add clb_gateway_update_cluster

gateway/clb_gateway_update_cluster.py

@@ -0,0 +1,323 @@
+import asyncio
+import sys
+import time
+import requests
+import clb_utils
+import logging
+import os
+import docker
+import gateway_config
+
+from concurrent.futures import ThreadPoolExecutor
+
+
+health_instances = []
+ess_instances = []
+remove_container_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 = clb_utils.get_ip_address(client=client, instance_id=instance_id)
+    while True:
+        health_check_url = f"http://{ip_address}:8080/longvideoapi/test"
+        try:
+            http_code = requests.get(health_check_url).status_code
+        except:
+            logging.info(f"images is downloading ip：{ip_address}")
+            http_code = 0
+
+        if http_code == 200:
+            health_instances.append((instance_id, ip_address))
+            logging.info(f"health check success, instance: {instance_id}/{ip_address}")
+            break
+        else:
+            time.sleep(10)
+
+
+async def ess_instance(ecs_client, clb_client, ess_count, max_workers, version):
+    """
+    扩容机器并运行新服务
+    :param ecs_client: 购买机器客户端连接
+    :param clb_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 = clb_utils.create_multiple_instances(
+        amount=ess_count,
+        client=ecs_client,
+        **gateway_config.instance_config_j,
+    )
+    time.sleep(60)
+
+    # 2. 发送启动脚本到机器上
+    clb_utils.send_file_to_ecs(client=ecs_client, instance_id_list=ess_instance_ids, **gateway_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(gateway_config.start_sh['target_dir'], gateway_config.start_sh['name'])
+    server_start_commend = f"sh {server_start_sh} {version}"
+    clb_utils.run_command(client=ecs_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
+        [(clb_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(10)
+        clb_utils.add_backend_servers_with_clbs(client=clb_client,
+                                            clb_id_list=gateway_config.clb_id_list,
+                                            instances=health_instances)
+        health_instance_ids = [instance_id for instance_id, _ in health_instances]
+        add_weight_list = [(10, 5), (20, 5), (40, 5), (60, 5), (80, 5), (100, 5)]
+        clb_utils.set_instance_weight_process_with_clbs(client=clb_client,
+                                                    clb_id_list=gateway_config.clb_id_list,
+                                                    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_container_image(client, instance_id, container_name_list):
+    """
+    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_list: 容器名称 type-string
+    :return:
+    """
+    ip_address = clb_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 True:
+        if i >= container_remove_retry:
+            logging.error(f"容器不存在或者无法删除当前容器, instance = {instance_id}/{ip_address}")
+            sys.exit()
+        try:
+            flag = False
+            for container_name in container_name_list:
+                try:
+                    container_id = client.containers.get(container_name)
+                    container_id.remove(force=True)
+                    flag = True
+                    break
+                except:
+                    continue
+            if flag:
+                break
+        except Exception as e:
+            i += 1
+
+    # 删除旧镜像
+    images_remove_retry = 3
+    j = 0
+    while True:
+        if j >= images_remove_retry:
+            logging.error(f"镜像不存在，无法获取到镜像ID, instance = {instance_id}/{ip_address}")
+            sys.exit()
+        try:
+            images = client.images.list()
+            for image in images:
+                client.images.remove(force=True, image=image.tags[0])
+                time.sleep(2)
+            global remove_container_instances
+            remove_container_instances.append(instance_id)
+            break
+        except Exception as e:
+            i += 1
+
+
+async def update_instance(ecs_client, clb_client, instance_ids, max_workers, version):
+    """
+    线上机器更新
+    :param ecs_client:
+    :param clb_client: slb客户端连接
+    :param instance_ids: instanceId type-list
+    :param max_workers:
+    :param version: 版本标记
+    :return:
+    """
+    media_index = len(instance_ids)//2
+    instance_ids_group = [instance_ids[:media_index], instance_ids[media_index:]]
+    update_finished_count = 0
+    for instance_id_list in instance_ids_group:
+        logging.info(f"update instances: {instance_id_list}")
+        # 1. 摘流量
+        clb_utils.set_instance_weight_process_with_clbs(client=clb_client,
+                                                    clb_id_list=gateway_config.clb_id_list,
+                                                    instance_id_list=instance_id_list,
+                                                    weight_list=[(0, 15)])
+        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
+        global remove_container_instances
+        remove_container_instances = []
+        # container_name = 'longvideoapi'
+        container_name_list = ['vlogapi', 'longvideoapi']
+        loop = asyncio.get_running_loop()
+        executor = ThreadPoolExecutor(max_workers=max_workers)
+        tasks = [
+            loop.run_in_executor(executor, remove_container_image, *args) for args in
