Merge branch '263_use_median_for_statsseries' into Refactor-job-footprint

internal/graph/generated/generated.go

@@ -249,9 +249,10 @@ type ComplexityRoot struct {
 	}
 
 	StatsSeries struct {
-		Max  func(childComplexity int) int
-		Mean func(childComplexity int) int
-		Min  func(childComplexity int) int
+		Max    func(childComplexity int) int
+		Mean   func(childComplexity int) int
+		Median func(childComplexity int) int
+		Min    func(childComplexity int) int
 	}
 
 	SubCluster struct {
@@ -1326,6 +1327,13 @@ func (e *executableSchema) Complexity(typeName, field string, childComplexity in
 
 		return e.complexity.StatsSeries.Mean(childComplexity), true
 
+	case "StatsSeries.median":
+		if e.complexity.StatsSeries.Median == nil {
+			break
+		}
+
+		return e.complexity.StatsSeries.Median(childComplexity), true
+
 	case "StatsSeries.min":
 		if e.complexity.StatsSeries.Min == nil {
 			break
@@ -1860,9 +1868,10 @@ type MetricStatistics {
 }
 
 type StatsSeries {
-  mean: [NullableFloat!]!
-  min:  [NullableFloat!]!
-  max:  [NullableFloat!]!
+  mean:   [NullableFloat!]!
+  median: [NullableFloat!]!
+  min:    [NullableFloat!]!
+  max:    [NullableFloat!]!
 }
 
 type MetricFootprints {
@@ -4662,6 +4671,8 @@ func (ec *executionContext) fieldContext_JobMetric_statisticsSeries(ctx context.
 			switch field.Name {
 			case "mean":
 				return ec.fieldContext_StatsSeries_mean(ctx, field)
+			case "median":
+				return ec.fieldContext_StatsSeries_median(ctx, field)
 			case "min":
 				return ec.fieldContext_StatsSeries_min(ctx, field)
 			case "max":
@@ -8649,6 +8660,50 @@ func (ec *executionContext) fieldContext_StatsSeries_mean(ctx context.Context, f
 	return fc, nil
 }
 
+func (ec *executionContext) _StatsSeries_median(ctx context.Context, field graphql.CollectedField, obj *schema.StatsSeries) (ret graphql.Marshaler) {
+	fc, err := ec.fieldContext_StatsSeries_median(ctx, field)
+	if err != nil {
+		return graphql.Null
+	}
+	ctx = graphql.WithFieldContext(ctx, fc)
+	defer func() {
+		if r := recover(); r != nil {
+			ec.Error(ctx, ec.Recover(ctx, r))
+			ret = graphql.Null
+		}
+	}()
+	resTmp, err := ec.ResolverMiddleware(ctx, func(rctx context.Context) (interface{}, error) {
+		ctx = rctx // use context from middleware stack in children
+		return obj.Median, nil
+	})
+	if err != nil {
+		ec.Error(ctx, err)
+		return graphql.Null
+	}
+	if resTmp == nil {
+		if !graphql.HasFieldError(ctx, fc) {
+			ec.Errorf(ctx, "must not be null")
+		}
+		return graphql.Null
+	}
+	res := resTmp.([]schema.Float)
+	fc.Result = res
+	return ec.marshalNNullableFloat2ᚕgithubᚗcomᚋClusterCockpitᚋccᚑbackendᚋpkgᚋschemaᚐFloatᚄ(ctx, field.Selections, res)
+}
+
+func (ec *executionContext) fieldContext_StatsSeries_median(ctx context.Context, field graphql.CollectedField) (fc *graphql.FieldContext, err error) {
+	fc = &graphql.FieldContext{
+		Object:     "StatsSeries",
+		Field:      field,
+		IsMethod:   false,
+		IsResolver: false,
+		Child: func(ctx context.Context, field graphql.CollectedField) (*graphql.FieldContext, error) {
+			return nil, errors.New("field of type NullableFloat does not have child fields")
+		},
+	}
+	return fc, nil
+}
+
 func (ec *executionContext) _StatsSeries_min(ctx context.Context, field graphql.CollectedField, obj *schema.StatsSeries) (ret graphql.Marshaler) {
 	fc, err := ec.fieldContext_StatsSeries_min(ctx, field)
 	if err != nil {
@@ -14431,6 +14486,11 @@ func (ec *executionContext) _StatsSeries(ctx context.Context, sel ast.SelectionS
 			if out.Values[i] == graphql.Null {
 				out.Invalids++
 			}
+		case "median":
+			out.Values[i] = ec._StatsSeries_median(ctx, field, obj)
+			if out.Values[i] == graphql.Null {
+				out.Invalids++
+			}
 		case "min":
 			out.Values[i] = ec._StatsSeries_min(ctx, field, obj)
 			if out.Values[i] == graphql.Null {

internal/metricdata/metricdata.go

@@ -263,7 +263,7 @@ func cacheKey(
 
 // For /monitoring/job/<job> and some other places, flops_any and mem_bw need
 // to be available at the scope 'node'. If a job has a lot of nodes,
-// statisticsSeries should be available so that a min/mean/max Graph can be
+// statisticsSeries should be available so that a min/median/max Graph can be
 // used instead of a lot of single lines.
 func prepareJobData(
 	job *schema.Job,

internal/util/statistics.go

@@ -7,6 +7,10 @@ package util
 import (
 	"github.com/ClusterCockpit/cc-backend/pkg/schema"
 	"golang.org/x/exp/constraints"
+
+	"fmt"
+	"math"
+	"sort"
 )
 
 func Min[T constraints.Ordered](a, b T) T {
@@ -34,3 +38,36 @@ func LoadJobStat(job *schema.JobMeta, metric string) float64 {
 
 	return 0.0
 }
+
+func sortedCopy(input []float64) []float64 {
+	sorted := make([]float64, len(input))
+	copy(sorted, input)
+	sort.Float64s(sorted)
+	return sorted
+}
+
+func Mean(input []float64) (float64, error) {
+	if len(input) == 0 {
+		return math.NaN(), fmt.Errorf("input array is empty: %#v", input)
+	}
+	sum := 0.0
+	for _, n := range input {
+		sum += n
+	}
+	return sum / float64(len(input)), nil
+}
+
+func Median(input []float64) (median float64, err error) {
+	c := sortedCopy(input)
+	// Even numbers: add the two middle numbers, divide by two (use mean function)
+	// Odd numbers: Use the middle number
+	l := len(c)
+	if l == 0 {
+		return math.NaN(), fmt.Errorf("input array is empty: %#v", input)
+	} else if l%2 == 0 {
+		median, _ = Mean(c[l/2-1 : l/2+1])
+	} else {
+		median = c[l/2]
+	}
+	return median, nil
+}