Add low severity alerts for cloud sql metrics

terraform/gcp/modules/monitoring/infra/alerts.tf

@@ -66,6 +66,45 @@ resource "google_monitoring_alert_policy" "ssl_cert_expiry_alert" {
 
 ### Cloud SQL Alerts
 
+# Cloud SQL Database CPU Utilization > 80%
+resource "google_monitoring_alert_policy" "cloud_sql_cpu_utilization" {
+  # In the absence of data, incident will auto-close in 7 days
+  alert_strategy {
+    auto_close = "604800s"
+  }
+  combiner = "OR"
+
+  conditions {
+    condition_threshold {
+      aggregations {
+        alignment_period   = "300s"
+        per_series_aligner = "ALIGN_MEAN"
+      }
+      comparison      = "COMPARISON_GT"
+      duration        = "0s"
+      filter          = "metric.type=\"cloudsql.googleapis.com/database/cpu/utilization\" resource.type=\"cloudsql_database\""
+      threshold_value = "0.9"
+      trigger {
+        count   = "1"
+        percent = "0"
+      }
+    }
+    display_name = "Cloud SQL Database - CPU Utilization [MEAN]"
+  }
+  display_name = "Cloud SQL Database CPU Utilization > 80%"
+  documentation {
+    content   = "Cloud SQL Database CPU Utilization is >80%. Please increase CPU capacity."
+    mime_type = "text/markdown"
+  }
+  enabled               = "true"
+  notification_channels = local.notification_channels
+  project               = var.project_id
+
+  user_labels = {
+    severity = "warning"
+  }
+}
+
 # Cloud SQL Database CPU Utilization > 90%
 resource "google_monitoring_alert_policy" "cloud_sql_cpu_utilization" {
   # In the absence of data, incident will auto-close in 7 days
@@ -101,6 +140,53 @@ resource "google_monitoring_alert_policy" "cloud_sql_cpu_utilization" {
   project               = var.project_id
 }
 
+
+# Cloud SQL Database Memory Utilization > 80%
+resource "google_monitoring_alert_policy" "cloud_sql_memory_utilization" {
+  # In the absence of data, incident will auto-close in 7 days
+  alert_strategy {
+    auto_close = "604800s"
+  }
+
+  combiner = "OR"
+
+  conditions {
+    condition_threshold {
+      aggregations {
+        alignment_period   = "300s"
+        per_series_aligner = "ALIGN_MEAN"
+      }
+
+      comparison      = "COMPARISON_GT"
+      duration        = "0s"
+      filter          = "metric.type=\"cloudsql.googleapis.com/database/memory/utilization\" resource.type=\"cloudsql_database\""
+      threshold_value = "0.9"
+
+      trigger {
+        count   = "1"
+        percent = "0"
+      }
+    }
+
+    display_name = "Cloud SQL Database - Memory utilization [MEAN]"
+  }
+
+  display_name = "Cloud SQL Database Memory Utilization > 80%"
+
+  documentation {
+    content   = "Cloud SQL Database Memory Utilization is >80%. Please increase memory capacity."
+    mime_type = "text/markdown"
+  }
+
+  enabled               = "true"
+  notification_channels = local.notification_channels
+  project               = var.project_id
+
+  user_labels = {
+    severity = "warning"
+  }
+}
+
 # Cloud SQL Database Memory Utilization > 90%
 resource "google_monitoring_alert_policy" "cloud_sql_memory_utilization" {
   # In the absence of data, incident will auto-close in 7 days
@@ -143,6 +229,54 @@ resource "google_monitoring_alert_policy" "cloud_sql_memory_utilization" {
   project               = var.project_id
 }
 
+# Cloud SQL Database Disk has < 30GiB Free
+resource "google_monitoring_alert_policy" "cloud_sql_disk_utilization" {
+  # In the absence of data, incident will auto-close in 7 days
+  alert_strategy {
+    auto_close = "604800s"
+  }
+
+  combiner = "OR"
+
+  # Disk has less that 30GiB free
+  conditions {
+    # < 30GiB disk space free
+    condition_monitoring_query_language {
+      duration = "0s"
+      query    = <<-EOT
+        fetch cloudsql_database
+        | { bytes: metric 'cloudsql.googleapis.com/database/disk/bytes_used'
+          ; quota: metric 'cloudsql.googleapis.com/database/disk/quota'
+          ; utilization: metric 'cloudsql.googleapis.com/database/disk/utilization' }
+        | join
+        | group_by 5m, [q_mean: mean(value.quota), b_mean: mean(value.bytes_used), u_mean: mean(value.utilization)]
+        | every 5m
+        | group_by [resource.database_id], [free_space: sub(mean(q_mean), mean(b_mean)), u: mean(u_mean)]
+        | condition and(free_space < 30 'GiBy', u > 0.98)
+      EOT
+      trigger {
+        count   = "1"
+        percent = "0"
+      }
+    }
+
+    display_name = "Cloud SQL Database - Disk free space and utilization [MEAN]"
+  }
+
+  display_name = "Cloud SQL Database Disk has < 30GiB Free and Utilization > 98%"
+
+  documentation {
+    content   = "Cloud SQL disk has less than 30GiB free space remaining. Please increase capacity. Note that autoresize should be enabled for the database. Ensure there is no issue with the autoresize process."
+    mime_type = "text/markdown"
+  }
+
+  enabled               = "true"
+  notification_channels = local.notification_channels
+  project               = var.project_id
+
+  severity = "warning"
+}
+
 # Cloud SQL Database Disk has < 20GiB Free
 resource "google_monitoring_alert_policy" "cloud_sql_disk_utilization" {
   # In the absence of data, incident will auto-close in 7 days