add predicted data samples when time series cycle duration is Hour

pkg/prediction/dsp/prediction.go

@@ -1,6 +1,7 @@
 package dsp
 
 import (
+	"encoding/json"
 	"fmt"
 	"math"
 	"sync"
@@ -210,7 +211,10 @@ func (p *periodicSignalPrediction) updateAggregateSignals(id string, tsList []*c
 	var predictedTimeSeriesList []*common.TimeSeries
 
 	for _, ts := range tsList {
-		klog.V(4).InfoS("Got time series.", "queryExpr", id, "samples", ts.Samples, "labels", ts.Labels)
+		if klog.V(6).Enabled() {
+			sampleData, err := json.Marshal(ts.Samples)
+			klog.V(6).Infof("Got time series, queryExpr: %v, samples: %v, labels: %v, err: %v", id, string(sampleData), ts.Labels, err)
+		}
 		var chosenEstimator Estimator
 		var signal *Signal
 		var nCycles int
@@ -232,7 +236,7 @@ func (p *periodicSignalPrediction) updateAggregateSignals(id string, tsList []*c
 			signal = SamplesToSignal(ts.Samples, config.historyResolution)
 			signal, nCycles = signal.Truncate(cycleDuration)
 			if nCycles >= 2 {
-				chosenEstimator = bestEstimator(config.estimators, signal, nCycles, cycleDuration)
+				chosenEstimator = bestEstimator(id, config.estimators, signal, nCycles, cycleDuration)
 			}
 		}
 
@@ -241,13 +245,26 @@ func (p *periodicSignalPrediction) updateAggregateSignals(id string, tsList []*c
 			intervalSeconds := int64(config.historyResolution.Seconds())
 			nextTimestamp := ts.Samples[len(ts.Samples)-1].Timestamp + intervalSeconds
 
-			samples := make([]common.Sample, len(estimatedSignal.Samples))
-			for i := range estimatedSignal.Samples {
-				samples[i] = common.Sample{
-					Value:     estimatedSignal.Samples[i],
-					Timestamp: nextTimestamp,
+			// Hack(temporary):
+			// Because the dsp predict only append one cycle points, when the cycle is hour, then estimate signal samples only contains at most one hour points
+			// This leads to tsp predictWindowSeconds more than 3600 will be always out of date. because the predicted data end point timestamp is always ts.Samples[len(ts.Samples)-1].Timestamp+ Hour in one model update interval loop
+			// If its cycle is hour, then we append 24 hour points to avoid the out of dated predicted data
+			// Alternative option 1: Reduce predictWindowSeconds in tsp less than one hour and model update interval to one hour too.
+			// Alternative option 2. Do not support hour cycle any more, because it is too short in production env. now the dsp can not handle hour cycle well.
+			cycles := 1
+			if cycleDuration == Hour {
+				cycles = 24
+			}
+			n := len(estimatedSignal.Samples)
+			samples := make([]common.Sample, n*cycles)
+			for c := 0; c < cycles; c++ {
+				for i := range estimatedSignal.Samples {
+					samples[i+c*n] = common.Sample{
+						Value:     estimatedSignal.Samples[i],
+						Timestamp: nextTimestamp,
+					}
+					nextTimestamp += intervalSeconds
 				}
-				nextTimestamp += intervalSeconds
 			}
 
 			predictedTimeSeriesList = append(predictedTimeSeriesList, &common.TimeSeries{
@@ -268,7 +285,7 @@ func (p *periodicSignalPrediction) updateAggregateSignals(id string, tsList []*c
 	}
 }
 
-func bestEstimator(estimators []Estimator, signal *Signal, totalCycles int, cycleDuration time.Duration) Estimator {
+func bestEstimator(id string, estimators []Estimator, signal *Signal, totalCycles int, cycleDuration time.Duration) Estimator {
 	samplesPerCycle := len(signal.Samples) / totalCycles
 
 	history := &Signal{
@@ -287,15 +304,15 @@ func bestEstimator(estimators []Estimator, signal *Signal, totalCycles int, cycl
 		estimated := estimators[i].GetEstimation(history, cycleDuration)
 		if estimated != nil {
 			pe, err := accuracy.PredictionError(actual.Samples, estimated.Samples)
-			klog.InfoS("Testing estimators ...", "estimator", estimators[i].String(), "error", pe)
+			klog.V(6).InfoS("Testing estimators ...", "key", id, "estimator", estimators[i].String(), "pe", pe, "error", err)
 			if err == nil && pe < minPE {
 				minPE = pe
 				bestEstimator = estimators[i]
 			}
 		}
 	}
 
-	klog.InfoS("Got the best estimator.", "estimator", bestEstimator.String(), "error", minPE)
+	klog.V(4).InfoS("Got the best estimator.", "key", id, "estimator", bestEstimator.String(), "minPE", minPE, "totalCycles", totalCycles)
 	return bestEstimator
 }