函数计算自动化运维实战2 -- 事件触发eip自动转移

函数计算

阿里云函数计算是一个事件驱动的全托管计算服务。通过函数计算，您无需管理服务器等基础设施，只需编写代码并上传。函数计算会为您准备好计算资源，以弹性、可靠的方式运行您的代码，并提供日志查询，性能监控，报警等功能。借助于函数计算，您可以快速构建任何类型的应用和服务，无需管理和运维。更棒的是，您只需要为代码实际运行消耗的资源付费，而代码未运行则不产生费用。

云监控

阿里云云监控为云上用户提供开箱即用的企业级开放型一站式监控解决方案。涵盖IT设施基础监控，外网网络质量拨测监控，基于事件、自定义指标、日志的业务监控。为您全方位提供更高效、更全面、更省钱的监控服务。

云监控提供了丰富事件，并且事件还在不断丰富中（云产品系统事件监控）， 丰富的事件触发自定义处理的函数，可以实现更加完美的的自动化运维。

专题传送门 => 函数计算进行自动化运维专题

示例场景

假设两台ECS机器A, B, A 机器出现了宕机，这个时候需要把A机器上的eip迁移到备用机器B上，这个利用云监控的报警和函数计算可以实现EIP的自动迁移， 云监控检测到A宕机这个事件，然后触发函数执行，函数实现eip的自动迁移

操作步骤

• 创建函数(函数代码在文末),函数创建可参考函数计算helloworld

注：记得给函数的service的role设置操作ecs和eip的权限

• 登录云监控控制台, 创建报警规则, 监控的事件为ecs 重启开始
  

• mock调试
  

• 模拟真实的ecs事件
  请参考演练系统事件处理程序？ So Easy~

代码

# -*- coding: utf-8 -*-
import logging
import json, random, string, time
from aliyunsdkcore import client
from aliyunsdkvpc.request.v20160428.AssociateEipAddressRequest import AssociateEipAddressRequest
from aliyunsdkvpc.request.v20160428.UnassociateEipAddressRequest import UnassociateEipAddressRequest
from aliyunsdkvpc.request.v20160428.DescribeEipAddressesRequest import DescribeEipAddressesRequest
from aliyunsdkcore.auth.credentials import StsTokenCredential

LOGGER = logging.getLogger()
clt = None

def handler(event, context):
  creds = context.credentials
  sts_token_credential = StsTokenCredential(creds.access_key_id, creds.access_key_secret, creds.security_token)
  '''
  {
    "product": "ECS",
    "content": {
        "executeFinishTime": "2018-06-08T01:25:37Z",
        "executeStartTime": "2018-06-08T01:23:37Z",
        "ecsInstanceName": "timewarp",
        "eventId": "e-t4nhcpqcu8fqushpn3mm",
        "eventType": "InstanceFailure.Reboot",
        "ecsInstanceId": "i-bp18l0uopocfc98xxxx" 
    },
    "resourceId": "acs:ecs:cn-hangzhou:123456789:instance/i-bp18l0uopocfc98xxxx",
    "level": "CRITICAL",
    "instanceName": "instanceName",
    "status": "Executing",
    "name": "Instance:SystemFailure.Reboot:Executing", 
    "regionId": "cn-hangzhou"
  }
  '''

  evt = json.loads(event)
  content = evt.get("content");
  ecsInstanceId = content.get("ecsInstanceId");
  regionId = evt.get("regionId");
  
  global clt
  clt = client.AcsClient(region_id=regionId, credential=sts_token_credential)
  
  name = evt.get("name");
  eipId = "eip-bp1nexxxc7zjfnex6i0";
  standbyEcsId = "i-bp19yycxx7yroukxpv"
  name = name.lower()
  
  if name in ['Disk:Stalled:Executing'.lower(), 'Instance:SystemFailure.Reboot:Executing'.lower(), "Instance:InstanceFailure.Reboot:Executing".lower()]:
    print("move eip to standbyEcs");
    move_eip(ecsInstanceId, standbyEcsId, eipId) # move eip to standbyEcs
  else:
    # other event to do 
    pass
  
def getEipStatus(eip):
  request = DescribeEipAddressesRequest()
  request.set_AllocationId(eip)
  request.add_query_param("RegionId", "cn-hangzhou")
  response = _send_request(request)
  if isinstance(response, dict) and "RequestId" in response:
    EipAddresses = response.get('EipAddresses', {})
    EipAddress = EipAddresses['EipAddress'][0]
    status = EipAddress['Status']
    return status
  else:
    LOGGER.error("getEipAddressDesc {} fail".format(eip))

def unAssociateEip(ecs_id, eip):
  request = UnassociateEipAddressRequest()
  request.set_AllocationId(eip)
  request.set_InstanceType('EcsInstance')
  request.set_InstanceId(ecs_id)
  response = _send_request(request)

  if isinstance(response, dict) and "RequestId" in response:
    LOGGER.info("UnassociateEipAddress {} from {} succ".format(ecs_id, eip))
  else:
    LOGGER.error("UnassociateEipAddress {} from {} fail".format(ecs_id, eip))

def associateEip(ecs_id, eip):
  associte_request = AssociateEipAddressRequest()
  associte_request.set_AllocationId(eip)
  associte_request.set_InstanceType('EcsInstance')
  associte_request.set_InstanceId(ecs_id)
  associte_response = _send_request(associte_request)
  if isinstance(associte_response, dict) and "RequestId" in associte_response:
    LOGGER.info("AssociateEipAddress {} to {} succ".format(eip, ecs_id))
    return True
  return False

def move_eip(from_ecs, to_ecs, eip):
  unAssociateEip(from_ecs, eip)
  # wait unAssociateEip ...
  time.sleep(3)
  # retry 30s util sucess
  for i in xrange(10):
    eip_status = getEipStatus(eip).lower()
    if eip_status == 'available':
      if associateEip(to_ecs, eip):
        LOGGER.info("AssociateEipAddress {} to {} succ".format(eip, to_ecs))
        return
    else:
      LOGGER.info("eip status = {}".format(eip_status))
      time.sleep(3)

  LOGGER.info("AssociateEipAddress {} to {} fail".format(eip, to_ecs))

# send open api request
def _send_request(request):
    request.set_accept_format('json')
    try:
        response_str = clt.do_action_with_exception(request)
        LOGGER.info(response_str)
        response_detail = json.loads(response_str)
        return response_detail
    except Exception as e:
        LOGGER.error(e)