Google GCP 自动开机 Python 脚本

大家可关注我的新Telegram频道，以获取更多相关信息和更新。感谢您的关注！
https://t.me/kkkkkcat/

完整文章：https://kkcat.blog/posts/google-gcp-spot-script-2024/

介绍

本文将介绍一个Python脚本，自动检查并启动处于非运行状态的GCE(Google Cloud Compute Engine)实例，以确保其持续可用。

需求

• 在开始之前，请确保您已经具备以下条件：

• 安装了Google Cloud SDK。

• 配置了正确的GCP项目，并下载了服务账号的JSON密钥文件。

• 在您的系统中安装了google-cloud-compute库：

pip install google-cloud-compute

脚本功能

这个Python脚本将执行以下任务：

列出所有区域中的所有GCE实例。
检查每个实例的状态。
启动所有处于非运行状态的实例。
每5分钟重复执行上述任务，以确保实例的持续可用性。

设置服务账号密钥：

1. API和服务 -> 凭证 -> 服务帐户(默认那个即可) -> 金钥 -> 新增金钥(json档)
2. project ID看图取

脚本代码

import os
import time
from google.cloud import compute_v1

def start_instance_if_not_running(project_id, zone, instance_name):
    instance_client = compute_v1.InstancesClient()
    instance = instance_client.get(project=project_id, zone=zone, instance=instance_name)
    
    if instance.status != "RUNNING":
        print(f"Instance {instance_name} is not running. Starting the instance.")
        operation = instance_client.start(project=project_id, zone=zone, instance=instance_name)
        wait_for_operation(project_id, zone, operation.name)
        instance = instance_client.get(project=project_id, zone=zone, instance=instance_name)
        print(f"Instance {instance_name} is now {instance.status}.")
    else:
        print(f"Instance {instance_name} is already running.")
    
    return instance.status

def wait_for_operation(project_id, zone, operation_name):
    operation_client = compute_v1.ZoneOperationsClient()
    while True:
        result = operation_client.get(project=project_id, zone=zone, operation=operation_name)
        if result.status == "DONE":
            if result.error:
                raise Exception(f"Error during operation: {result.error}")
            print("Operation finished successfully.")
            break
        time.sleep(5)

def list_and_start_instances(project_id):
    compute_client = compute_v1.InstancesClient()
    zones_client = compute_v1.ZonesClient()

    result_table = "Instance name | Zone | Status\n\n"

    zones = zones_client.list(project=project_id)
    instances_count = 0

    for zone in zones:
        zone_name = zone.name
        request = compute_v1.ListInstancesRequest(project=project_id, zone=zone_name)
        response = compute_client.list(request=request)

        for instance in response:
            instances_count += 1
            status = start_instance_if_not_running(project_id, zone_name, instance.name)
            result_table += f"{instance.name} | {zone_name} | {status}\n"

    print(f"Total number of instances: {instances_count}")
    print(result_table)

if __name__ == "__main__":
    os.environ["GOOGLE_APPLICATION_CREDENTIALS"] = "xxxxxxxxxx.json"
    project_id = "xxxxxxxxxxxxx"
    
    while True:
        list_and_start_instances(project_id)
        time.sleep(300)

代码解释

1. 设置服务账号密钥：

os.environ["GOOGLE_APPLICATION_CREDENTIALS"] = "xxxxxxxxxx.json"

这行代码指定了服务账号密钥文件的路径，以便脚本能够使用正确的凭证访问GCP资源。

2. 启动非运行状态实例：

def start_instance_if_not_running(project_id, zone, instance_name):
    # 获取实例状态
    instance_client = compute_v1.InstancesClient()
    instance = instance_client.get(project=project_id, zone=zone, instance=instance_name)
    
    if instance.status != "RUNNING":
        print(f"Instance {instance_name} is not running. Starting the instance.")
        operation = instance_client.start(project=project_id, zone=zone, instance=instance_name)
        wait_for_operation(project_id, zone, operation.name)
        instance = instance_client.get(project=project_id, zone=zone, instance=instance_name)
        print(f"Instance {instance_name} is now {instance.status}.")
    else:
        print(f"Instance {instance_name} is already running.")
    
    return instance.status

3. 等待操作完成：

   def wait_for_operation(project_id, zone, operation_name):
    operation_client = compute_v1.ZoneOperationsClient()
    while True:
        result = operation_client.get(project=project_id, zone=zone, operation=operation_name)
        if result.status == "DONE":
            if result.error:
                raise Exception(f"Error during operation: {result.error}")
            print("Operation finished successfully.")
            break
        time.sleep(5)
   

4. 列出并启动实例：

   def list_and_start_instances(project_id):
    compute_client = compute_v1.InstancesClient()
    zones_client = compute_v1.ZonesClient()
   
    result_table = "| Instance name | Status |\n"
   
    zones = zones_client.list(project=project_id)
    instances_count = 0
   
    for zone in zones:
        zone_name = zone.name
        request = compute_v1.ListInstancesRequest(project=project_id, zone=zone_name)
        response = compute_client.list(request=request)
   
        for instance in response:
            instances_count += 1
            status = start_instance_if_not_running(project_id, zone_name, instance.name)
            result_table += f"| {instance.name} | {zone_name} | {status} |\n"
   
    print(f"Total number of instances: {instances_count}")
    print(result_table)
   

   该函数遍历所有区域中的所有实例，并检查每个实例的状态，必要时启动实例。

5. 主循环：

   if __name__ == "__main__":
    project_id = "xxxxxxxxxxxxx"
    
    while True:
        list_and_start_instances(project_id)
        time.sleep(300)
   

   这个主循环每5分钟运行一次list_and_start_instances函数，以确保所有实例持续可用。

结论

通过这个Python脚本，可以自动化管理GCP中的虚拟机实例，确保它们始终处于运行状态。
如果您有更多的实例或区域需要管理，可以根据需要扩展这个脚本。