[ -z "$active_worker" ] || [ "$active_worker" = "$worker_candidate" ] \
|| docker rm "$active_worker" >/dev/null 2>&1 || true
[ "$old_container_stopped" = false ] || docker rm "$active_container" >/dev/null
if [ "$worker_required" = true ]; then
next_worker="$worker_candidate"
else
next_worker=""
fi
printf '%s|%s|%s|%s|%s\n' "$target_slot" "$candidate" "$target_port" "$IMAGE" "$next_worker" > "$STATE_FILE.tmp"
chmod 640 "$STATE_FILE.tmp"
mv -f "$STATE_FILE.tmp" "$STATE_FILE"
echo "蓝绿切换完成: service=$SERVICE active=$target_slot container=$candidate port=$target_port"