server-deploy/rov_server_update.py

import asyncio
import sys
import time
import requests
import utils
import logging
import os
import docker
import rov_server_config

from concurrent.futures import ThreadPoolExecutor


health_instances = []
ess_instances = []

# slb_id = 'lb-bp1werfophtsjzfr76njm'
# # 修改负载均衡权限
# slb_client_params = {
#     'access_key_id': 'LTAIuPbTPL3LDDKN',
#     'access_key_secret': 'ORcNedKwWuwVtcq4IRFtUDZgS0b1le',
#     'region_id': 'cn-hangzhou'
# }
# # 购买机器权限
# create_client_params = {
#     'access_key_id': 'LTAI4GBWbFvvXoXsSVBe1o9f',
#     'access_key_secret': 'kRAikWitb4kDxaAyBqNrmLmllMEDO3',
#     'region_id': 'cn-hangzhou'
# }
#
# # 机器配置
# instance_config = {
#     # 使用的镜像信息
#     'image_id': 'm-bp1e5jx8eqhq22l91xw7',
#     # 设置实例规格
#     'instance_type': 'ecs.ic5.xlarge',
#     # 选择的交换机
#     'vswitch_id': 'vsw-bp19lpjwtc6j0p0m9mdc2',
#     # 当前VPC类型的安全组
#     'security_group_id': 'sg-bp1irhrkr4vfj272hk4y',
#     # 硬盘的大小，单位：G
#     'disk_size': '200',
#     # 服务器命名
#     'instance_name': 'ESS-rov-server-[1,2]',
#     # 服务器所在区域
#     'zone_id': 'cn-hangzhou-h',
#     # 磁盘类型：云盘
#     'disk_category': 'cloud_efficiency',
#     # 密钥
#     'key_pair_name': 'stuuudy'
# }
#
# # 服务启动脚本
# start_sh_dir = os.path.dirname(os.path.realpath(__file__))
# start_sh_filename = 'rov_server_start.sh'
# with open(file=os.path.join(start_sh_dir, start_sh_filename), mode='r', encoding='utf-8') as rf:
#     file_content = rf.read()
# start_sh = {
#     'target_dir': '/home/piaoquan_server_sh',
#     'name': start_sh_filename,
#     'content': file_content,
# }


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 set_instance_weight_process(client, instance_id_list, weight_list):
    """
    修改服务器的权重值
    :param client: slb客户端连接
    :param instance_id_list: instance id list
    :param weight_list: 权重修改列表 type-list [(weight, sleep_time), ...]
    :return:
    """
    for weight, sleep_time in weight_list:
        flag = True
        while flag:
            try:
                utils.set_weight_for_instances(client=client,
                                               slb_id=rov_server_config.slb_id,
                                               instance_id_list=instance_id_list,
                                               weight=weight)
                time.sleep(sleep_time)
                flag = False
            except Exception as e:
                time.sleep(10)
                continue


async def run_server(create_client, slb_client, instance_ids, max_workers):
    # 1. 发送启动脚本到机器上
    utils.send_file_to_ecs(client=create_client, instance_id_list=instance_ids, **rov_server_config.start_sh)
    logging.info(f"send start shell file finished, instances: {instance_ids}")
    # 2. 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}"
    utils.run_command(client=create_client, instance_ids=instance_ids, command=server_start_commend)
    # 3. 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 instance_ids]
    ]
    await asyncio.wait(tasks)
    logging.info(f"health instances count: {len(health_instances)}, {health_instances}")
    # 4. 挂载流量
    if len(health_instances) == len(instance_ids):
        # 所有机器探活成功
        time.sleep(60)
        add_weight_list = [(10, 30), (20, 20), (40, 10), (60, 10), (80, 10), (100, 10)]
        set_instance_weight_process(client=slb_client, instance_id_list=instance_ids, weight_list=add_weight_list)
        logging.info(f"add weight instances count: {len(health_instances)}")
    else:
        logging.info(f"instances count: {len(instance_ids)},\nhealth instances count: {len(health_instances)}")
        sys.exit()