+            [(clb_client, instance_id, container_name_list) for instance_id in instance_id_list]
+        ]
+        await asyncio.wait(tasks)
+        logging.info(f"remove container & images finished, instances: {remove_container_instances},"
+                     f" count: {len(remove_container_instances)}")
+        if len(remove_container_instances) < len(instance_id_list):
+            logging.error(f"remove container image failed| "
+                          f"request count: {len(instance_id_list)}, removed count: {len(remove_container_instances)}")
+            sys.exit()
+        # 3. 发送启动脚本到机器上
+        clb_utils.send_file_to_ecs(client=ecs_client, instance_id_list=instance_id_list, **gateway_config.start_sh)
+        logging.info(f"send start shell file finished, instances: {instance_id_list}, count: {len(instance_id_list)}")
+        # 4. 启动服务
+        server_start_sh = os.path.join(gateway_config.start_sh['target_dir'], gateway_config.start_sh['name'])
+        server_start_commend = f"sh {server_start_sh} {version}"
+        clb_utils.run_command(client=ecs_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
+            [(clb_client, instance_id) for instance_id in instance_id_list]
+        ]
+        await asyncio.wait(tasks)
+        logging.info(f"health instances: {health_instances}, count: {len(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(10)
+            clb_utils.add_backend_servers_with_clbs(client=clb_client,
+                                                clb_id_list=gateway_config.clb_id_list,
+                                                instances=health_instances)
+            health_instance_ids = [instance_id for instance_id, _ in health_instances]
+            add_weight_list = [(10, 5), (20, 5), (40, 5), (60, 5), (80, 5), (100, 5)]
+            clb_utils.set_instance_weight_process_with_clbs(client=clb_client,
+                                                        clb_id_list=gateway_config.clb_id_list,
+                                                        instance_id_list=health_instance_ids,
+                                                        weight_list=add_weight_list)
+            logging.info(f"finished instances: {health_instances}, count: {len(health_instances)}")
+            update_finished_count += len(health_instances)
+            logging.info(f"update finished: {update_finished_count}/{len(instance_ids)}")
+        else:
+            logging.info(f"health instances: {health_instances}, count: {len(health_instances)}")
+            sys.exit()
+
+
+def remove_instances(ecs_client, clb_client, instance_ids):
+    """
+    停止并释放机器
+    :param ecs_client:
+    :param clb_client:
+    :param instance_ids: instanceId type-list
+    :return: None
+    """
+    # 1. 摘流量
+    clb_utils.set_instance_weight_process_with_clbs(client=clb_client,
+                                                clb_id_list=gateway_config.clb_id_list,
+                                                instance_id_list=instance_ids,
+                                                weight_list=[(0, 20)])
+    logging.info(f"set weight = 0 finished, instances: {instance_ids}")
+    time.sleep(10)
+    # 2.移除slb
+    clb_utils.remove_backend_servers_with_clbs(client=clb_client,
+                                           clb_id_list=gateway_config.clb_id_list,
+                                           instances=instance_ids)
+    # 3. 停止机器
+    clb_utils.stop_instances(client=ecs_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 = clb_utils.get_instances_status(client=ecs_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(5)
+        else:
+            logging.error(response)
+            sys.exit()
+    # 5. 释放机器
+    response = clb_utils.release_instances(client=ecs_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:
+        version = sys.argv[1]
+        clb_client = clb_utils.connect_client(access_key_id=gateway_config.clb_client_params['access_key_id'],
+                                          access_key_secret=gateway_config.clb_client_params['access_key_secret'],
+                                          region_id=gateway_config.clb_client_params['region_id'])
+        ecs_client = clb_utils.connect_client(access_key_id=gateway_config.ecs_client_params['access_key_id'],
+                                             access_key_secret=gateway_config.ecs_client_params['access_key_secret'],
+                                             region_id=gateway_config.ecs_client_params['region_id'])
+
+        # 1. 获取clb下所有机器
+        online_instance_ids = clb_utils.get_instance_ids(client=clb_client, clb_id=gateway_config.clb_id_list[0])
+        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(ecs_client=ecs_client, clb_client=clb_client,
+                                 ess_count=ess_instance_count, max_workers=2, version=version))
+        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(ecs_client=ecs_client, clb_client=clb_client,
+                                    instance_ids=online_instance_ids, max_workers=8, version=version))
+        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(ecs_client=ecs_client, clb_client=clb_client, instance_ids=ess_instances)
+        logging.info(f"stop & release instances end!")
+    except Exception as e:
+        logging.error(e)
+        sys.exit()
+
+
+if __name__ == '__main__':
+    main()

