Introduce explicit preview option to disable ingestion sampling (#3713)

trask · web-flow · commit ae3b5a3de49e · 2024-05-31T01:13:06.000Z
diff --git a/.github/workflows/build-common.yml b/.github/workflows/build-common.yml
@@ -8,7 +8,7 @@ on:
         required: false
         
 env:
-  EXPORTER_VERSION: 1.0.0-beta.23 # to be updated with the latest version
+  EXPORTER_VERSION: 1.0.0-beta.24 # to be updated with the latest version
 
 jobs:
   spotless:
diff --git a/agent/agent-tooling/build.gradle.kts b/agent/agent-tooling/build.gradle.kts
@@ -21,7 +21,7 @@ dependencies {
   implementation(project(":agent:agent-profiler:agent-diagnostics"))
   implementation(project(":etw:java"))
 
-  implementation("com.azure:azure-monitor-opentelemetry-exporter:1.0.0-beta.23")
+  implementation("com.azure:azure-monitor-opentelemetry-exporter:1.0.0-beta.24")
   compileOnly("io.opentelemetry.javaagent:opentelemetry-javaagent-bootstrap")
   compileOnly("io.opentelemetry.javaagent:opentelemetry-javaagent-tooling")
   compileOnly("io.opentelemetry.javaagent.instrumentation:opentelemetry-javaagent-servlet-common-bootstrap")
diff --git a/agent/agent-tooling/gradle.lockfile b/agent/agent-tooling/gradle.lockfile
@@ -9,7 +9,7 @@ com.azure:azure-core-http-netty:1.14.2=runtimeClasspath
 com.azure:azure-core:1.48.0=runtimeClasspath
 com.azure:azure-identity:1.12.1=runtimeClasspath
 com.azure:azure-json:1.1.0=runtimeClasspath
-com.azure:azure-monitor-opentelemetry-exporter:1.0.0-beta.23=runtimeClasspath
+com.azure:azure-monitor-opentelemetry-exporter:1.0.0-beta.24=runtimeClasspath
 com.azure:azure-sdk-bom:1.2.23=runtimeClasspath
 com.azure:azure-storage-blob:12.25.4=runtimeClasspath
 com.azure:azure-storage-common:12.24.4=runtimeClasspath
diff --git a/agent/agent-tooling/src/main/java/com/microsoft/applicationinsights/agent/internal/classicsdk/BytecodeUtilImpl.java b/agent/agent-tooling/src/main/java/com/microsoft/applicationinsights/agent/internal/classicsdk/BytecodeUtilImpl.java
@@ -506,9 +506,9 @@ private static void track(
     }
 
     if (isPartOfTheCurrentTrace && applySampling && span instanceof ReadableSpan) {
-      Long itemCount = ((ReadableSpan) span).getAttribute(AiSemanticAttributes.ITEM_COUNT);
-      if (itemCount != null) {
-        telemetryBuilder.setSampleRate(100.0f / itemCount);
+      Double sampleRate = ((ReadableSpan) span).getAttribute(AiSemanticAttributes.SAMPLE_RATE);
+      if (sampleRate != null) {
+        telemetryBuilder.setSampleRate(sampleRate.floatValue());
       }
     }
 
diff --git a/agent/agent-tooling/src/main/java/com/microsoft/applicationinsights/agent/internal/configuration/Configuration.java b/agent/agent-tooling/src/main/java/com/microsoft/applicationinsights/agent/internal/configuration/Configuration.java
@@ -159,13 +159,13 @@ public static class Role {
 
   public static class Sampling {
 
-    // fixed percentage of requests
-    @Nullable public Double percentage;
-
     // default is 5 requests per second (set in ConfigurationBuilder if neither percentage nor
     // requestsPerSecond was configured)
     @Nullable public Double requestsPerSecond;
 
+    // fixed percentage of requests
+    @Nullable public Double percentage;
+
     // this config option only existed in one BETA release (3.4.0-BETA)
     @Deprecated @Nullable public Double limitPerSecond;
 
@@ -202,6 +202,8 @@ public static class SamplingPreview {
     // future goal: make parentBased sampling the default if item count is received via tracestate
     public boolean parentBased;
 