async def ess_instance(create_client, slb_client, ess_count, max_workers):
    """
    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
    :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,
    )
    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}"
    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)
    #     add_weight_list = [(10, 30), (20, 20), (40, 10), (60, 10), (80, 10), (100, 10)]
    #     set_instance_weight_process(client=slb_client, instance_id_list=ess_instance_ids, weight_list=add_weight_list)
    #     global ess_instances
    #     ess_instances.extend(ess_instance_ids)
    #     logging.info(f"ess count: {ess_count}, "
    #                  f"create count: {len(ess_instance_ids)}, "
    #                  f"finished count: {len(health_instances)}")
    # 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_container_image(client, instance_id, container_name):
    """
    移除旧容器并删除旧镜像
    :param client: 客户端连接
    :param instance_id: instanceId type-string
    :param container_name: 容器名称 type-string
    :return:
    """
    ip_address = utils.get_ip_address(client=client, instance_id=instance_id)
    logging.info(f"服务器信息：{instance_id}/{ip_address}")
    client = docker.DockerClient(base_url=f'tcp://{ip_address}:2375', timeout=60)
    # 移除旧的容器
    container_remove_retry = 3
    i = 0
    while i < container_remove_retry:
        try:
            container_id = client.containers.get(container_name)
            container_id.remove(force=True)
            break
        except Exception as e:
            i += 1
            print("容器不存在或者无法删除当前容器")
    # 删除旧镜像
    images_remove_retry = 3
    j = 0
    while j < images_remove_retry:
        try:
            images = client.images.list()
            for image in images:
                client.images.remove(force=True, image=image.tags[0])
                time.sleep(2)
        except Exception as e:
            i += 1
            print("镜像不存在，无法获取到镜像ID")


async def update_instance(create_client, slb_client, instance_ids, max_workers):
    """
    线上机器更新
    :param create_client:
    :param slb_client: slb客户端连接
    :param instance_ids: instanceId type-list
    :param max_workers:
    :return:
    """
    media_index = len(instance_ids)//2
    instance_ids_group = [instance_ids[:media_index], instance_ids[media_index:]]
    for instance_id_list in instance_ids_group:
        # 1. 摘流量
        set_instance_weight_process(client=slb_client, instance_id_list=instance_id_list, weight_list=[(0, 60)])
        logging.info(f"set weight with 0 finished, instances: {instance_id_list}")
        # 2. 异步移除旧容器并删除旧镜像
        container_name = 'rov-server'
        loop = asyncio.get_running_loop()
        executor = ThreadPoolExecutor(max_workers=max_workers)
        tasks = [
            loop.run_in_executor(executor, remove_container_image, *args) for args in
            [(slb_client, instance_id, container_name) for instance_id in instance_id_list]
        ]
        await asyncio.wait(tasks)
        logging.info(f"remove container & images finished, instances: {instance_id_list}")
        # 3. 发送启动脚本到机器上
        utils.send_file_to_ecs(client=create_client, instance_id_list=instance_id_list, **rov_server_config.start_sh)
        logging.info(f"send start shell file finished, instances: {instance_id_list}")
        # 4. 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}"
        utils.run_command(client=create_client, instance_ids=instance_id_list, command=server_start_commend)
        # 5. 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 instance_id_list]
        ]
        await asyncio.wait(tasks)
        logging.info(f"health instances count: {len(health_instances)}, {health_instances}")
        # 6. 挂载流量
        if len(health_instances) == len(instance_id_list):
            # 所有机器探活成功
            time.sleep(60)
            add_weight_list = [(10, 30), (20, 20), (40, 10), (60, 10), (80, 10), (100, 10)]
            set_instance_weight_process(client=slb_client, instance_id_list=instance_id_list,
                                        weight_list=add_weight_list)
            logging.info(f"finished count: {len(health_instances)}")
        else:
            logging.info(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
    set_instance_weight_process(client=slb_client, instance_id_list=instance_ids, weight_list=[(0, 60)])
    logging.info(f"set weight = 0 finished, instances: {instance_ids}")
    time.sleep(10)
    # 2. 停止机器
    utils.stop_instances(client=create_client, instance_ids=instance_ids)
    logging.info(f"instances stop finished, instances: {instance_ids}")
    # 3. 判断机器运行状态是否为Stopped
    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')
        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}")
        else:
            logging.info(f"stopped instances count = {len(instance_ids)}, instances: {stopped_instances}")
            sys.exit()
    else:
        logging.error(response)
        sys.exit()
    # 4. 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():
    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/2
    logging.info(f"ess instances start ...")
    ess_instance_count = online_instance_count // 2
    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=2))
    logging.info(f"ess instances end!")

    # # 3. 原有机器进行更新
    # logging.info(f"update online instances start ...")
    # asyncio.run(update_instance(create_client=create_client, slb_client=slb_client,
    #                             instance_ids=online_instance_ids, max_workers=2))
    # logging.info(f"update online instances end!")

    # 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!")


if __name__ == '__main__':
    main()