fix(metrics): Avoid 0 value when instance is not available

qonto · Oct 18, 2023 · ed61550 · ed61550
1 parent 59551d4
commit ed61550
Show file tree

Hide file tree

Showing 3 changed files with 103 additions and 53 deletions.
diff --git a/internal/app/cloudwatch/rds.go b/internal/app/cloudwatch/rds.go
@@ -27,55 +27,55 @@ type CloudWatchMetrics struct {
 }
 
 type RdsMetrics struct {
-	CPUUtilization            float64
-	DBLoad                    float64
-	DBLoadCPU                 float64
-	DBLoadNonCPU              float64
-	DatabaseConnections       float64
-	FreeStorageSpace          float64
-	FreeableMemory            float64
-	MaximumUsedTransactionIDs float64
-	ReadIOPS                  float64
-	ReadThroughput            float64
-	ReplicaLag                float64
-	ReplicationSlotDiskUsage  float64
-	SwapUsage                 float64
-	WriteIOPS                 float64
-	WriteThroughput           float64
+	CPUUtilization            *float64
+	DBLoad                    *float64
+	DBLoadCPU                 *float64
+	DBLoadNonCPU              *float64
+	DatabaseConnections       *float64
+	FreeStorageSpace          *float64
+	FreeableMemory            *float64
+	MaximumUsedTransactionIDs *float64
+	ReadIOPS                  *float64
+	ReadThroughput            *float64
+	ReplicaLag                *float64
+	ReplicationSlotDiskUsage  *float64
+	SwapUsage                 *float64
+	WriteIOPS                 *float64
+	WriteThroughput           *float64
 }
 
 func (m *RdsMetrics) Update(field string, value float64) error {
 	switch field {
 	case "DBLoad":
-		m.DBLoad = value
+		m.DBLoad = &value
 	case "DBLoadCPU":
-		m.DBLoadCPU = value
+		m.DBLoadCPU = &value
 	case "DBLoadNonCPU":
-		m.DBLoadNonCPU = value
+		m.DBLoadNonCPU = &value
 	case "CPUUtilization":
-		m.CPUUtilization = value
+		m.CPUUtilization = &value
 	case "DatabaseConnections":
-		m.DatabaseConnections = value
+		m.DatabaseConnections = &value
 	case "FreeStorageSpace":
-		m.FreeStorageSpace = value
+		m.FreeStorageSpace = &value
 	case "FreeableMemory":
-		m.FreeableMemory = value
+		m.FreeableMemory = &value
 	case "SwapUsage":
-		m.SwapUsage = value
+		m.SwapUsage = &value
 	case "WriteIOPS":
-		m.WriteIOPS = value
+		m.WriteIOPS = &value
 	case "ReadIOPS":
-		m.ReadIOPS = value
+		m.ReadIOPS = &value
 	case "ReplicaLag":
-		m.ReplicaLag = value
+		m.ReplicaLag = &value
 	case "ReplicationSlotDiskUsage":
-		m.ReplicationSlotDiskUsage = value
+		m.ReplicationSlotDiskUsage = &value
 	case "MaximumUsedTransactionIDs":
-		m.MaximumUsedTransactionIDs = value
+		m.MaximumUsedTransactionIDs = &value
 	case "ReadThroughput":
-		m.ReadThroughput = value
+		m.ReadThroughput = &value
 	case "WriteThroughput":
-		m.WriteThroughput = value
+		m.WriteThroughput = &value
 	default:
 		return fmt.Errorf("can't process '%s' metrics: %w", field, errUnknownMetric)
 	}

