使用 Terraform 的 CloudWatch 指标警报

Posted 2023-03-24

【中文标题】使用 Terraform 的 CloudWatch 指标警报

【英文标题】：CloudWatch metric alarm using Terraform

【发布时间】：2019-12-26 20:08:31

【问题描述】：

由于某种原因尝试使用 Terraform 设置一些 CloudWatch 警报时，它找不到指标并且警报仍然卡在数据不足的情况下。 Terraform 不会输出任何错误，如果我在 AWS 中手动搜索，我可以找到指标。我在这里错过了什么？

一个简单的健康主机警报指向目标组的例子：

#healthy host alarm
resource "aws_cloudwatch_metric_alarm" "health" 
  alarm_name          = "$var.tag_app_healthy_host"
  comparison_operator = "LessThanThreshold"
  evaluation_periods  = "1"
  metric_name         = "HealthyHostCount"
  namespace           = "AWS/ApplicationELB"
  period              = "60"
  statistic           = "Maximum"
  threshold           = "1"
  alarm_description   = "Healthy host count for EC2 machine"
  alarm_actions       = ["$data.aws_sns_topic.blabla.arn"]
  ok_actions          = ["$data.aws_sns_topic.blabla.arn"]

  dimensions = 
    TargetGroup  = "$aws_lb_target_group.alb_target.arn_suffix"
  

当我选择另一个资源（EC2、RDS）和另一个指标时，我会收到指向正确指标的 CloudWatch 警报，并且不会因为数据不足而卡住。

【参考方案1】：

HealthyHostCount metric 仅在TargetGroup, LoadBalancer 维度或TargetGroup, AvailabilityZone, LoadBalancer 下可用，因此您至少还需要添加LoadBalancer 维度才能访问此指标。

所以你的 Terraform 代码应该是：

#healthy host alarm
resource "aws_cloudwatch_metric_alarm" "health" 
  alarm_name          = "$var.tag_app_healthy_host"
  comparison_operator = "LessThanThreshold"
  evaluation_periods  = "1"
  metric_name         = "HealthyHostCount"
  namespace           = "AWS/ApplicationELB"
  period              = "60"
  statistic           = "Maximum"
  threshold           = "1"
  alarm_description   = "Healthy host count for EC2 machine"
  alarm_actions       = ["$data.aws_sns_topic.blabla.arn"]
  ok_actions          = ["$data.aws_sns_topic.blabla.arn"]

  dimensions = 
    LoadBalancer = "$aws_lb.example.arn_suffix"
    TargetGroup  = "$aws_lb_target_group.alb_target.arn_suffix"
  

【讨论】：

你怎么知道哪些指标需要哪些维度？

这真的应该作为一个单独的问题来问。