Truncate CI Visibility meta tag values

dd-trace-core/src/main/java/datadog/trace/civisibility/writer/ddintake/CiTestCycleMapperV1.java

@@ -3,6 +3,8 @@
 import static datadog.communication.http.OkHttpUtils.gzippedMsgpackRequestBodyOf;
 import static datadog.communication.http.OkHttpUtils.msgpackRequestBodyOf;
 import static datadog.json.JsonMapper.toJson;
+import static datadog.trace.api.civisibility.CIConstants.MAX_META_STRING_VALUE_LENGTH;
+import static datadog.trace.util.Strings.truncate;
 
 import datadog.communication.serialization.GrowableBuffer;
 import datadog.communication.serialization.Writable;
@@ -350,21 +352,22 @@ public void accept(Metadata metadata) {
       // we just need to be sure that the size is the same as the number of elements
       for (Map.Entry<String, String> entry : metadata.getBaggage().entrySet()) {
         writable.writeString(entry.getKey(), null);
-        writable.writeString(entry.getValue(), null);
+        writable.writeString(truncate(entry.getValue(), MAX_META_STRING_VALUE_LENGTH), null);
       }
       if (null != metadata.getHttpStatusCode()) {
         writable.writeUTF8(HTTP_STATUS);
-        writable.writeUTF8(metadata.getHttpStatusCode());
+        writable.writeUTF8(truncate(metadata.getHttpStatusCode(), MAX_META_STRING_VALUE_LENGTH));
       }
       for (Map.Entry<String, Object> entry : tags.entrySet()) {
         Object value = entry.getValue();
         if (!(value instanceof Number)) {
           writable.writeString(entry.getKey(), null);
           if (!(value instanceof Iterable)) {
-            writable.writeObjectString(value, null);
+            writable.writeString(
+                truncate(String.valueOf(value), MAX_META_STRING_VALUE_LENGTH), null);
           } else {
             String serializedValue = toJson((Collection<String>) value);
-            writable.writeString(serializedValue, null);
+            writable.writeString(truncate(serializedValue, MAX_META_STRING_VALUE_LENGTH), null);
           }
         }
       }

dd-trace-core/src/main/java/datadog/trace/common/writer/FileBasedPayloadDispatcher.java

@@ -1,6 +1,8 @@
 package datadog.trace.common.writer;
 
 import static datadog.json.JsonMapper.toJson;
+import static datadog.trace.api.civisibility.CIConstants.MAX_META_STRING_VALUE_LENGTH;
+import static datadog.trace.util.Strings.truncate;
 
 import datadog.json.JsonWriter;
 import datadog.trace.api.Config;
@@ -383,20 +385,21 @@ public void accept(Metadata metadata) {
       w.beginObject();
       for (Map.Entry<String, String> entry : metadata.getBaggage().entrySet()) {
         if (!isExcludedTag(entry.getKey())) {
-          w.name(entry.getKey()).value(entry.getValue());
+          w.name(entry.getKey()).value(truncate(entry.getValue(), MAX_META_STRING_VALUE_LENGTH));
         }
       }
       if (metadata.getHttpStatusCode() != null) {
-        w.name(Tags.HTTP_STATUS).value(metadata.getHttpStatusCode().toString());
+        w.name(Tags.HTTP_STATUS)
+            .value(truncate(metadata.getHttpStatusCode().toString(), MAX_META_STRING_VALUE_LENGTH));
       }
       for (Map.Entry<String, Object> entry : tags.entrySet()) {
         Object value = entry.getValue();
         if (!(value instanceof Number) && !isExcludedTag(entry.getKey())) {
           w.name(entry.getKey());
           if (value instanceof Iterable) {
-            w.value(toJson((Collection<String>) value));
+            w.value(truncate(toJson((Collection<String>) value), MAX_META_STRING_VALUE_LENGTH));
           } else {
-            w.value(String.valueOf(value));
+            w.value(truncate(String.valueOf(value), MAX_META_STRING_VALUE_LENGTH));
           }
         }
       }

dd-trace-core/src/test/groovy/datadog/trace/civisibility/writer/ddintake/CiTestCycleMapperV1PayloadTest.groovy