diff --git a/internal/app/exporter/exporter.go b/internal/app/exporter/exporter.go
@@ -389,9 +389,12 @@ func (c *rdsCollector) Collect(ch chan<- prometheus.Metric) {
 	// RDS metrics
 	ch <- prometheus.MustNewConstMetric(c.apiCall, prometheus.CounterValue, c.counters.rdsAPIcalls, c.awsAccountID, c.awsRegion, "rds")
 	for dbidentifier, instance := range c.metrics.rds.Instances {
+		if instance.LogFilesSize != nil {
+			ch <- prometheus.MustNewConstMetric(c.logFilesSize, prometheus.GaugeValue, float64(*instance.LogFilesSize), c.awsAccountID, c.awsRegion, dbidentifier)
+		}
+
 		ch <- prometheus.MustNewConstMetric(c.allocatedStorage, prometheus.GaugeValue, float64(instance.AllocatedStorage), c.awsAccountID, c.awsRegion, dbidentifier)
 		ch <- prometheus.MustNewConstMetric(c.information, prometheus.GaugeValue, 1, c.awsAccountID, c.awsRegion, dbidentifier, instance.DbiResourceID, instance.DBInstanceClass, instance.Engine, instance.EngineVersion, instance.StorageType, strconv.FormatBool(instance.MultiAZ), strconv.FormatBool(instance.DeletionProtection), instance.Role, instance.SourceDBInstanceIdentifier, strconv.FormatBool(instance.PendingModifiedValues), instance.PendingMaintenanceAction, strconv.FormatBool(instance.PerformanceInsightsEnabled))
-		ch <- prometheus.MustNewConstMetric(c.logFilesSize, prometheus.GaugeValue, float64(instance.LogFilesSize), c.awsAccountID, c.awsRegion, dbidentifier)
 		ch <- prometheus.MustNewConstMetric(c.maxAllocatedStorage, prometheus.GaugeValue, float64(instance.MaxAllocatedStorage), c.awsAccountID, c.awsRegion, dbidentifier)
 		ch <- prometheus.MustNewConstMetric(c.maxIops, prometheus.GaugeValue, float64(instance.MaxIops), c.awsAccountID, c.awsRegion, dbidentifier)
 		ch <- prometheus.MustNewConstMetric(c.status, prometheus.GaugeValue, float64(instance.Status), c.awsAccountID, c.awsRegion, dbidentifier)
@@ -402,21 +405,65 @@ func (c *rdsCollector) Collect(ch chan<- prometheus.Metric) {
 	// Cloudwatch metrics
 	ch <- prometheus.MustNewConstMetric(c.apiCall, prometheus.CounterValue, c.counters.cloudwatchAPICalls, c.awsAccountID, c.awsRegion, "cloudwatch")
 	for dbidentifier, instance := range c.metrics.cloudwatchInstances.Instances {
-		ch <- prometheus.MustNewConstMetric(c.databaseConnections, prometheus.GaugeValue, instance.DatabaseConnections, c.awsAccountID, c.awsRegion, dbidentifier)
-		ch <- prometheus.MustNewConstMetric(c.freeStorageSpace, prometheus.GaugeValue, instance.FreeStorageSpace, c.awsAccountID, c.awsRegion, dbidentifier)
-		ch <- prometheus.MustNewConstMetric(c.freeableMemory, prometheus.GaugeValue, instance.FreeableMemory, c.awsAccountID, c.awsRegion, dbidentifier)
-		ch <- prometheus.MustNewConstMetric(c.maximumUsedTransactionIDs, prometheus.GaugeValue, instance.MaximumUsedTransactionIDs, c.awsAccountID, c.awsRegion, dbidentifier)
-		ch <- prometheus.MustNewConstMetric(c.readThroughput, prometheus.GaugeValue, instance.ReadThroughput, c.awsAccountID, c.awsRegion, dbidentifier)
-		ch <- prometheus.MustNewConstMetric(c.replicaLag, prometheus.GaugeValue, instance.ReplicaLag, c.awsAccountID, c.awsRegion, dbidentifier)
-		ch <- prometheus.MustNewConstMetric(c.replicationSlotDiskUsage, prometheus.GaugeValue, instance.ReplicationSlotDiskUsage, c.awsAccountID, c.awsRegion, dbidentifier)
-		ch <- prometheus.MustNewConstMetric(c.swapUsage, prometheus.GaugeValue, instance.SwapUsage, c.awsAccountID, c.awsRegion, dbidentifier)
-		ch <- prometheus.MustNewConstMetric(c.readIOPS, prometheus.GaugeValue, instance.ReadIOPS, c.awsAccountID, c.awsRegion, dbidentifier)
-		ch <- prometheus.MustNewConstMetric(c.writeIOPS, prometheus.GaugeValue, instance.WriteIOPS, c.awsAccountID, c.awsRegion, dbidentifier)
-		ch <- prometheus.MustNewConstMetric(c.writeThroughput, prometheus.GaugeValue, instance.WriteThroughput, c.awsAccountID, c.awsRegion, dbidentifier)
-		ch <- prometheus.MustNewConstMetric(c.DBLoad, prometheus.GaugeValue, instance.DBLoad, c.awsAccountID, c.awsRegion, dbidentifier)
-		ch <- prometheus.MustNewConstMetric(c.cpuUtilisation, prometheus.GaugeValue, instance.CPUUtilization, c.awsAccountID, c.awsRegion, dbidentifier)
-		ch <- prometheus.MustNewConstMetric(c.dBLoadCPU, prometheus.GaugeValue, instance.DBLoadCPU, c.awsAccountID, c.awsRegion, dbidentifier)
-		ch <- prometheus.MustNewConstMetric(c.dBLoadNonCPU, prometheus.GaugeValue, instance.DBLoadNonCPU, c.awsAccountID, c.awsRegion, dbidentifier)
+		if instance.DatabaseConnections != nil {
+			ch <- prometheus.MustNewConstMetric(c.databaseConnections, prometheus.GaugeValue, *instance.DatabaseConnections, c.awsAccountID, c.awsRegion, dbidentifier)
+		}
+
+		if instance.FreeStorageSpace != nil {
+			ch <- prometheus.MustNewConstMetric(c.freeStorageSpace, prometheus.GaugeValue, *instance.FreeStorageSpace, c.awsAccountID, c.awsRegion, dbidentifier)
+		}
+
+		if instance.FreeableMemory != nil {
+			ch <- prometheus.MustNewConstMetric(c.freeableMemory, prometheus.GaugeValue, *instance.FreeableMemory, c.awsAccountID, c.awsRegion, dbidentifier)
+		}
+
+		if instance.MaximumUsedTransactionIDs != nil {
+			ch <- prometheus.MustNewConstMetric(c.maximumUsedTransactionIDs, prometheus.GaugeValue, *instance.MaximumUsedTransactionIDs, c.awsAccountID, c.awsRegion, dbidentifier)
+		}
+
+		if instance.ReadThroughput != nil {
+			ch <- prometheus.MustNewConstMetric(c.readThroughput, prometheus.GaugeValue, *instance.ReadThroughput, c.awsAccountID, c.awsRegion, dbidentifier)
+		}
+
+		if instance.ReplicaLag != nil {
+			ch <- prometheus.MustNewConstMetric(c.replicaLag, prometheus.GaugeValue, *instance.ReplicaLag, c.awsAccountID, c.awsRegion, dbidentifier)
+		}
+
+		if instance.ReplicationSlotDiskUsage != nil {
+			ch <- prometheus.MustNewConstMetric(c.replicationSlotDiskUsage, prometheus.GaugeValue, *instance.ReplicationSlotDiskUsage, c.awsAccountID, c.awsRegion, dbidentifier)
+		}
+
+		if instance.SwapUsage != nil {
+			ch <- prometheus.MustNewConstMetric(c.swapUsage, prometheus.GaugeValue, *instance.SwapUsage, c.awsAccountID, c.awsRegion, dbidentifier)
+		}
+
+		if instance.ReadIOPS != nil {
+			ch <- prometheus.MustNewConstMetric(c.readIOPS, prometheus.GaugeValue, *instance.ReadIOPS, c.awsAccountID, c.awsRegion, dbidentifier)
+		}
+
+		if instance.WriteIOPS != nil {
+			ch <- prometheus.MustNewConstMetric(c.writeIOPS, prometheus.GaugeValue, *instance.WriteIOPS, c.awsAccountID, c.awsRegion, dbidentifier)
+		}
+
+		if instance.WriteThroughput != nil {
+			ch <- prometheus.MustNewConstMetric(c.writeThroughput, prometheus.GaugeValue, *instance.WriteThroughput, c.awsAccountID, c.awsRegion, dbidentifier)
+		}
+
+		if instance.DBLoad != nil {
+			ch <- prometheus.MustNewConstMetric(c.DBLoad, prometheus.GaugeValue, *instance.DBLoad, c.awsAccountID, c.awsRegion, dbidentifier)
+		}
+
+		if instance.CPUUtilization != nil {
+			ch <- prometheus.MustNewConstMetric(c.cpuUtilisation, prometheus.GaugeValue, *instance.CPUUtilization, c.awsAccountID, c.awsRegion, dbidentifier)
+		}
+
+		if instance.DBLoadCPU != nil {
+			ch <- prometheus.MustNewConstMetric(c.dBLoadCPU, prometheus.GaugeValue, *instance.DBLoadCPU, c.awsAccountID, c.awsRegion, dbidentifier)
+		}
+
+		if instance.DBLoadNonCPU != nil {
+			ch <- prometheus.MustNewConstMetric(c.dBLoadNonCPU, prometheus.GaugeValue, *instance.DBLoadNonCPU, c.awsAccountID, c.awsRegion, dbidentifier)
+		}
 	}
 
 	// usage metrics

diff --git a/internal/app/rds/rds.go b/internal/app/rds/rds.go
@@ -31,7 +31,7 @@ type RdsInstanceMetrics struct {
 	StorageThroughput                int64
 	MaxAllocatedStorage              int64
 	MaxIops                          int64
-	LogFilesSize                     int64
+	LogFilesSize                     *int64
 	PendingMaintenanceAction         string
 	PendingModifiedValues            bool
 	BackupRetentionPeriod            int32
@@ -243,8 +243,8 @@ func (r *RDSFetcher) computeInstanceMetrics(dbInstance aws_rds_types.DBInstance,
 }
 
 // getLogFilesSize returns the size of all logs on the specified instance
-func (r *RDSFetcher) getLogFilesSize(dbidentifier string) (int64, error) {
-	var filesSize int64
+func (r *RDSFetcher) getLogFilesSize(dbidentifier string) (*int64, error) {
+	var filesSize *int64
 
 	input := &aws_rds.DescribeDBLogFilesInput{DBInstanceIdentifier: &dbidentifier}
 
@@ -254,15 +254,18 @@ func (r *RDSFetcher) getLogFilesSize(dbidentifier string) (int64, error) {
 	if err != nil {
 		var notFoundError *aws_rds_types.DBInstanceNotFoundFault
 		if errors.As(err, &notFoundError) { // Replica in "creating" status may return notFoundError exception
-			return 0, nil
+			return filesSize, nil
 		}
 
-		return 0, fmt.Errorf("can't describe db logs files for %s: %w", dbidentifier, err)
+		return filesSize, fmt.Errorf("can't describe db logs files for %s: %w", dbidentifier, err)
 	}
 
-	if result != nil {
+	if result != nil && len(result.DescribeDBLogFiles) > 0 {
+		if filesSize == nil {
+			filesSize = new(int64)
+		}
 		for _, file := range result.DescribeDBLogFiles {
-			filesSize += file.Size
+			*filesSize += file.Size
 		}
 	}