Question

我希望创建具有多个警报和操作的多个指标过滤器。到目前为止，这就是我所拥有的。尝试参数化“ alarm_actions”部分而没有手动手动键入警报操作ARN时，我不断收到错误消息：

# variables.tf

# 2 Log groups - group1 & group2 
variable "log_groups" {
  type  = list(object({})
  
  default = [
    {
      name = "group1"
      retention_in_days = 10
    },
    {
      name = "group2"
      retention_in_days = 7
    },
  ]
}

# Metric filters - 2 for group1 & 1 for group2
variable "metric_filters" {
  type  = list(object({}))

  default = [
    # Group1 - Metric filters
    {
      name    = "group1-filter1"
      pattern = "abc"
      group   = "group1"
    },
    {
      name    = "group1-filter2"
      pattern = "xyz"
      group   = "group1"
    },

    # Group2 - Metric filters
    {
      name    = "group2-filter1"
      pattern = "abcdef"
      group   = "group2"
    },
  ]
}

# Alarms - with variable number of actions
variable "alarms" {
  type  = list(object({}))

  default = [
    {
      name          = "group1-filter1-alarm"
      period        = 120
      alarm_actions = ["action1"]
    },
    {
      name          = "group1-filter2-alarm"
      period        = 300
      alarm_actions = ["action1", "action2"]
    },
    {
      name          = "group2-filter1-alarm"
      period        = 60
      alarm_actions = []
    },
  ]
}

现在很简单地为日志组和度量过滤器创建资源（尽管我希望避免在变量中重复日志组名称，但我可以接受）

resource "aws_cloudwatch_log_group" "log_groups" {
  for_each          = { for x in var.log_groups : x.name => x }
  name              = each.value.name
  retention_in_days = each.value.retention_in_days
}

resource "aws_cloudwatch_log_metric_filter" "metric_filters" {
  for_each       = { for x in var.metric_filters : x.name => x }
  name           = each.value.name
  pattern        = each.value.pattern
  log_group_name = each.value.group

  metric_transformation {
    name          = each.value.name
    namespace     = "xyz-namespace"
    value         = "1"
  }
}

问题在于创建Alarm_actions而没有重复我的代码。如何将SNS通知的ARN一次声明为变量，然后在变量中重用它们（我猜terraform不允许变量内部包含变量）。或者，我可以使用data功能通过名称获取SNS，但是在alarm资源块中嵌套的alert_actions循环for失败：

data "aws_sns_topic" "action1" {
  name = "action1"
}

data "aws_sns_topic" "action2" {
  name = "action2"
}

resource "aws_cloudwatch_metric_alarm" "alarms" {
  for_each            = { for x in var.alarms : x.name => x }
  alarm_name          = each.value.name
  comparison_operator = "GreaterThan"
  evaluation_periods  = "1"
  metric_name         = each.value.name
  namespace           = "xyz-namespace"
  period              = each.value.period
  statistic           = "Sum"
  threshold           = "1"

  alarm_actions = [for a in each.value.alarm_actions : "data.aws_sns_topic.${a}.arn"]
}

我得到的错误是

Error: "alarm_actions.0" does not match EC2 automation ARN ("^arn:[\\w-]+:automate:[\\w-]+:ec2:(reboot|recover|stop|terminate)$"): "data.aws_sns_topic.action1.arn"

即看起来表达式data.aws_sns_topic.${a}.arn获得${a}的值，然后将其视为字符串

Answer 1

如果愿意，可以串联ARN：
"arn:aws:sns:{region}:{account}:{name}"
应该很简单，您只需要再添加一些项目
variable "alarms" { type = list(object({ name = string period = number alarm_actions = list(string) })) default = [ { name = "group1-filter1-alarm" period = 120 alarm_actions = ["action1"] },{ name = "group1-filter2-alarm" period = 300 alarm_actions = ["action1", "action2"] }, ] } data "aws_sns_topic" "actions" { for_each = toset(flatten([for x in var.alarms : x.alarm_actions])) name = each.value } resource "aws_cloudwatch_metric_alarm" "alarms" { for_each = { for x in var.alarms : x.name => x } alarm_name = each.value.name comparison_operator = "GreaterThanThreshold" evaluation_periods = "1" metric_name = each.value.name namespace = "xyz-namespace" period = each.value.period statistic = "Sum" threshold = "1" alarm_actions = [for a in each.value.alarm_actions : data.aws_sns_topic.actions[a].arn] } output "actions" { value = data.aws_sns_topic.actions }

另一种选择是对数据进行循环，如下所示：

Terraform v0.12.24
 + provider.aws v2.54.0

经过测试可用于：

FrameLayout

Terraform：使用变量创建多个度量过滤器和警报

1 个答案: