From 80718d4dedf2b89429b0ac7bc2aa2851fdc684bc Mon Sep 17 00:00:00 2001
From: Henning Baldersheim <balder@yahoo-inc.com>
Date: Tue, 16 Apr 2024 00:06:05 +0200
Subject: - Remove duplicates of the timestamp. - Avoid 17M of duplicates of
 0.0. - Avoid duplicating the optionally sanitized metric name.

---
 .../metric/model/prometheus/PrometheusUtil.java         | 17 +++++++++++++----
 .../ai/vespa/metricsproxy/service/MetricsParser.java    |  4 +++-
 2 files changed, 16 insertions(+), 5 deletions(-)

(limited to 'metrics-proxy')

diff --git a/metrics-proxy/src/main/java/ai/vespa/metricsproxy/metric/model/prometheus/PrometheusUtil.java b/metrics-proxy/src/main/java/ai/vespa/metricsproxy/metric/model/prometheus/PrometheusUtil.java
index d7436ccf404..f1050933a01 100644
--- a/metrics-proxy/src/main/java/ai/vespa/metricsproxy/metric/model/prometheus/PrometheusUtil.java
+++ b/metrics-proxy/src/main/java/ai/vespa/metricsproxy/metric/model/prometheus/PrometheusUtil.java
@@ -19,7 +19,15 @@ import java.util.stream.Collectors;
  */
 public class PrometheusUtil {
 
+    private static String sanitize(String name, Map<String, String> sanitizedCache) {
+        return sanitizedCache.computeIfAbsent(name, key -> {
+            String sanitized = Collector.sanitizeMetricName(name);
+            return (name.equals(sanitized)) ? name : sanitized;
+        });
+    }
+
     public static PrometheusModel toPrometheusModel(List<MetricsPacket> metricsPackets) {
+        Map<String, String> sanitizedMetrics = new HashMap<>();
         Map<ServiceId, List<MetricsPacket>> packetsByService = metricsPackets.stream()
                 .collect(Collectors.groupingBy(packet -> packet.service));
 
@@ -28,13 +36,14 @@ public class PrometheusUtil {
         Map<String, List<Sample>> samples = new HashMap<>();
         packetsByService.forEach(((serviceId, packets) -> {
 
-            var serviceName = Collector.sanitizeMetricName(serviceId.id);
+            var serviceName = sanitize(serviceId.id, sanitizedMetrics);
             for (var packet : packets) {
+                Long timeStamp = packet.timestamp * 1000;
                 var dimensions = packet.dimensions();
                 List<String> labels = new ArrayList<>(dimensions.size());
                 List<String> labelValues = new ArrayList<>(dimensions.size());
                 for (var entry : dimensions.entrySet()) {
-                    var labelName = Collector.sanitizeMetricName(entry.getKey().id);
+                    var labelName = sanitize(entry.getKey().id, sanitizedMetrics);
                     labels.add(labelName);
                     labelValues.add(entry.getValue());
                 }
@@ -42,7 +51,7 @@ public class PrometheusUtil {
                 labelValues.add(serviceName);
 
                 for (var metric : packet.metrics().entrySet()) {
-                    var metricName = Collector.sanitizeMetricName(metric.getKey().id);
+                    var metricName = sanitize(metric.getKey().id, sanitizedMetrics);
                     List<Sample> sampleList;
                     if (samples.containsKey(metricName)) {
                         sampleList = samples.get(metricName);
@@ -51,7 +60,7 @@ public class PrometheusUtil {
                         samples.put(metricName, sampleList);
                         metricFamilySamples.add(new MetricFamilySamples(metricName, Collector.Type.UNKNOWN, "", sampleList));
                     }
-                    sampleList.add(new Sample(metricName, labels, labelValues, metric.getValue().doubleValue(), packet.timestamp * 1000));
+                    sampleList.add(new Sample(metricName, labels, labelValues, metric.getValue().doubleValue(), timeStamp));
                 }
             }
             if (!packets.isEmpty()) {
diff --git a/metrics-proxy/src/main/java/ai/vespa/metricsproxy/service/MetricsParser.java b/metrics-proxy/src/main/java/ai/vespa/metricsproxy/service/MetricsParser.java
index 6c3b759e97b..0e33d7dbf2f 100644
--- a/metrics-proxy/src/main/java/ai/vespa/metricsproxy/service/MetricsParser.java
+++ b/metrics-proxy/src/main/java/ai/vespa/metricsproxy/service/MetricsParser.java
@@ -28,6 +28,7 @@ import static ai.vespa.metricsproxy.metric.model.DimensionId.toDimensionId;
  * @author Jo Kristian Bergum
  */
 public class MetricsParser {
+    private static final Double ZERO_DOUBLE = 0d;
     public interface Collector {
         void accept(Metric metric);
     }
@@ -186,7 +187,8 @@ public class MetricsParser {
             if (token == JsonToken.VALUE_NUMBER_INT) {
                 metrics.add(Map.entry(metricName, parser.getLongValue()));
             } else if (token == JsonToken.VALUE_NUMBER_FLOAT) {
-                metrics.add(Map.entry(metricName, parser.getValueAsDouble()));
+                double value = parser.getValueAsDouble();
+                metrics.add(Map.entry(metricName, value == ZERO_DOUBLE ? ZERO_DOUBLE : value));
             } else {
                 throw new IllegalArgumentException("Value for aggregator '" + fieldName + "' is not a number");
             }
-- 
cgit v1.2.3