@@ -23,6 +23,8 @@ import java.nio.ByteBuffer
 import java.nio.channels.WritableByteChannel
 
 import static datadog.trace.bootstrap.instrumentation.api.InstrumentationTags.DD_MEASURED
+import static datadog.trace.api.civisibility.CIConstants.MAX_META_STRING_VALUE_LENGTH
+import static datadog.trace.util.Strings.truncate
 import static datadog.trace.common.writer.TraceGenerator.generateRandomSpan
 import static datadog.trace.common.writer.TraceGenerator.generateRandomTraces
 import static org.junit.jupiter.api.Assertions.assertEquals
@@ -110,6 +112,56 @@ class CiTestCycleMapperV1PayloadTest extends DDSpecification {
     assert spanContent.containsKey("parent_id")
   }
 
+  def "truncates meta string values and preserves metrics and top level ids"() {
+    setup:
+    String longValue = "a" * (MAX_META_STRING_VALUE_LENGTH + 1)
+    String exactValue = "b" * MAX_META_STRING_VALUE_LENGTH
+    def span = generateRandomSpan(InternalSpanTypes.TEST, [
+      (Tags.TEST_SESSION_ID): DDTraceId.from(123),
+      (Tags.TEST_MODULE_ID) : 456,
+      (Tags.TEST_SUITE_ID)  : 789,
+      "custom.tag"          : longValue,
+      "exact.tag"           : exactValue,
+      "custom.metric"       : 42,
+    ])
+
+    when:
+    Map<String, Object> deserializedSpan = whenASpanIsWritten(span)
+
+    then:
+    verifyTopLevelTags(deserializedSpan, DDTraceId.from(123), 456, 789)
+
+    def spanContent = (Map<String, Object>) deserializedSpan.get("content")
+    def deserializedMetrics = (Map<String, Object>) spanContent.get("metrics")
+    def deserializedMeta = (Map<String, Object>) spanContent.get("meta")
+
+    assert deserializedMeta.get("custom.tag") == longValue.substring(0, MAX_META_STRING_VALUE_LENGTH)
+    assert deserializedMeta.get("custom.tag").length() == MAX_META_STRING_VALUE_LENGTH
+    assert deserializedMeta.get("exact.tag") == exactValue
+    assert deserializedMetrics.get("custom.metric") == 42
+  }
+
+  def "truncates payload metadata values"() {
+    setup:
+    String longValue = "m" * (MAX_META_STRING_VALUE_LENGTH + 1)
+    CiVisibilityWellKnownTags wellKnownTags = new CiVisibilityWellKnownTags(
+      longValue, longValue, longValue,
+      longValue, longValue, longValue,
+      longValue, longValue, longValue, longValue)
+    CiTestCycleMapperV1 mapper = new CiTestCycleMapperV1(wellKnownTags, false)
+    List<List<TraceGenerator.PojoSpan>> traces = Collections.singletonList(
+      Collections.singletonList(generateRandomSpan(InternalSpanTypes.TEST, Collections.emptyMap())))
+    PayloadVerifier verifier = new PayloadVerifier(wellKnownTags, traces, mapper)
+    MsgPackWriter packer = new MsgPackWriter(new FlushingBuffer(100 << 10, verifier))
+
+    when:
+    packer.format(traces.get(0), mapper)
+    packer.flush()
+
+    then:
+    verifier.verifyTracesConsumed()
+  }
+
   def "verify test_suite_end event is written correctly"() {
     setup:
     def span = generateRandomSpan(InternalSpanTypes.TEST_SUITE_END, [
@@ -275,25 +327,25 @@ class CiTestCycleMapperV1PayloadTest extends DDSpecification {
 