gateway/clb_gateway_update_one.py

@@ -0,0 +1,179 @@
+#!/bin/env python
+#coding=utf-8
+#edite  panwang
+import docker
+import sys
+import  requests
+import json
+import queue
+import threading
+from aliyunsdkcore import client
+from aliyunsdkecs.request.v20140526.DescribeNetworkInterfacesRequest import DescribeNetworkInterfacesRequest
+from aliyunsdkcore.request import CommonRequest
+from aliyunsdkslb.request.v20140515.DescribeLoadBalancerAttributeRequest import DescribeLoadBalancerAttributeRequest
+import time
+
+
+AccessKey = 'LTAIuPbTPL3LDDKN'
+AccessSecret = 'ORcNedKwWuwVtcq4IRFtUDZgS0b1le'
+RegionId = 'cn-hangzhou'
+version = sys.argv[1]
+slbIDs = ['lb-bp1jtzhp9krunyv3mim2q','lb-bp1mfk1gmd47twfh6bgwv']
+#slbIDs = ['lb-bp136gme0r0n02aew4vjc','lb-bp1mfk1gmd47twfh6bgwv']
+apps = 'piaoquan-gateway'
+repository = 'registry-vpc.cn-hangzhou.aliyuncs.com/stuuudy/{}'.format(apps)
+registry = 'registry-vpc.cn-hangzhou.aliyuncs.com/stuuudy/{}:{}'.format(apps, version)
+
+
+
+clt = client.AcsClient (AccessKey, AccessSecret, RegionId)
+
+class MyThread (threading.Thread):
+    def __init__(self, func):
+        threading.Thread.__init__ (self)
+
+        self.func = func
+
+    def run(self):
+        self.func ()
+
+def  checkHealth(ipadd):
+    while  True:
+        health_url = 'http://%s:9000/healthcheck' %(ipadd)
+        header = {"Content-Type":"application/json"}
+        try:
+            health_code = requests.get(health_url).status_code
+        except  Exception  as  e:
+           continue
+        if  health_code == 200:
+            print("httpcode 200,开始挂载流量")
+            return  False
+#服务更新完之后逐步修改服务器的权重值，直接加载100会出现502。权重值每次增加10，每2s修改一次
+def setInstanceWeightProcess(slb_id, instance_id):
+    for  i in range(1,6):
+        weight = i*20
+        setWeight(slb_id,instance_id,weight)
+        print("当前权重： ", weight)
+        time.sleep(5)
+
+#设置权重 instance_id :服务器id，weight：权重值
+def setWeight(slb_id,instance_id ,weight):
+
+    BackendServers = [{"ServerId": instance_id, "Weight": weight}]
+
+    request = CommonRequest ()
+    request.set_accept_format ('json')
+    request.set_domain ('slb.aliyuncs.com')
+    request.set_version ('2014-05-15')
+    request.set_method ('POST')
+    request.set_action_name ('SetBackendServers')
+    request.add_query_param ('BackendServers', BackendServers)
+    request.add_query_param ('LoadBalancerId', slb_id)
+    try:
+        response = clt.do_action (request)
+    except Exception as e:
+        print (e)
+
+
+def  getInstanceId(slb_id):
+    request = DescribeLoadBalancerAttributeRequest()
+    request.set_accept_format('json')
+    request.set_LoadBalancerId(slb_id)
+    response = clt.do_action_with_exception(request)
+    return json.loads (response)
+
+
+#获取实例IP地址
+def  getIpadd(instance_id):
+    request = DescribeNetworkInterfacesRequest()
+    request.set_accept_format('json')
+    request.set_InstanceId(instance_id)
+    response = clt.do_action_with_exception(request)
+    request_content = json.loads(response)
+    IpAddr  = request_content['NetworkInterfaceSets']['NetworkInterfaceSet'][0]['PrivateIpAddress']
+    return  IpAddr
+
+
+#更新服务
+def  update(instance_id):
+    time.sleep(10)
+    global success_count
+
+    ipadd = getIpadd(instance_id)
+    print("服务器信息：" + "%s/%s" %(instance_id, ipadd))
+    client = docker.DockerClient(base_url='tcp://%s:2375'  %(ipadd),timeout=60 )
+    try:
+    #更新前移除旧的容器
+        id = client.containers.get(apps)
+        id.remove(force = True)
+
+    except Exception as e:
+        print("容器不存在或者无法删除当前容器")
+
+    try:
+        #登录镜像仓库
+        client.login(username='stuuudys' ,password='Qingqu@2019', registry='registry-vpc.cn-hangzhou.aliyuncs.com')
+        #启动一个容器
+        client.containers.run(registry, detach = True, cap_add = 'SYS_PTRACE', network_mode = 'host', name = apps,volumes={'/datalog/': {'bind': '/datalog/', 'mode': 'rw'}})
+        print("开始健康检查")
+        checkHealth(ipadd)
+        print("%s ：权重修改中......" %(ipadd))
+
+        for slbID in slbIDs:
+            setInstanceWeightProcess(slbID, instance_id)
+        success_count = success_count + 1
+        print("更新进度" + "%s/%s"  %(success_count, total))
+    except Exception as e:
+        print(e)
+        sys.exit()
+        #容器启动失败立即退出更新
+
+
+def  pull_image():
+    instanceId = q1.get()
+    ipaddr = getIpadd(instanceId)
+    cd_url = "tcp://{}:2375".format(ipaddr)
+    client = docker.DockerClient(base_url=cd_url, timeout=30)
+
+    try:
+        client.images.pull(repository, tag=version)
+        print(ipaddr, "pull  images success ")
+
+        return  True
+    except Exception as e:
+        print(e, "images pull  fail")
+        return  False
+
+
+if __name__ == '__main__':
+
+
+    #更新完成计数
+    success_count = 0
+    threads = []
+    res = getInstanceId (slbIDs[0])
+    #slb下服务器总数
+    total = len(res["BackendServers"]["BackendServer"])
+    InstanceIDs = []
+    q1 = queue.Queue()
+    if res["BackendServers"]["BackendServer"]:
+        for i in range ((len (res["BackendServers"]["BackendServer"]))):
+            InstanceID = res["BackendServers"]["BackendServer"][i]["ServerId"]
+            InstanceIDs.append(InstanceID)
+            q1.put(InstanceID)
+    print(InstanceIDs)
+    #多线程预先pull images
+    for i in range (len(InstanceIDs)):
+        thread = MyThread (pull_image)
+        thread.start ()
+        threads.append(thread)
+    for thread in threads:
+            thread.join ()
+
+    #单线程更新
+    for instanceID in InstanceIDs:
+        for slbID in slbIDs:
+            setWeight(slbID,instanceID,0)
+        update(instanceID)
+
+

