Terraform：为多个实例创建 CloudWatch 警报时出错

Question

我正在两个区域创建多个 ec2 实例。我想关联 CloudWatch 警报以进行状态检查和 CPU 利用率。

下面提到了cloudwatch的目录结构和代码，main.tf代表模块的调用

我有 2 个问题，包括创建 cloudwatch 警报的逻辑。

目录结构：

├── main.tf
├── modules
│   ├── alb
│   │   ├── aws_alb.tf
│   │   ├── aws_instance.tf
│   │   ├── bootstrap.sh
│   │   ├── cloudwatch.tf
│   │   ├── main.tf
│   │   ├── output.tf
│   │   ├── security-group.tf
│   │   ├── sns.tf
│   │   └── variables.tf
│   └── route53
│       ├── main.tf
│       └── variables.tf
└── variables.tf

main.tf

module "north-virginia" {
  source          = "./modules/alb"
  region          = "us-east-1"
  az              = ["us-east-1a", "us-east-1b", "us-east-1c"]
}

module "oregon" {
  source          = "./modules/alb"
  region          = "us-west-2"
  az              = ["us-west-2a", "us-west-2b", "us-west-2c"]
}

modules/alb/aws_instance.tf

resource "aws_instance" "web" {
  ami               = "${data.aws_ami.amzn2.id}"
  instance_type     = "${var.instance_type}"
  count             = 3
  availability_zone = "${element(var.az, count.index)}"
  tags {
    Name = "${count.index}"
  }
}

modules/alb/cloudwatch.tf

resource "aws_cloudwatch_metric_alarm" "cpu_utilization" {
  count               = "${length(local.instance_id_var)}"
  alarm_name          = "${element(split(",", join(",", aws_instance.web.*.id)), count.index)}"
  comparison_operator = "GreaterThanOrEqualToThreshold"
  evaluation_periods  = "2"
  metric_name         = "CPUUtilization"
  namespace           = "AWS/EC2"
  period              = "120"
  statistic           = "Average"
  threshold           = "60"
  alarm_description   = "This metric monitors ec2 cpu utilization"

  dimensions {
    InstanceId = "${element(aws_instance.web.*.id, count.index)}"
  }
}

resource "aws_cloudwatch_metric_alarm" "status_check" {
  count               = 3
  alarm_name          = "${element(split(",", join(",", aws_instance.web.*.id)), count.index)}"
  comparison_operator = "GreaterThanOrEqualToThreshold"
  evaluation_periods  = "2"
  metric_name         = "StatusCheckFailed"
  namespace           = "AWS/EC2"
  period              = "120"
  statistic           = "Average"
  threshold           = "1"
  alarm_description   = "This metric monitors ec2 status check."

  dimensions {
    InstanceId = "${element(aws_instance.web.*.id, count.index)}"
  }
}

预期行为：我希望每个实例在每个区域都应附加 2 个以上的警报。

错误行为：它使用实例在每个区域创建并附加 3 个警报。

北弗吉尼亚地区 - 一个用于 CPU 两个用于 StatusCheck。
对于俄勒冈地区 - 两个用于 StatusCheck，一个用于 CPU 利用率。

每次我应用它都会创建警报，反之亦然。

我遇到以下错误，如果我在更新警报时等待 2 分钟或者如果我使用 terraform apply -parallelism=1

，该错误将得到解决

错误：

4 error(s) occurred:

* module.north-virginia.aws_cloudwatch_metric_alarm.status_check[0]: 1 error(s) occurred:

* aws_cloudwatch_metric_alarm.status_check.0: Creating metric alarm failed: ValidationError: A separate request to update this alarm is in progress.
status code: 400, request id: ea6c4502-dede-11e8-9262-c55251d6673a
* module.north-virginia.aws_cloudwatch_metric_alarm.cpu_utilization[1]: 1 error(s) occurred:

* aws_cloudwatch_metric_alarm.cpu_utilization.1: Creating metric alarm failed: ValidationError: A separate request to update this alarm is in progress.
status code: 400, request id: ea6c6c09-dede-11e8-a13f-bbb86ff53045
* module.oregon.aws_cloudwatch_metric_alarm.status_check[1]: 1 error(s) occurred:

* aws_cloudwatch_metric_alarm.status_check.1: Creating metric alarm failed: ValidationError: A separate request to update this alarm is in progress.
status code: 400, request id: ed198a56-dede-11e8-b95a-9d366b9f2e85
* module.oregon.aws_cloudwatch_metric_alarm.cpu_utilization[3]: 1 error(s) occurred:

* aws_cloudwatch_metric_alarm.cpu_utilization.3: Creating metric alarm failed: ValidationError: A separate request to update this alarm is in progress.
status code: 400, request id: ed193c4d-dede-11e8-9c63-21cde1551122

我在这里遗漏的任何想法或任何约定将不胜感激。

Answer 1

首先，我会通过 removing/commenting 来简化您的测试 module "oregon"。一旦你得到 virginia 一个正确的，然后重新添加它。

其次，我将切换您模块中的代码以计算计数作为 var.az 的长度。这应该针对您拥有的 3 个资源：aws_instance 和 2 个 CloudWatch 警报。例如：

count = "${length(var.az)}"

这样您就可以在调用模块的代码中更改 AZ 的数量，并动态更改创建的实例的数量。

第三，name 您为 CloudWatch 设置的警报看起来是一样的。尝试区分它们。例如：

alarm_name = "${element(split(",", join(",", aws_instance.web.*.id)), count.index)}-cpu-util"
alarm_name = "${element(split(",", join(",", aws_instance.web.*.id)), count.index)}-status-check"

PS> 在测试之间，确保您已清除所有可能已创建的资源以确保您是运行一个干净的测试。

Terraform：为多个实例创建 CloudWatch 警报时出错

Terraform: Error while creating CloudWatch alarm for multiple instances

amazon-cloudwatch

terraform