UPD: added EC2 monitoring with cwagent

modules/ManagementGovernance/Monitoring.EC2/main.tf

@@ -67,3 +67,145 @@ resource "aws_cloudwatch_metric_alarm" "ec2-CPUUtilization" {
     ignore_changes = [tags]
   }
 }
+
+# cwagent metrics
+data "aws_instance" "ec2-instance" {
+  instance_id = var.ec2-instance-id
+}
+
+resource "aws_cloudwatch_metric_alarm" "ec2-mem_free" {
+  alarm_name                = "${var.cw-alarm-prefix}:EC2:mem_free:${var.ec2-instance-id}"
+  comparison_operator       = "LessThanThreshold"
+  evaluation_periods        = "2"
+  metric_name               = "mem_free"
+  period                    = "900"
+  statistic                 = "Average"
+  threshold                 = var.threshold-mem_free
+  alarm_description         = "EC2:mem_free"
+  namespace                 = "CWAgent"
+  insufficient_data_actions = []
+  actions_enabled           = var.actions-enabled
+  alarm_actions             = [var.alarm-actions-standard]
+  ok_actions                = [var.alarm-actions-standard]
+  treat_missing_data        = "notBreaching"
+  dimensions = {
+    InstanceId   = var.ec2-instance-id
+    ImageId      = data.aws_instance.ec2-instance.ami
+    InstanceType = data.aws_instance.ec2-instance.instance_type
+  }
+  tags = var.default-tags
+  lifecycle {
+    ignore_changes = [tags]
+  }
+}
+
+resource "aws_cloudwatch_metric_alarm" "ec2-swap_free" {
+  # zero is fine as most ec2 instances are deployed without any swap
+  alarm_name          = "${var.cw-alarm-prefix}:EC2:swap_free:${var.ec2-instance-id}"
+  comparison_operator = "LessThanThreshold"
+  evaluation_periods  = "2"
+  # metric_name               = "swap_free"
+  # period                    = "900"
+  # statistic                 = "Average"
+  threshold         = var.threshold-swap_free
+  alarm_description = "EC2:swap_free"
+  # namespace                 = "CWAgent"
+  insufficient_data_actions = []
+  actions_enabled           = var.actions-enabled
+  alarm_actions             = [var.alarm-actions-standard]
+  ok_actions                = [var.alarm-actions-standard]
+  treat_missing_data        = "notBreaching"
+  metric_query {
+    id = "m1"
+    metric {
+      metric_name = "swap_free"
+      namespace   = "CWAgent"
+      period      = 900
+      stat        = "Average"
+      dimensions = {
+        InstanceId   = var.ec2-instance-id
+        ImageId      = data.aws_instance.ec2-instance.ami
+        InstanceType = data.aws_instance.ec2-instance.instance_type
+      }
+    }
+  }
+  metric_query {
+    id          = "e1"
+    expression  = "IF(m1==0, ${var.threshold-swap_free}, m1)"
+    label       = "swap_free_if_not_zero"
+    return_data = "true"
+  }
+
+  #dimensions = {
+  #  InstanceId = var.ec2-instance-id
+  #  ImageId = data.aws_instance.ec2-instance.ami
+  #  InstanceType = data.aws_instance.ec2-instance.instance_type
+  #}
+  tags = var.default-tags
+  lifecycle {
+    ignore_changes = [tags]
+  }
+}
+
+resource "aws_cloudwatch_metric_alarm" "ec2-disk_free" {
+  alarm_name                = "${var.cw-alarm-prefix}:EC2:disk_free:${var.ec2-instance-id}"
+  comparison_operator       = "LessThanThreshold"
+  evaluation_periods        = "2"
+  metric_name               = "disk_free"
+  period                    = "900"
+  statistic                 = "Average"
+  threshold                 = var.threshold-disk_free
+  alarm_description         = "EC2:disk_free"
+  namespace                 = "CWAgent"
+  insufficient_data_actions = []
+  actions_enabled           = var.actions-enabled
+  alarm_actions             = [var.alarm-actions-urgent]
+  ok_actions                = [var.alarm-actions-urgent]
+  treat_missing_data        = "notBreaching"
+  dimensions = {
+    InstanceId   = var.ec2-instance-id
+    ImageId      = data.aws_instance.ec2-instance.ami
+    InstanceType = data.aws_instance.ec2-instance.instance_type
+    # The following will not work. AWS expects the device name from OS, not from AWS perspective
+    # device = one(data.aws_instance.ec2-instance.root_block_device[*].device_name)
+    # device = ""
+    fstype = "xfs"
+    path   = "/"
+  }
+  tags = var.default-tags
+  lifecycle {
+    ignore_changes = [tags]
+  }
+}
+
+
+resource "aws_cloudwatch_metric_alarm" "ec2-disk_inodes_free" {
+  alarm_name                = "${var.cw-alarm-prefix}:EC2:disk_inodes_free:${var.ec2-instance-id}"
+  comparison_operator       = "LessThanThreshold"
+  evaluation_periods        = "2"
+  metric_name               = "disk_inodes_free"
+  period                    = "300"
+  statistic                 = "Average"
+  threshold                 = var.threshold-disk_inodes_free
+  alarm_description         = "EC2:disk_inodes_free"
+  namespace                 = "CWAgent"
+  insufficient_data_actions = []
+  actions_enabled           = var.actions-enabled
+  alarm_actions             = [var.alarm-actions-urgent]
+  ok_actions                = [var.alarm-actions-urgent]
+  treat_missing_data        = "notBreaching"
+  dimensions = {
+    InstanceId   = var.ec2-instance-id
+    ImageId      = data.aws_instance.ec2-instance.ami
+    InstanceType = data.aws_instance.ec2-instance.instance_type
+    # The following will not work. AWS expects the device name from OS, not from AWS perspective
+    # device = one(data.aws_instance.ec2-instance.root_block_device[*].device_name)
+    # device = ""
+    fstype = "xfs"
+    path   = "/"
+  }
+  tags = var.default-tags
+  lifecycle {
+    ignore_changes = [tags]
+  }
+}

modules/ManagementGovernance/Monitoring.EC2/variables.tf

@@ -1,22 +1,26 @@
-variable cw-alarm-prefix {}
+variable "cw-alarm-prefix" {}
-variable actions-enabled {}
+variable "actions-enabled" {}
-variable ec2-instance-id {}
+variable "ec2-instance-id" {}
-variable alarm-actions-urgent {
+variable "alarm-actions-urgent" {
   type    = string
   default = "arn:aws:sns:ap-east-1:843733946244:rackspace-support-urgent"
 }
-variable alarm-actions-emergency {
+variable "alarm-actions-emergency" {
   type    = string
   default = "arn:aws:sns:ap-east-1:843733946244:rackspace-support-emergency"
 }
-variable alarm-actions-standard {
+variable "alarm-actions-standard" {
   type    = string
   default = "arn:aws:sns:ap-east-1:843733946244:rackspace-support-standard"
 }
-variable alarm-actions-general {
+variable "alarm-actions-general" {
   type    = string
   default = "arn:aws:sns:ap-east-1:843733946244:rackspace-support"
 }
-variable default-tags {}
+variable "default-tags" {}
 
-variable threshold-CPUUtilization {}
+variable "threshold-CPUUtilization" {}
+variable "threshold-mem_free" {}
+variable "threshold-swap_free" {}
+variable "threshold-disk_free" {}
+variable "threshold-disk_inodes_free" {}

modules/ManagementGovernance/Monitoring.NLB/main.tf

@@ -17,7 +17,7 @@ module "nlb-targetgroups" {
 resource "aws_cloudwatch_metric_alarm" "nlb-HealthyHostCount" {
   for_each                  = module.nlb-targetgroups.result-set
   alarm_name                = "${var.cw-alarm-prefix}:NLBTG:HealthyHostCount:${split(":", each.value)[5]}"
-  comparison_operator       = "LessThanThreshold"
+  comparison_operator       = "LessThanOrEqualToThreshold"
   evaluation_periods        = "1"
   metric_name               = "HealthyHostCount"
   period                    = "300"

modules/ManagementGovernance/Monitoring.RDS/main.tf

@@ -70,7 +70,7 @@ resource "aws_cloudwatch_metric_alarm" "rds-memory" {
 resource "aws_cloudwatch_metric_alarm" "rds-DiskQueueDepth" {
   alarm_name                = "${var.cw-alarm-prefix}:RDS:DiskQueueDepth:${var.rds-instance-name}"
   comparison_operator       = "GreaterThanThreshold"
-  evaluation_periods        = "1"
+  evaluation_periods        = "2"
   metric_name               = "DiskQueueDepth"
   period                    = "300"
   statistic                 = "Average"