gateway/clb_utils.py

@@ -272,11 +272,11 @@ def get_ip_address(client, instance_id):
     return ip_address
 
 
-def set_weight_for_instances(client, slb_id, instance_id_list, weight):
+def set_weight_for_instances(client, clb_id, instance_id_list, weight):
     """
     同时设置多台服务器的slb权重，权重一样
     :param client: 客户端连接
-    :param slb_id: slb_id
+    :param clb_id: clb_id
     :param instance_id_list: 服务器id list
     :param weight: 权重值
     :return: None
@@ -293,7 +293,7 @@ def set_weight_for_instances(client, slb_id, instance_id_list, weight):
         request.set_method('POST')
         request.set_action_name('SetBackendServers')
         request.add_query_param('BackendServers', BackendServers)
-        request.add_query_param('LoadBalancerId', slb_id)
+        request.add_query_param('LoadBalancerId', clb_id)
         response = send_request(client=client, request=request)
 
 
@@ -364,11 +364,11 @@ def get_instances_status(client, instance_ids):
     return response
 
 
-def set_instance_weight_process(client, slb_id, instance_id_list, weight_list):
+def set_instance_weight_process(client, clb_id, instance_id_list, weight_list):
     """
     修改服务器的权重值
-    :param client: slb客户端连接
-    :param slb_id: slb id
+    :param client: clb客户端连接
+    :param clb_id: clb id
     :param instance_id_list: instance id list
     :param weight_list: 权重修改列表 type-list [(weight, sleep_time), ...]
     :return:
@@ -378,7 +378,7 @@ def set_instance_weight_process(client, slb_id, instance_id_list, weight_list):
         flag = True
         while flag:
             try:
-                set_weight_for_instances(client=client, slb_id=slb_id, instance_id_list=instance_id_list, weight=weight)
+                set_weight_for_instances(client=client, clb_id=clb_id, instance_id_list=instance_id_list, weight=weight)
                 time.sleep(sleep_time)
                 flag = False
             except Exception as e:
@@ -444,7 +444,7 @@ def set_instance_weight_process_with_clbs(client, clb_id_list, instance_id_list,
     """
     修改服务器的权重值
     :param client: clb客户端连接
-    :param clb_id_list: slb id list
+    :param clb_id_list: clb id list
     :param instance_id_list: instance id list
     :param weight_list: 权重修改列表 type-list [(weight, sleep_time), ...]
     :return:
@@ -455,7 +455,7 @@ def set_instance_weight_process_with_clbs(client, clb_id_list, instance_id_list,
             flag = True
             while flag:
                 try:
-                    set_weight_for_instances(client=client, slb_id=clb_id, instance_id_list=instance_id_list, weight=weight)
+                    set_weight_for_instances(client=client, clb_id=clb_id, instance_id_list=instance_id_list, weight=weight)
                     logging.info(f"slb: {clb_id} finished！")
                     flag = False
                 except Exception as e: