import logging import requests import time import docker import asyncio import sys import os import rov_server_config import utils from concurrent.futures import ThreadPoolExecutor logging.basicConfig(level=logging.INFO, format='%(asctime)s %(filename)s[line:%(lineno)d] %(levelname)s %(message)s', datefmt='%a, %d %b %Y %H:%M:%S') health_instances = [] ess_instances = [] def server_health_check(client, instance_id): """ 服务健康检查 :param client: 客户端连接 :param instance_id: instanceId :return: """ global health_instances ip_address = utils.get_ip_address(client=client, instance_id=instance_id) while True: health_check_url = f"http://{ip_address}:5001/healthcheck" try: http_code = requests.get(health_check_url).status_code except: logging.info("images is downloading") http_code = 0 if http_code == 200: health_instances.append((instance_id, ip_address)) break else: time.sleep(10) def restart(slb_client, instance_id, image_name): """重启容器并探活""" # 1. 获取ip ip_address = utils.get_ip_address(client=slb_client, instance_id=instance_id) logging.info(f"ip address：{instance_id}/{ip_address}") # 3. 重启容器 docker_client = docker.DockerClient(base_url=f'tcp://{ip_address}:2375', timeout=60) image_id = docker_client.containers.get(image_name) image_id.restart() logging.info(f"docker restart finished, ip address：{instance_id}/{ip_address}") time.sleep(5) # 4. 探活 server_health_check(slb_client, instance_id) logging.info(f"health check finished, ip address：{instance_id}/{ip_address}") async def server_restart_all(slb_client, instance_id_list, image_name, max_workers): try: # 1. 摘流量 utils.set_instance_weight_process(client=slb_client, slb_id=rov_server_config.slb_id, instance_id_list=instance_id_list, weight_list=[(0, 60)]) logging.info(f"set weight with 0 finished.") # 2. 异步重启 & 探活 global health_instances loop = asyncio.get_running_loop() executor = ThreadPoolExecutor(max_workers=max_workers) tasks = [ loop.run_in_executor(executor, restart, *args) for args in [(slb_client, instance_id, image_name) for instance_id in instance_id_list] ] await asyncio.wait(tasks) logging.info(f"health instances count: {len(health_instances)}, {health_instances}") # 3. 挂载流量 if len(health_instances) == len(instance_id_list): time.sleep(30) add_weight_list = [(10, 30), (20, 20), (40, 10), (60, 10), (80, 10), (100, 10)] utils.set_instance_weight_process(client=slb_client, slb_id=rov_server_config.slb_id, instance_id_list=[instance[0] for instance in health_instances], weight_list=add_weight_list) logging.info(f"set weight finished instance count: {len(health_instances)}, {health_instances}") else: logging.info(f"ess count: {len(instance_id_list)}, " f"health count: {len(health_instances)}") sys.exit() except Exception as e: logging.error(f"server restart fail!") logging.error(e) async def ess_instance(create_client, slb_client, ess_count, max_workers, version): """ 扩容机器并运行新服务 :param create_client: 购买机器客户端连接 :param slb_client: 修改负载均衡权限 :param ess_count: 扩容数量 :param max_workers: 线程数 :param version: 版本标记 :return: """ # 1. 购买机器并启动 ess_instance_ids = utils.create_multiple_instances( amount=ess_count, client=create_client, **rov_server_config.instance_config_i, ) time.sleep(60) # 2. 发送启动脚本到机器上 utils.send_file_to_ecs(client=create_client, instance_id_list=ess_instance_ids, **rov_server_config.start_sh) logging.info(f"send start shell file finished, instances: {ess_instance_ids}") # 3. 启动服务 server_start_sh = os.path.join(rov_server_config.start_sh['target_dir'], rov_server_config.start_sh['name']) server_start_commend = f"sh {server_start_sh} {version}" utils.run_command(client=create_client, instance_ids=ess_instance_ids, command=server_start_commend) # 4. 异步探活 global health_instances health_instances = [] loop = asyncio.get_running_loop() executor = ThreadPoolExecutor(max_workers=max_workers) tasks = [ loop.run_in_executor(executor, server_health_check, *args) for args in [(slb_client, instance_id) for instance_id in ess_instance_ids] ] await asyncio.wait(tasks) logging.info(f"health instances count: {len(health_instances)}, {health_instances}") # 5. 