GCE修改Hosterror等待时间

主机错误 (compute.instances.hostError) 意味着托管虚拟机的物理机器上存在硬件或软件问题，进而导致虚拟机崩溃。如果主机错误涉及硬件完全无法运行或其他硬件问题，则可能会阻止虚拟机实时迁移。如果您的虚拟机设置为自动重启（这是默认设置），Google 通常会在检测到错误后的 3 分钟内重启您的虚拟机。重启可能最多需要 5.5 分钟，具体取决于问题。

有时，虚拟机可能会在检测到主机错误之前无响应。您可以使用 --host-error-timeout-seconds 标志（预览版）缩短 Compute Engine 等待重启或终止虚拟机的时间。此标志设置 Compute Engine 在检测到虚拟机无响应后等待重启或终止虚拟机的最长时间。如需了解详情，请参阅设置可用性政策。

1. 缩短hosterr等待时间

• hostErrorTimeoutSeconds（预览版）：设置 Compute Engine 在检测到虚拟机无响应后等待重启或终止虚拟机的最长时间（以秒为单位）。
• [默认] 未设置，Compute Engine 最多会等待 5.5 分钟（330 秒），然后再重启无响应的虚拟机。
• 90 和 330 之间的秒数（以 30 为增量），这用于设置 Compute Engine 在重启无响应的虚拟机之前等待的时间。

修改示例：

gcloud beta compute instances set-scheduling wanggaoli-dasan  \
     --zone=asia-northeast1-b \
     --host-error-timeout-seconds=90

日志

{
  "protoPayload": {
    "@type": "type.googleapis.com/google.cloud.audit.AuditLog",
    "authenticationInfo": {
      "principalEmail": "wanggaoli@yunion-hk.com",
      "principalSubject": "user:wanggaoli@yunion-hk.com"
    },
    "requestMetadata": {
      "callerIp": "34.168.207.222",
      "callerSuppliedUserAgent": "google-cloud-sdk gcloud/405.0.0 command/gcloud.beta.compute.instances.set-scheduling invocation-id/63e829c5d6044343a26f86aedb13cdfd environment/devshell environment-version/None interactive/True from-script/False python/3.9.2 term/screen (Linux 5.10.133+),gzip(gfe)",
      "requestAttributes": {
        "time": "2022-10-21T05:55:12.761312Z",
        "auth": {}
      },
      "destinationAttributes": {}
    },
    "serviceName": "compute.googleapis.com",
    "methodName": "beta.compute.instances.setScheduling",
    "authorizationInfo": [
      {
        "permission": "compute.instances.setScheduling",
        "granted": true,
        "resourceAttributes": {
          "service": "compute",
          "name": "projects/mec-test-344202/zones/asia-northeast1-b/instances/wanggaoli-dasan",
          "type": "compute.instances"
        }
      }
    ],
    "resourceName": "projects/mec-test-344202/zones/asia-northeast1-b/instances/wanggaoli-dasan",
    "request": {
      "@type": "type.googleapis.com/compute.instances.setScheduling",
      "hostErrorTimeoutSeconds": "90"
    },
    "response": {
      "selfLinkWithId": "https://www.googleapis.com/compute/beta/projects/mec-test-344202/zones/asia-northeast1-b/operations/5345066212986423983",
      "status": "RUNNING",
      "targetLink": "https://www.googleapis.com/compute/beta/projects/mec-test-344202/zones/asia-northeast1-b/instances/wanggaoli-dasan",
      "zone": "https://www.googleapis.com/compute/beta/projects/mec-test-344202/zones/asia-northeast1-b",
      "progress": "0",
      "startTime": "2022-10-20T22:55:12.620-07:00",
      "id": "5345066212986423983",
      "@type": "type.googleapis.com/operation",
      "name": "operation-1666331711770-5eb8515c21079-cb78d67c-81a0d17f",
      "selfLink": "https://www.googleapis.com/compute/beta/projects/mec-test-344202/zones/asia-northeast1-b/operations/operation-1666331711770-5eb8515c21079-cb78d67c-81a0d17f",
      "targetId": "4609340071823249091",
      "insertTime": "2022-10-20T22:55:12.609-07:00",
      "operationType": "setScheduling",
      "user": "wanggaoli@yunion-hk.com"
    },
    "resourceLocation": {
      "currentLocations": [
        "asia-northeast1-b"
      ]
    }
  },
  "insertId": "f9l5ube2i15w",
  "resource": {
    "type": "gce_instance",
    "labels": {
      "zone": "asia-northeast1-b",
      "project_id": "mec-test-344202",
      "instance_id": "4609340071823249091"
    }
  },
  "timestamp": "2022-10-21T05:55:11.824199Z",
  "severity": "NOTICE",
  "logName": "projects/mec-test-344202/logs/cloudaudit.googleapis.com%2Factivity",
  "operation": {
    "id": "operation-1666331711770-5eb8515c21079-cb78d67c-81a0d17f",
    "producer": "compute.googleapis.com",
    "first": true
  },
  "receiveTimestamp": "2022-10-21T05:55:13.732887733Z"
}

{
  "protoPayload": {
    "@type": "type.googleapis.com/google.cloud.audit.AuditLog",
    "authenticationInfo": {
      "principalEmail": "wanggaoli@yunion-hk.com",
      "principalSubject": "user:wanggaoli@yunion-hk.com"
    },
    "requestMetadata": {
      "callerIp": "34.168.207.222",
      "callerSuppliedUserAgent": "google-cloud-sdk gcloud/405.0.0 command/gcloud.beta.compute.instances.set-scheduling invocation-id/63e829c5d6044343a26f86aedb13cdfd environment/devshell environment-version/None interactive/True from-script/False python/3.9.2 term/screen (Linux 5.10.133+),gzip(gfe)",
      "requestAttributes": {},
      "destinationAttributes": {}
    },
    "serviceName": "compute.googleapis.com",
    "methodName": "beta.compute.instances.setScheduling",
    "resourceName": "projects/mec-test-344202/zones/asia-northeast1-b/instances/wanggaoli-dasan",
    "request": {
      "@type": "type.googleapis.com/compute.instances.setScheduling"
    }
  },
  "insertId": "-l2ndfwddik4",
  "resource": {
    "type": "gce_instance",
    "labels": {
      "project_id": "mec-test-344202",
      "zone": "asia-northeast1-b",
      "instance_id": "4609340071823249091"
    }
  },
  "timestamp": "2022-10-21T05:55:15.286348Z",
  "severity": "NOTICE",
  "logName": "projects/mec-test-344202/logs/cloudaudit.googleapis.com%2Factivity",
  "operation": {
    "id": "operation-1666331711770-5eb8515c21079-cb78d67c-81a0d17f",
    "producer": "compute.googleapis.com",
    "last": true
  },
  "receiveTimestamp": "2022-10-21T05:55:15.745905867Z"
}

修改多台机器：

# inlist.txt为VM主机名列表
while read name;do gcloud beta compute instances set-scheduling $name --host-error-timeout-seconds=90 --zone=asia-northeast1-b; done < inslist.txt

2. 获取迁移通知

此元数据键的值会在维护事件开始前 60 秒发生更改，让您的应用代码能够在维护事件之前触发您要执行的任何任务，例如备份数据或更新日志。