         assertEquals(10, unpacker.unpackMapHeader())
         assertEquals("env", unpacker.unpackString())
-        assertEquals(wellKnownTags.env as String, unpacker.unpackString())
+        assertEquals(truncate(wellKnownTags.env as String, MAX_META_STRING_VALUE_LENGTH), unpacker.unpackString())
         assertEquals("runtime-id", unpacker.unpackString())
-        assertEquals(wellKnownTags.runtimeId as String, unpacker.unpackString())
+        assertEquals(truncate(wellKnownTags.runtimeId as String, MAX_META_STRING_VALUE_LENGTH), unpacker.unpackString())
         assertEquals("language", unpacker.unpackString())
-        assertEquals(wellKnownTags.language as String, unpacker.unpackString())
+        assertEquals(truncate(wellKnownTags.language as String, MAX_META_STRING_VALUE_LENGTH), unpacker.unpackString())
         assertEquals(Tags.RUNTIME_NAME, unpacker.unpackString())
-        assertEquals(wellKnownTags.runtimeName as String, unpacker.unpackString())
+        assertEquals(truncate(wellKnownTags.runtimeName as String, MAX_META_STRING_VALUE_LENGTH), unpacker.unpackString())
         assertEquals(Tags.RUNTIME_VENDOR, unpacker.unpackString())
-        assertEquals(wellKnownTags.runtimeVendor as String, unpacker.unpackString())
+        assertEquals(truncate(wellKnownTags.runtimeVendor as String, MAX_META_STRING_VALUE_LENGTH), unpacker.unpackString())
         assertEquals(Tags.RUNTIME_VERSION, unpacker.unpackString())
-        assertEquals(wellKnownTags.runtimeVersion as String, unpacker.unpackString())
+        assertEquals(truncate(wellKnownTags.runtimeVersion as String, MAX_META_STRING_VALUE_LENGTH), unpacker.unpackString())
         assertEquals(Tags.OS_ARCHITECTURE, unpacker.unpackString())
-        assertEquals(wellKnownTags.osArch as String, unpacker.unpackString())
+        assertEquals(truncate(wellKnownTags.osArch as String, MAX_META_STRING_VALUE_LENGTH), unpacker.unpackString())
         assertEquals(Tags.OS_PLATFORM, unpacker.unpackString())
-        assertEquals(wellKnownTags.osPlatform as String, unpacker.unpackString())
+        assertEquals(truncate(wellKnownTags.osPlatform as String, MAX_META_STRING_VALUE_LENGTH), unpacker.unpackString())
         assertEquals(Tags.OS_VERSION, unpacker.unpackString())
-        assertEquals(wellKnownTags.osVersion as String, unpacker.unpackString())
+        assertEquals(truncate(wellKnownTags.osVersion as String, MAX_META_STRING_VALUE_LENGTH), unpacker.unpackString())
         assertEquals(DDTags.TEST_IS_USER_PROVIDED_SERVICE, unpacker.unpackString())
-        assertEquals(wellKnownTags.isUserProvidedService as String, unpacker.unpackString())
+        assertEquals(truncate(wellKnownTags.isUserProvidedService as String, MAX_META_STRING_VALUE_LENGTH), unpacker.unpackString())
 
         assertEquals("events", unpacker.unpackString())
 
@@ -307,7 +359,7 @@ class CiTestCycleMapperV1PayloadTest extends DDSpecification {
           TraceGenerator.PojoSpan expectedSpan = expectedTrace.get(k)
           assertEquals(3, unpacker.unpackMapHeader())
           assertEquals("type", unpacker.unpackString())
-          if ("test" == expectedSpan.getType()) {
+          if ("test" == String.valueOf(expectedSpan.getType())) {
             assertEquals("test", unpacker.unpackString())
           } else {
             assertEquals("span", unpacker.unpackString())

dd-trace-core/src/test/java/datadog/trace/common/writer/FileBasedPayloadDispatcherTest.java

@@ -12,6 +12,7 @@
 import com.fasterxml.jackson.databind.ObjectMapper;
 import datadog.trace.api.DDTraceId;
 import datadog.trace.api.TagMap;
+import datadog.trace.api.civisibility.CIConstants;
 import datadog.trace.api.intake.TrackType;
 import datadog.trace.bootstrap.instrumentation.api.InternalSpanTypes;
 import datadog.trace.bootstrap.instrumentation.api.Tags;
@@ -143,6 +144,45 @@ void citestcycleStripsCiGitOsRuntimeTagsAndWellKnownFields(@TempDir Path outputD
     assertEquals(99L, metrics.get("kept.metric").asLong());
   }
 
+  @Test
+  void citestcycleTruncatesMetaValuesAndPreservesMetricsAndTopLevelIds(@TempDir Path outputDir)
+      throws IOException {
+    FileBasedPayloadDispatcher dispatcher =
+        new FileBasedPayloadDispatcher(outputDir.toString(), "tests", TrackType.CITESTCYCLE);
+    String longValue = longString(CIConstants.MAX_META_STRING_VALUE_LENGTH + 1, 'a');
+    String exactValue = longString(CIConstants.MAX_META_STRING_VALUE_LENGTH, 'b');
+    Map<String, Object> tags = new HashMap<>();
+    tags.put(Tags.TEST_SESSION_ID, DDTraceId.from(123));
+    tags.put(Tags.TEST_MODULE_ID, 456L);
+    tags.put(Tags.TEST_SUITE_ID, 789L);
+    tags.put("custom.tag", longValue);
+    tags.put("exact.tag", exactValue);
+    tags.put("custom.metric", 42L);
+    CoreSpan<?> span = mockSpan(InternalSpanTypes.TEST, tags);
+
+    dispatcher.addTrace(Collections.singletonList(span));
+    dispatcher.flush();
+
+    JsonNode content =
+        JSON.readTree(listFiles(outputDir).get(0).toFile()).get("events").get(0).get("content");
+    JsonNode meta = content.get("meta");
+    JsonNode metrics = content.get("metrics");
+
+    assertEquals(
+        longValue.substring(0, CIConstants.MAX_META_STRING_VALUE_LENGTH),
+        meta.get("custom.tag").asText());
+    assertEquals(
+        CIConstants.MAX_META_STRING_VALUE_LENGTH, meta.get("custom.tag").asText().length());
+    assertEquals(exactValue, meta.get("exact.tag").asText());
+    assertEquals(42L, metrics.get("custom.metric").asLong());
+    assertFalse(meta.has(Tags.TEST_SESSION_ID));
+    assertFalse(meta.has(Tags.TEST_MODULE_ID));
+    assertFalse(meta.has(Tags.TEST_SUITE_ID));
+    assertEquals(123L, content.get(Tags.TEST_SESSION_ID).asLong());
+    assertEquals(456L, content.get(Tags.TEST_MODULE_ID).asLong());
+    assertEquals(789L, content.get(Tags.TEST_SUITE_ID).asLong());
+  }
+
   @Test
   void citestcycleAssignsEventTypesForSessionModuleSuiteTestSpanSpans(@TempDir Path outputDir)
       throws IOException {
@@ -295,4 +335,10 @@ private static List<Path> listFiles(Path dir) throws IOException {
     }
     return files;
   }
+
+  private static String longString(int length, char value) {
+    char[] chars = new char[length];
+    Arrays.fill(chars, value);
+    return new String(chars);
+  }
 }

internal-api/src/main/java/datadog/trace/api/Config.java

@@ -720,6 +720,7 @@
 import datadog.environment.JavaVirtualMachine;
 import datadog.environment.OperatingSystem;
 import datadog.environment.SystemProperties;
+import datadog.trace.api.civisibility.CIConstants;
 import datadog.trace.api.civisibility.CiVisibilityWellKnownTags;
 import datadog.trace.api.config.GeneralConfig;
 import datadog.trace.api.config.OtlpConfig;
@@ -3250,7 +3251,7 @@ PROFILING_DATADOG_PROFILER_ENABLED, isDatadogProfilerSafeInCurrentEnvironment())
 
     int defaultStackTraceLengthLimit =
         instrumenterConfig.isCiVisibilityEnabled()
-            ? 5000 // EVP limit
+            ? CIConstants.MAX_META_STRING_VALUE_LENGTH // EVP limit
             : Integer.MAX_VALUE; // no effective limit (old behavior)
     this.stackTraceLengthLimit =
         configProvider.getInteger(STACK_TRACE_LENGTH_LIMIT, defaultStackTraceLengthLimit);

internal-api/src/main/java/datadog/trace/api/civisibility/CIConstants.java

@@ -2,6 +2,12 @@
 
 public interface CIConstants {
 
+  /**
+   * Maximum length (in characters) of a meta string value sent to the CI Visibility intake; longer
+   * values are truncated. Matches the Event Platform (EVP) per-tag-value limit.
+   */
+  int MAX_META_STRING_VALUE_LENGTH = 5000;
+
   String SELENIUM_BROWSER_DRIVER = "selenium";
 
   String FAIL_FAST_TEST_ORDER = "FAILFAST";

internal-api/src/main/java/datadog/trace/api/civisibility/CiVisibilityWellKnownTags.java

@@ -1,6 +1,7 @@
 package datadog.trace.api.civisibility;
 
 import datadog.trace.bootstrap.instrumentation.api.UTF8BytesString;
+import datadog.trace.util.Strings;
 
 public class CiVisibilityWellKnownTags {
 
@@ -26,16 +27,25 @@ public CiVisibilityWellKnownTags(
       CharSequence osPlatform,
       CharSequence osVersion,
       CharSequence isUserProvidedService) {
-    this.runtimeId = UTF8BytesString.create(runtimeId);
-    this.env = UTF8BytesString.create(env);
-    this.language = UTF8BytesString.create(language);
-    this.runtimeName = UTF8BytesString.create(runtimeName);
-    this.runtimeVersion = UTF8BytesString.create(runtimeVersion);
-    this.runtimeVendor = UTF8BytesString.create(runtimeVendor);
-    this.osArch = UTF8BytesString.create(osArch);
-    this.osPlatform = UTF8BytesString.create(osPlatform);
-    this.osVersion = UTF8BytesString.create(osVersion);
-    this.isUserProvidedService = UTF8BytesString.create(isUserProvidedService);
+    this.runtimeId = truncated(runtimeId);
+    this.env = truncated(env);
+    this.language = truncated(language);
+    this.runtimeName = truncated(runtimeName);
+    this.runtimeVersion = truncated(runtimeVersion);
+    this.runtimeVendor = truncated(runtimeVendor);
+    this.osArch = truncated(osArch);
+    this.osPlatform = truncated(osPlatform);
+    this.osVersion = truncated(osVersion);
+    this.isUserProvidedService = truncated(isUserProvidedService);
+  }
+
+  /**
+   * Truncates a well-known tag value to the EVP per-value limit once, up front, and stores it
+   * pre-encoded so the intake serializers can marshal it as-is on every payload without truncating.
+   */
+  private static UTF8BytesString truncated(CharSequence value) {
+    return UTF8BytesString.create(
+        Strings.truncate(value, CIConstants.MAX_META_STRING_VALUE_LENGTH));
   }
 
   public UTF8BytesString getEnv() {

internal-api/src/main/java/datadog/trace/util/Strings.java

@@ -2,6 +2,7 @@
 
 import static java.nio.charset.StandardCharsets.US_ASCII;
 
+import datadog.trace.bootstrap.instrumentation.api.UTF8BytesString;
 import java.nio.charset.StandardCharsets;
 import java.security.MessageDigest;
 import java.security.NoSuchAlgorithmException;
@@ -92,6 +93,18 @@ public static CharSequence truncate(CharSequence input, int limit) {
     return input.subSequence(0, limit);
   }
 
+  /**
+   * Truncates a pre-encoded {@link UTF8BytesString}. Returns the same instance when within the
+   * limit, so callers writing it back out keep the zero-copy fast path; only the rare over-limit
+   * case re-encodes the truncated value.
+   */
+  public static UTF8BytesString truncate(UTF8BytesString input, int limit) {
+    if (input == null || input.length() <= limit) {
+      return input;
+    }
+    return UTF8BytesString.create(input.subSequence(0, limit));
+  }
+
   /**
    * Checks that a string is not blank, i.e. contains at least one character that is not a
    * whitespace