挂载流量 if len(health_instances) == len(ess_instance_ids): # 所有机器探活成功 time.sleep(60) utils.add_backend_servers(client=slb_client, slb_id=rov_server_config.slb_id, instances=health_instances) health_instance_ids = [instance_id for instance_id, _ in health_instances] add_weight_list = [(10, 30), (20, 20), (40, 10), (60, 10), (80, 10), (100, 10)] utils.set_instance_weight_process(client=slb_client, slb_id=rov_server_config.slb_id, instance_id_list=health_instance_ids, weight_list=add_weight_list) global ess_instances ess_instances.extend(health_instance_ids) logging.info(f"ess count: {ess_count}, " f"create count: {len(ess_instance_ids)}, " f"finished count: {len(health_instance_ids)}") else: logging.info(f"ess count: {ess_count}, " f"create count: {len(ess_instance_ids)}, " f"health count: {len(health_instances)}") sys.exit() def remove_instances(create_client, slb_client, instance_ids): """ 停止并释放机器 :param create_client: :param slb_client: :param instance_ids: instanceId type-list :return: None """ # 1. 摘流量 utils.set_instance_weight_process(client=slb_client, slb_id=rov_server_config.slb_id, instance_id_list=instance_ids, weight_list=[(0, 60)]) logging.info(f"set weight = 0 finished, instances: {instance_ids}") time.sleep(10) # 2.移除slb utils.remove_backend_servers(client=slb_client, slb_id=rov_server_config.slb_id, instances=instance_ids) # 3. 停止机器 utils.stop_instances(client=create_client, instance_ids=instance_ids) logging.info(f"instances stop finished, instances: {instance_ids}") # 4. 判断机器运行状态是否为Stopped while True: response = utils.get_instances_status(client=create_client, instance_ids=instance_ids) if response.get('Code') is None: instances_list = response.get('InstanceStatuses').get('InstanceStatus') # logging.info(instances_list) stopped_instances = [instance.get('InstanceId') for instance in instances_list if instance.get('Status') == 'Stopped'] if len(stopped_instances) == len(instance_ids): logging.info(f"instances stopped status set success, instances: {stopped_instances}") break else: logging.info(f"stopped instances count = {len(stopped_instances)}, instances: {stopped_instances}") time.sleep(10) else: logging.error(response) sys.exit() # 5. 释放机器 response = utils.release_instances(client=create_client, instance_ids=stopped_instances) if response.get('Code') is None: logging.info(f"release instances finished, instances: {stopped_instances}") else: logging.error(f"release instances fail!!!") sys.exit() def main(): try: slb_client = utils.connect_client(access_key_id=rov_server_config.slb_client_params['access_key_id'], access_key_secret=rov_server_config.slb_client_params['access_key_secret'], region_id=rov_server_config.slb_client_params['region_id']) create_client = utils.connect_client(access_key_id=rov_server_config.create_client_params['access_key_id'], access_key_secret=rov_server_config.create_client_params[ 'access_key_secret'], region_id=rov_server_config.create_client_params['region_id']) # 1. 获取slb下所有机器 online_instance_ids = utils.get_instance_ids(client=slb_client, slb_id=rov_server_config.slb_id) online_instance_count = len(online_instance_ids) logging.info(f"online instance count: {online_instance_count}.") logging.info(f"online instance ids: {online_instance_ids}") # 2. 扩容机器并启动新服务 扩容数量：线上机器数量 logging.info(f"ess instances start ...") ess_instance_count = online_instance_count if ess_instance_count > 100: ess_instance_count = 100 logging.info(f"ess instance count: {ess_instance_count}") asyncio.run(ess_instance(create_client=create_client, slb_client=slb_client, ess_count=ess_instance_count, max_workers=ess_instance_count, version='latest')) logging.info(f"ess instances end!") # 3. 重启线上原有服务器 image_name = 'rov-server' asyncio.run(server_restart_all(slb_client=slb_client, instance_id_list=online_instance_ids, image_name=image_name, max_workers=online_instance_count)) logging.info(f"server restart finished!") # 4. 停止并释放扩容机器 logging.info(f"stop & release instances start ...") remove_instances(create_client=create_client, slb_client=slb_client, instance_ids=ess_instances) logging.info(f"stop & release instances end!") except Exception as e: logging.error(e) if __name__ == '__main__': main()