+    public boolean ingestionSamplingEnabled = true;
+
     @Deprecated public List<SamplingOverride> overrides = new ArrayList<>();
   }
 
diff --git a/agent/agent-tooling/src/main/java/com/microsoft/applicationinsights/agent/internal/configuration/ConfigurationBuilder.java b/agent/agent-tooling/src/main/java/com/microsoft/applicationinsights/agent/internal/configuration/ConfigurationBuilder.java
@@ -269,7 +269,7 @@ private static void logConfigurationWarnings(Configuration config) {
               + "If you're on premise, you can use APPLICATIONINSIGHTS_AUTHENTICATION_STRING environment variable to pass the client ID and secret, "
               + "e.g. APPLICATIONINSIGHTS_AUTHENTICATION_STRING=Authorization=AAD;ClientId={CLIENT_ID};ClientSecret={CLIENT_SECRET}.");
     }
-    if (config.sampling.percentage != null && config.sampling.requestsPerSecond != null) {
+    if (config.sampling.requestsPerSecond != null && config.sampling.percentage != null) {
       configurationLogger.warn(
           "Sampling \"requestsPerSecond\" and \"percentage\" should not be used at the same time."
               + " Please remove one of them.");
diff --git a/agent/agent-tooling/src/main/java/com/microsoft/applicationinsights/agent/internal/exporter/AgentLogExporter.java b/agent/agent-tooling/src/main/java/com/microsoft/applicationinsights/agent/internal/exporter/AgentLogExporter.java
@@ -105,13 +105,7 @@ public CompletableResultCode export(Collection<LogRecordData> logs) {
           continue;
         }
 
-        Long itemCount = null;
-        if (samplingPercentage != null) {
-          // samplingPercentage cannot be 0 here
-          itemCount = Math.round(100.0 / samplingPercentage);
-        }
-
-        TelemetryItem telemetryItem = mapper.map(log, stack, itemCount);
+        TelemetryItem telemetryItem = mapper.map(log, stack, samplingPercentage);
         telemetryItemConsumer.accept(telemetryItem);
 
         exportingLogLogger.recordSuccess();
diff --git a/agent/agent-tooling/src/main/java/com/microsoft/applicationinsights/agent/internal/init/AzureMonitorLogProcessor.java b/agent/agent-tooling/src/main/java/com/microsoft/applicationinsights/agent/internal/init/AzureMonitorLogProcessor.java
@@ -40,9 +40,9 @@ public void onEmit(Context context, ReadWriteLogRecord logRecord) {
     ReadableSpan readableSpan = (ReadableSpan) currentSpan;
     logRecord.setAttribute(
         AiSemanticAttributes.OPERATION_NAME, OperationNames.getOperationName(readableSpan));
-    Long itemCount = readableSpan.getAttribute(AiSemanticAttributes.ITEM_COUNT);
-    if (itemCount != null) {
-      logRecord.setAttribute(AiSemanticAttributes.ITEM_COUNT, itemCount);
+    Double sampleRate = readableSpan.getAttribute(AiSemanticAttributes.SAMPLE_RATE);
+    if (sampleRate != null) {
+      logRecord.setAttribute(AiSemanticAttributes.SAMPLE_RATE, sampleRate);
     }
   }
 
diff --git a/agent/agent-tooling/src/main/java/com/microsoft/applicationinsights/agent/internal/sampling/AiSampler.java b/agent/agent-tooling/src/main/java/com/microsoft/applicationinsights/agent/internal/sampling/AiSampler.java
@@ -27,24 +27,34 @@
 // * adds item count to span attribute if it is sampled
 public class AiSampler implements Sampler {
 
+  private static final double SAMPLE_RATE_TO_DISABLE_INGESTION_SAMPLING = 99.99;
+
+  private final boolean ingestionSamplingEnabled;
   private final boolean localParentBased;
   private final SamplingPercentage requestSamplingPercentage;
   // when localParentBased=false, then this applies to all dependencies, not only parentless
   private final SamplingPercentage parentlessDependencySamplingPercentage;
-  private final Cache<Long, SamplingResult> recordAndSampleWithItemCountMap = Cache.bounded(100);
+  private final Cache<Double, SamplingResult> recordAndSampleWithSampleRateMap = Cache.bounded(100);
 
   public AiSampler(
       SamplingPercentage requestSamplingPercentage,
-      SamplingPercentage parentlessDependencySamplingPercentage) {
-    this(requestSamplingPercentage, parentlessDependencySamplingPercentage, true);
+      SamplingPercentage parentlessDependencySamplingPercentage,
+      boolean ingestionSamplingEnabled) {
+    this(
+        requestSamplingPercentage,
+        parentlessDependencySamplingPercentage,
+        ingestionSamplingEnabled,
+        true);
   }
 
   public AiSampler(
       SamplingPercentage requestSamplingPercentage,
       SamplingPercentage parentlessDependencySamplingPercentage,
+      boolean ingestionSamplingEnabled,
       boolean localParentBased) {
     this.requestSamplingPercentage = requestSamplingPercentage;
     this.parentlessDependencySamplingPercentage = parentlessDependencySamplingPercentage;
+    this.ingestionSamplingEnabled = ingestionSamplingEnabled;
     this.localParentBased = localParentBased;
   }
 
@@ -76,24 +86,29 @@ public SamplingResult shouldSample(
               : parentlessDependencySamplingPercentage.get();
     }
 
-    if (sp == SamplingPercentage.USE_INGESTION_SAMPLING) {
-      return SamplingResult.recordAndSample();
-    }
-
     if (sp == 0) {
       return SamplingResult.drop();
     }
 
-    if (!shouldRecordAndSample(traceId, sp)) {
+    if (sp != 100 && !shouldRecordAndSample(traceId, sp)) {
       return SamplingResult.drop();
     }
 
-    // sp cannot be 0 here
-    long itemCount = Math.round(100.0 / sp);
-    SamplingResult samplingResult = recordAndSampleWithItemCountMap.get(itemCount);
+    if (sp == 100 && ingestionSamplingEnabled) {
+      return SamplingResult.recordAndSample();
+    }
+
+    if (sp == 100) {
+      // ingestion sampling is applied when sample rate is 100 (or missing)
+      // so we set it to 99.99 which will bypass ingestion sampling
+      // (and will still be stored as item count 1)
+      sp = SAMPLE_RATE_TO_DISABLE_INGESTION_SAMPLING;
+    }
+
+    SamplingResult samplingResult = recordAndSampleWithSampleRateMap.get(sp);
     if (samplingResult == null) {
-      samplingResult = new RecordAndSampleWithItemCount(itemCount);
-      recordAndSampleWithItemCountMap.put(itemCount, samplingResult);
+      samplingResult = new RecordAndSampleWithItemCount(sp);
+      recordAndSampleWithSampleRateMap.put(sp, samplingResult);
     }
     return samplingResult;
   }
@@ -112,9 +127,10 @@ private static SamplingResult useLocalParentDecisionIfPossible(Context parentCon
       return SamplingResult.drop();
     }
     if (parentSpan instanceof ReadableSpan) {
-      Long itemCount = ((ReadableSpan) parentSpan).getAttribute(AiSemanticAttributes.ITEM_COUNT);
-      if (itemCount != null) {
-        return new RecordAndSampleWithItemCount(itemCount);
+      Double parentSampleRate =
+          ((ReadableSpan) parentSpan).getAttribute(AiSemanticAttributes.SAMPLE_RATE);
+      if (parentSampleRate != null) {
+        return new RecordAndSampleWithItemCount(parentSampleRate);
       }
     }
     return null;
@@ -141,8 +157,8 @@ private static class RecordAndSampleWithItemCount implements SamplingResult {
 
     private final Attributes attributes;
 
-    RecordAndSampleWithItemCount(long itemCount) {
-      attributes = Attributes.builder().put(AiSemanticAttributes.ITEM_COUNT, itemCount).build();
+    RecordAndSampleWithItemCount(double sampleRate) {
+      attributes = Attributes.builder().put(AiSemanticAttributes.SAMPLE_RATE, sampleRate).build();
     }
 
     @Override
diff --git a/agent/agent-tooling/src/main/java/com/microsoft/applicationinsights/agent/internal/sampling/Samplers.java b/agent/agent-tooling/src/main/java/com/microsoft/applicationinsights/agent/internal/sampling/Samplers.java
@@ -18,15 +18,16 @@ public static Sampler getSampler(
       SamplingPercentage requestSamplingPercentage =
           SamplingPercentage.rateLimited(sampling.requestsPerSecond);
       SamplingPercentage parentlessDependencySamplingPercentage = SamplingPercentage.fixed(100);
-      sampler = new AiSampler(requestSamplingPercentage, parentlessDependencySamplingPercentage);
+      sampler =
+          new AiSampler(
+              requestSamplingPercentage,
+              parentlessDependencySamplingPercentage,
+              samplingPreview.ingestionSamplingEnabled);
     } else if (sampling.percentage != null) {
-      SamplingPercentage samplingPercentage;
-      if (sampling.percentage == 100) {
-        samplingPercentage = SamplingPercentage.useIngestionSampling();
-      } else {
-        samplingPercentage = SamplingPercentage.fixed(sampling.percentage);
-      }
-      sampler = new AiSampler(samplingPercentage, samplingPercentage);
+      SamplingPercentage samplingPercentage = SamplingPercentage.fixed(sampling.percentage);
+      sampler =
+          new AiSampler(
+              samplingPercentage, samplingPercentage, samplingPreview.ingestionSamplingEnabled);
     } else {
       throw new AssertionError("ConfigurationBuilder should have set the default sampling");
     }
diff --git a/agent/agent-tooling/src/main/java/com/microsoft/applicationinsights/agent/internal/sampling/SamplingOverrides.java b/agent/agent-tooling/src/main/java/com/microsoft/applicationinsights/agent/internal/sampling/SamplingOverrides.java
@@ -70,7 +70,7 @@ private MatcherGroup(SamplingOverride override) {
         }
       }
       samplingPercentage = SamplingPercentage.fixed(override.percentage);
-      sampler = new AiSampler(samplingPercentage, samplingPercentage, false);
+      sampler = new AiSampler(samplingPercentage, samplingPercentage, false, false);
     }
 
