Merge pull request #2358 from newrelic/release

Release 20240404

alert-policies/amazon-timestream/MagneticStoreRejectedUploadSystemFailures.yml

@@ -1,7 +1,7 @@
-name: High MagneticStoreRejectedUploadSystemFailures
+name: High Magnetic Store Rejected Upload System Failures
 
 description: |+
-  This alert is triggered when the MagneticStoreRejectedUploadSystemFailures is above 100 in 10 minutes.
+  This alert is triggered when the Magnetic Store Rejected Upload System Failures is above 100 in 10 minutes.
 
 type: STATIC
 nrql:

alert-policies/amazon-timestream/MagneticStoreRejectedUploadUserFailures.yml

@@ -1,7 +1,7 @@
-name: High MagneticStoreRejectedUploadUserFailures
+name: High Magnetic Store Rejected Upload User Failures
 
 description: |+
-  This alert is triggered when the MagneticStoreRejectedUploadUserFailures is above 100 in 10 minutes.
+  This alert is triggered when the Magnetic Store Rejected Upload User Failures is above 100 in 10 minutes.
 
 type: STATIC
 nrql:

alert-policies/amazon-timestream/SystemErrors.yml

@@ -1,4 +1,4 @@
-name: High SystemErrors
+name: High System Errors
 
 description: |+
   This alert is triggered when the system errors is above 100 in 10 minutes.

alert-policies/amazon-timestream/UserErrors.yml

@@ -1,4 +1,4 @@
-name: High UserErrors
+name: High User Errors
 
 description: |+
   This alert is triggered when the user errors is above 100 in 10 minutes.

alert-policies/apache-druid/failedQuery.yml

@@ -0,0 +1,26 @@
+name: Failed Query
+
+description: |+
+  This alert is triggered if the number of failed queries exceeds 5 for 5 minutes.
+type: STATIC
+nrql:
+  query: "SELECT average(druid_query_failed_count_total) FROM Metric"
+
+# Function used to aggregate the NRQL query value(s) for comparison to the terms.threshold (Default: SINGLE_VALUE)
+valueFunction: SINGLE_VALUE
+
+# List of Critical and Warning thresholds for the condition
+terms:
+  - priority: CRITICAL
+    # 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])
+violationTimeLimitSeconds: 86400

alert-policies/apache-druid/jvmMemoryUsage.yml

@@ -0,0 +1,35 @@
+name: Average Memory Usage (%)
+
+description: |+
+  This alert is triggered if memory usage exceeds 90% for 5 minutes.
+type: STATIC
+nrql:
+  query: "SELECT average(druid_jvm_mem_used) / average(druid_jvm_mem_max) * 100 as 'utilization' FROM Metric"
+
+# Function used to aggregate the NRQL query value(s) for comparison to the terms.threshold (Default: SINGLE_VALUE)
+valueFunction: SINGLE_VALUE
+
+# List of Critical and Warning thresholds for the condition
+terms:
+  - priority: CRITICAL
+    # Operator used to compare against the threshold.
+    operator: ABOVE
+    # Value that triggers a violation
+    threshold: 95
+    # Time in seconds; 120 - 3600
+    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: 90
+    # 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])
+violationTimeLimitSeconds: 86400