Google Cloud 不会替换有状态管理实例组内的实例

Question

使用以下代码创建的托管实例组，但在“替换 VMS”时出现此错误：

Failed to update instance group ssv-node-igm: UpdatePolicy with type set to PROACTIVE and replacementMethod different than RECREATE is not allowed when Stateful is used (instanceGroupManager has statefulPolicy or perInstanceConfigs). Set replacementMethod to RECREATE to apply this update.

此外，MIG 不会替换运行状况检查失败的 VM，但会给出“实例正在验证”以获取更多信息超过6小时。

代码：

  resource "google_compute_health_check" "autohealing" {
    name                = "${var.name_space}-ssv-health-check"
    check_interval_sec  = 5
    timeout_sec         = 5
    healthy_threshold   = 2
    unhealthy_threshold = 10
  
    tcp_health_check {
      port = "13000"
    }
  }
  
  resource "google_compute_instance_group_manager" "ssv-node" {
    name = "ssv-node-igm"
  
    base_instance_name = "${var.name_space}-ssv-node"
    zone               = "us-central1-a"
  
    version {
      instance_template = google_compute_instance_template.ssv-template.id
      name              = "${var.name_space}-ssv-node-igm-v1"
    }
  
    target_size = 1
  
    stateful_disk {
      device_name = "${var.name_space}-boot"
    }
  
    stateful_disk {
      device_name = "${var.name_space}-data"
    }
  
  
    named_port {
      name = "customhttp"
      port = 8888
    }
  
    auto_healing_policies {
      health_check      = google_compute_health_check.autohealing.id
      initial_delay_sec = 30
    }
  
    update_policy {
      type               = "PROACTIVE"
      minimal_action     = "REPLACE"
      max_surge_fixed    = 0
      max_unavailable_fixed = 1
      replacement_method = "RECREATE"
    }
  }

您可以发现更新策略与错误消息建议的操作完全相同，但仍然出现相同的错误。

提出任何建议！

Answer 1

我有一张 Google 支持票，问题是 GUI 问题。使用下面的 cli，它可以工作。

gcloud compute instance-groups managed rolling-action replace my-mig

除此之外，对于使用运行状况检查的任何人来说，还有一个注意事项：在 MIG 终止不运行状况的实例之前，您必须具有运行状况。换句话说，如果您的实例从未达到健康状态，它将永远保留。