pkg/trace/info: separate payload_too_large from decoding_error

pkg/trace/api/api.go

@@ -370,7 +370,11 @@ func (r *HTTPReceiver) handleTraces(v Version, w http.ResponseWriter, req *http.
  traces, err := r.decodeTraces(v, req)
  if err != nil {
  httpDecodingError(err, []string{tagTraceHandler, fmt.Sprintf("v:%s", v)}, w)
- atomic.AddInt64(&ts.TracesDropped.DecodingError, traceCount)
+ if err == ErrLimitedReaderLimitReached {
+ atomic.AddInt64(&ts.TracesDropped.PayloadTooLarge, traceCount)
+ } else {
+ atomic.AddInt64(&ts.TracesDropped.DecodingError, traceCount)
+ }
  log.Errorf("Cannot decode %s traces payload: %v", v, err)
  return
  }

pkg/trace/info/stats.go

@@ -191,6 +191,9 @@ func mapToString(m map[string]int64) string {
 type TracesDropped struct {
  // DecodingError is when the agent fails to decode a trace payload
  DecodingError int64
+ // PayloadTooLarge specifies the number of traces dropped due to the payload
+ // being too large to be accepted.
+ PayloadTooLarge int64
  // EmptyTrace is when the trace contains no spans
  EmptyTrace int64
  // TraceIDZero is when any spans in a trace have TraceId=0
@@ -204,11 +207,12 @@ type TracesDropped struct {
 // tagValues converts TracesDropped into a map representation with keys matching standardized names for all reasons
 func (s *TracesDropped) tagValues() map[string]int64 {
  return map[string]int64{
- "decoding_error": atomic.LoadInt64(&s.DecodingError),
- "empty_trace": atomic.LoadInt64(&s.EmptyTrace),
- "trace_id_zero": atomic.LoadInt64(&s.TraceIDZero),
- "span_id_zero": atomic.LoadInt64(&s.SpanIDZero),
- "foreign_span": atomic.LoadInt64(&s.ForeignSpan),
+ "payload_too_large": atomic.LoadInt64(&s.PayloadTooLarge),
+ "decoding_error": atomic.LoadInt64(&s.DecodingError),
+ "empty_trace": atomic.LoadInt64(&s.EmptyTrace),
+ "trace_id_zero": atomic.LoadInt64(&s.TraceIDZero),
+ "span_id_zero": atomic.LoadInt64(&s.SpanIDZero),
+ "foreign_span": atomic.LoadInt64(&s.ForeignSpan),
  }
 }
 

pkg/trace/info/stats_test.go

@@ -21,11 +21,12 @@ func TestTracesDropped(t *testing.T) {
 
  t.Run("tagValues", func(t *testing.T) {
  assert.Equal(t, map[string]int64{
- "empty_trace": 0,
- "decoding_error": 1,
- "foreign_span": 1,
- "trace_id_zero": 1,
- "span_id_zero": 1,
+ "empty_trace": 0,
+ "payload_too_large": 0,
+ "decoding_error": 1,
+ "foreign_span": 1,
+ "trace_id_zero": 1,
+ "span_id_zero": 1,
  }, s.tagValues())
  })
 

releasenotes/notes/apm-reason-payload-too-large-563bd2a463d41500.yaml

@@ -0,0 +1,12 @@
+# Each section from every releasenote are combined when the
+# CHANGELOG.rst is rendered. So the text needs to be worded so that
+# it does not depend on any information only available in another
+# section. This may mean repeating some details, but each section
+# must be readable independently of the other.
+#
+# Each section note must be formatted as reStructuredText.
+---
+enhancements:
+ - |
+ APM: The `datadog.trace_agent.normalizer.traces_dropped` metric now has a new
+ reason `payload_too_large` which was confusingly merged with `decoding_error`.