     Sampler getSampler() {
diff --git a/agent/agent-tooling/src/main/java/com/microsoft/applicationinsights/agent/internal/sampling/SamplingPercentage.java b/agent/agent-tooling/src/main/java/com/microsoft/applicationinsights/agent/internal/sampling/SamplingPercentage.java
@@ -12,8 +12,6 @@
 // the portal
 public interface SamplingPercentage {
 
-  long USE_INGESTION_SAMPLING = -1;
-
   double get();
 
   static SamplingPercentage fixed(double percentage) {
@@ -24,8 +22,4 @@ static SamplingPercentage fixed(double percentage) {
   static SamplingPercentage rateLimited(double targetPerSecondLimit) {
     return new RateLimitedSamplingPercentage(targetPerSecondLimit, 0.1);
   }
-
-  static SamplingPercentage useIngestionSampling() {
-    return () -> USE_INGESTION_SAMPLING;
-  }
 }
diff --git a/agent/agent-tooling/src/test/java/com/microsoft/applicationinsights/agent/internal/SamplingTestUtil.java b/agent/agent-tooling/src/test/java/com/microsoft/applicationinsights/agent/internal/SamplingTestUtil.java
@@ -37,8 +37,7 @@ public static Double getCurrentSamplingPercentage(Sampler sampler) {
             SpanKind.SERVER,
             Attributes.empty(),
             Collections.emptyList());
-    Long itemCount = samplingResult.getAttributes().get(AiSemanticAttributes.ITEM_COUNT);
-    return itemCount == null ? null : 100.0 / itemCount;
+    return samplingResult.getAttributes().get(AiSemanticAttributes.SAMPLE_RATE);
   }
 
   private SamplingTestUtil() {}
diff --git a/licenses/more-licenses.md b/licenses/more-licenses.md
@@ -350,7 +350,7 @@ _2024-05-27 03:31:29 UTC_
 > - **POM Project URL**: [https://github.com/Azure/azure-sdk-for-java](https://github.com/Azure/azure-sdk-for-java)
 > - **POM License**: MIT License - [https://opensource.org/licenses/MIT](https://opensource.org/licenses/MIT)
 
-**67** **Group:** `com.azure` **Name:** `azure-monitor-opentelemetry-exporter` **Version:** `1.0.0-beta.23` 
+**67** **Group:** `com.azure` **Name:** `azure-monitor-opentelemetry-exporter` **Version:** `1.0.0-beta.24` 
 > - **POM License**: MIT License - [https://opensource.org/licenses/MIT](https://opensource.org/licenses/MIT)
 
 **68** **Group:** `com.azure` **Name:** `azure-storage-blob` **Version:** `12.25.4` 
diff --git a/smoke-tests/apps/Sampling/src/main/java/com/microsoft/applicationinsights/smoketestapp/SimpleSamplingServlet.java b/smoke-tests/apps/Sampling/src/main/java/com/microsoft/applicationinsights/smoketestapp/SimpleSamplingServlet.java
diff --git a/smoke-tests/apps/Sampling/src/main/java/com/microsoft/applicationinsights/smoketestapp/SimpleServlet.java b/smoke-tests/apps/Sampling/src/main/java/com/microsoft/applicationinsights/smoketestapp/SimpleServlet.java
@@ -13,10 +13,10 @@
 import javax.servlet.http.HttpServletRequest;
 import javax.servlet.http.HttpServletResponse;
 
-@WebServlet("/no-sampling")
-public class NoSamplingServlet extends HttpServlet {
+@WebServlet("/simple")
+public class SimpleServlet extends HttpServlet {
 
-  private static final Logger logger = Logger.getLogger(NoSamplingServlet.class.getName());
+  private static final Logger logger = Logger.getLogger(SimpleServlet.class.getName());
 
   private final TelemetryClient client = new TelemetryClient();
 
diff --git a/smoke-tests/apps/Sampling/src/smokeTest/java/com/microsoft/applicationinsights/smoketest/IngestionSamplingDisabledTest.java b/smoke-tests/apps/Sampling/src/smokeTest/java/com/microsoft/applicationinsights/smoketest/IngestionSamplingDisabledTest.java
diff --git a/smoke-tests/apps/Sampling/src/smokeTest/java/com/microsoft/applicationinsights/smoketest/IngestionSamplingTest.java b/smoke-tests/apps/Sampling/src/smokeTest/java/com/microsoft/applicationinsights/smoketest/IngestionSamplingTest.java
diff --git a/smoke-tests/apps/Sampling/src/smokeTest/java/com/microsoft/applicationinsights/smoketest/SamplingTest.java b/smoke-tests/apps/Sampling/src/smokeTest/java/com/microsoft/applicationinsights/smoketest/SamplingTest.java
diff --git a/smoke-tests/apps/Sampling/src/smokeTest/resources/applicationinsights-ingestion-sampling-disabled.json b/smoke-tests/apps/Sampling/src/smokeTest/resources/applicationinsights-ingestion-sampling-disabled.json
diff --git a/smoke-tests/apps/SamplingOverrides/src/smokeTest/java/com/microsoft/applicationinsights/smoketest/SamplingOverrides2Test.java b/smoke-tests/apps/SamplingOverrides/src/smokeTest/java/com/microsoft/applicationinsights/smoketest/SamplingOverrides2Test.java

Original file line number	Diff line number	Diff line change
`@@ -506,9 +506,9 @@ private static void track(`
`506`	`506`	`}`
`507`	`507`
`508`	`508`	`if (isPartOfTheCurrentTrace && applySampling && span instanceof ReadableSpan) {`
`509`		`- Long itemCount = ((ReadableSpan) span).getAttribute(AiSemanticAttributes.ITEM_COUNT);`
`510`		`- if (itemCount != null) {`
`511`		`- telemetryBuilder.setSampleRate(100.0f / itemCount);`
	`509`	`+ Double sampleRate = ((ReadableSpan) span).getAttribute(AiSemanticAttributes.SAMPLE_RATE);`
	`510`	`+ if (sampleRate != null) {`
	`511`	`+ telemetryBuilder.setSampleRate(sampleRate.floatValue());`
`512`	`512`	`}`
`513`	`513`	`}`
`514`	`514`
Original file line number	Diff line number	Diff line change
`@@ -269,7 +269,7 @@ private static void logConfigurationWarnings(Configuration config) {`
`269`	`269`	`+ "If you're on premise, you can use APPLICATIONINSIGHTS_AUTHENTICATION_STRING environment variable to pass the client ID and secret, "`
`270`	`270`	`+ "e.g. APPLICATIONINSIGHTS_AUTHENTICATION_STRING=Authorization=AAD;ClientId={CLIENT_ID};ClientSecret={CLIENT_SECRET}.");`
`271`	`271`	`}`
`272`		`- if (config.sampling.percentage != null && config.sampling.requestsPerSecond != null) {`
	`272`	`+ if (config.sampling.requestsPerSecond != null && config.sampling.percentage != null) {`
`273`	`273`	`configurationLogger.warn(`
`274`	`274`	`"Sampling \"requestsPerSecond\" and \"percentage\" should not be used at the same time."`
`275`	`275`	`+ " Please remove one of them.");`
Original file line number	Diff line number	Diff line change
`@@ -40,9 +40,9 @@ public void onEmit(Context context, ReadWriteLogRecord logRecord) {`
`40`	`40`	`ReadableSpan readableSpan = (ReadableSpan) currentSpan;`
`41`	`41`	`logRecord.setAttribute(`
`42`	`42`	`AiSemanticAttributes.OPERATION_NAME, OperationNames.getOperationName(readableSpan));`
`43`		`- Long itemCount = readableSpan.getAttribute(AiSemanticAttributes.ITEM_COUNT);`
`44`		`- if (itemCount != null) {`
`45`		`- logRecord.setAttribute(AiSemanticAttributes.ITEM_COUNT, itemCount);`
	`43`	`+ Double sampleRate = readableSpan.getAttribute(AiSemanticAttributes.SAMPLE_RATE);`
	`44`	`+ if (sampleRate != null) {`
	`45`	`+ logRecord.setAttribute(AiSemanticAttributes.SAMPLE_RATE, sampleRate);`
`46`	`46`	`}`
`47`	`47`	`}`
`48`	`48`