Merge pull request #1992 from RamanaReddy8801/NR-143043

NR-143043: Updated Nodejs dashboards and alerts

alert-policies/node-js/HighCpuUtilization.yml

@@ -1,7 +1,7 @@
-name: High CPU Utilization
+name: CPU Utilization
 
 description: |+
-  This alert is triggered when the CPU Utilization is above 90%.
+  This alert is triggered when the CPU utilization exceeds 90% for 5 minutes.
 
 type: STATIC
 nrql:
@@ -22,6 +22,15 @@ terms:
     thresholdDuration: 300
     # How many data points must be in violation for the duration
     thresholdOccurrences: ALL
+  - priority: WARNING
+    # Operator used to compare against the threshold.
+    operator: ABOVE
+    # Value that triggers a violation.
+    threshold: 85
+    # Time in seconds; 120 - 3600
+    thresholdDuration: 300
+    # How many data points must be in violation for the duration?
+    thresholdOccurrences: ALL
 
 # Duration after which a violation automatically closes
 # Time in seconds; 300 - 2592000 (Default: 86400 [1 day])

alert-policies/node-js/HighMemoryUsage.yml

@@ -1,7 +1,7 @@
 name: Memory Usage
 
 description: |+
-  This alert is triggered when Memory usage is above 90%
+  This alert is triggered when the Memory usage exceeds 90% for 5 minutes.
 type: STATIC
 nrql:
   query: "SELECT average(memoryUsedBytes/memoryTotalBytes) * 100 FROM SystemSample"
@@ -20,6 +20,15 @@ terms:
     thresholdDuration: 300
     # How many data points must be in violation for the duration
     thresholdOccurrences: ALL
+  - priority: WARNING
+    # Operator used to compare against the threshold.
+    operator: ABOVE
+    # Value that triggers a violation.
+    threshold: 85
+    # Time in seconds; 120 - 3600
+    thresholdDuration: 300
+    # How many data points must be in violation for the duration?
+    thresholdOccurrences: ALL
 
 # Duration after which a violation automatically closes
 # Time in seconds; 300 - 2592000 (Default: 86400 [1 day])

alert-policies/node-js/LowApdexScore.yml

@@ -1,10 +1,10 @@
 name: Apdex Score
 
 description: |+
-  This alert is triggered when the Apdex score is below 0.5 for 5 minutes
+  This alert is triggered when the Apdex score is below 0.5 for 5 minutes.
 type: STATIC
 nrql:
-  query: "SELECT apdex(duration, t: 0.5) FROM Transaction WHERE appName like '%'"
+  query: "SELECT apdex(duration, t: 0.5) FROM Transaction FACET appName"
 
 # Function used to aggregate the NRQL query value(s) for comparison to the terms.threshold (Default: SINGLE_VALUE)
 valueFunction: SINGLE_VALUE

alert-policies/node-js/TransactionErrors.yml

@@ -1,14 +1,13 @@
-name: Transaction Errors
+name: Web Transaction Errors
 
 description: |+
-  This alert is triggered when the the transactions fail more than 10% of the time in 5 minutes.
+  This alert is triggered when the web transactions fail exceeds 10% for 5 minutes.
 
 type: STATIC
 nrql:
   query:
     "SELECT count(apm.service.error.count) / count(apm.service.transaction.duration) * 100
-    as 'Web errors' FROM Metric WHERE appName like '%'
-    AND (transactionType = 'Web')"
+    as 'Web errors' FROM Metric WHERE appName LIKE '%' AND (transactionType = 'Web') FACET appName"
 
 # Function used to aggregate the NRQL query value(s) for comparison to the terms.threshold (Default: SINGLE_VALUE)
 valueFunction: SINGLE_VALUE
@@ -24,6 +23,15 @@ terms:
     thresholdDuration: 300
     # How many data points must be in violation for the duration
     thresholdOccurrences: ALL
+  - priority: WARNING
+    # Operator used to compare against the threshold.
+    operator: ABOVE
+    # Value that triggers a violation.
+    threshold: 5 
+    # Time in seconds; 120 - 3600
+    thresholdDuration: 300
+    # How many data points must be in violation for the duration?
+    thresholdOccurrences: ALL
 
 # Duration after which a violation automatically closes
 # Time in seconds; 300 - 2592000 (Default: 86400 [1 day])