feat(inputs.statsd): Add support for DogStatsD v1.2 (#12991) (#15380)

plugins/inputs/statsd/README.md

@@ -81,6 +81,11 @@ See the [CONFIGURATION.md][CONFIGURATION.md] for more details.
   ## https://docs.datadoghq.com/developers/metrics/types/?tab=distribution#definition
   datadog_distributions = false
 
+  ## Keep or drop the container id as tag. Included as optional field
+  ## in DogStatsD protocol v1.2 if source is running in Kubernetes
+  ## https://docs.datadoghq.com/developers/dogstatsd/datagram_shell/?tab=metrics#dogstatsd-protocol-v12
+  datadog_keep_container_tag = false
+
   ## Statsd data translation templates, more info can be read here:
   ## https://github.com/influxdata/telegraf/blob/master/docs/TEMPLATE_PATTERN.md
   # templates = [
@@ -259,6 +264,7 @@ measurements and tags.
 - **parse_data_dog_tags** boolean: Enable parsing of tags in DataDog's dogstatsd format (<http://docs.datadoghq.com/guides/dogstatsd/>)
 - **datadog_extensions** boolean: Enable parsing of DataDog's extensions to dogstatsd format (<http://docs.datadoghq.com/guides/dogstatsd/>)
 - **datadog_distributions** boolean: Enable parsing of the Distribution metric in DataDog's dogstatsd format (<https://docs.datadoghq.com/developers/metrics/types/?tab=distribution#definition>)
+- **datadog_keep_container_tag** boolean: Keep or drop the container id as tag. Included as optional field in DogStatsD protocol v1.2 if source is running in Kubernetes.
 - **max_ttl** config.Duration: Max duration (TTL) for each metric to stay cached/reported without being updated.
 
 ## Statsd bucket -> InfluxDB line-protocol Templates

plugins/inputs/statsd/sample.conf

@@ -54,6 +54,11 @@
   ## https://docs.datadoghq.com/developers/metrics/types/?tab=distribution#definition
   datadog_distributions = false
 
+  ## Keep or drop the container id as tag. Included as optional field
+  ## in DogStatsD protocol v1.2 if source is running in Kubernetes
+  ## https://docs.datadoghq.com/developers/dogstatsd/datagram_shell/?tab=metrics#dogstatsd-protocol-v12
+  datadog_keep_container_tag = false
+
   ## Statsd data translation templates, more info can be read here:
   ## https://github.com/influxdata/telegraf/blob/master/docs/TEMPLATE_PATTERN.md
   # templates = [

plugins/inputs/statsd/statsd.go

@@ -96,6 +96,11 @@ type Statsd struct {
 	// https://docs.datadoghq.com/developers/metrics/types/?tab=distribution#definition
 	DataDogDistributions bool `toml:"datadog_distributions"`
 
+	// Either to keep or drop the container id as tag.
+	// Requires the DataDogExtension flag to be enabled.
+	// https://docs.datadoghq.com/developers/dogstatsd/datagram_shell/?tab=metrics#dogstatsd-protocol-v12
+	DataDogKeepContainerTag bool `toml:"datadog_keep_container_tag"`
+
 	// UDPPacketSize is deprecated, it's only here for legacy support
 	// we now always create 1 max size buffer and then copy only what we need
 	// into the in channel
@@ -577,6 +582,11 @@ func (s *Statsd) parseStatsdLine(line string) error {
 			if len(segment) > 0 && segment[0] == '#' {
 				// we have ourselves a tag; they are comma separated
 				parseDataDogTags(lineTags, segment[1:])
+			} else if len(segment) > 0 && strings.HasPrefix(segment, "c:") {
+				// This is optional container ID field
+				if s.DataDogKeepContainerTag {
+					lineTags["container"] = segment[2:]
+				}
 			} else {
 				recombinedSegments = append(recombinedSegments, segment)
 			}

plugins/inputs/statsd/statsd_test.go

@@ -1206,6 +1206,157 @@ func TestParse_DataDogTags(t *testing.T) {
 	}
 }
 
+func TestParse_DataDogContainerID(t *testing.T) {
+	tests := []struct {
+		name     string
+		line     string
+		keep     bool
+		expected []telegraf.Metric
+	}{
+		{
+			name: "counter",
+			line: "my_counter:1|c|#host:localhost,endpoint:/:tenant?/oauth/ro|c:f76b5a1c03caa192580874b253c158010ade668cf03080a57aa8283919d56e75",
+			keep: true,
+			expected: []telegraf.Metric{
+				testutil.MustMetric(
+					"my_counter",
+					map[string]string{
+						"endpoint":    "/:tenant?/oauth/ro",
+						"host":        "localhost",
+						"metric_type": "counter",
+						"container":   "f76b5a1c03caa192580874b253c158010ade668cf03080a57aa8283919d56e75",
+					},
+					map[string]interface{}{
+						"value": 1,
+					},
+					time.Now(),
+					telegraf.Counter,
+				),
+			},
+		},
+		{
+			name: "gauge",
+			line: "my_gauge:10.1|g|#live|c:f76b5a1c03caa192580874b253c158010ade668cf03080a57aa8283919d56e75",
+			keep: true,
+			expected: []telegraf.Metric{
+				testutil.MustMetric(
+					"my_gauge",
+					map[string]string{
+						"live":        "true",
+						"metric_type": "gauge",
+						"container":   "f76b5a1c03caa192580874b253c158010ade668cf03080a57aa8283919d56e75",
+					},
+					map[string]interface{}{
+						"value": 10.1,
+					},
+					time.Now(),
+					telegraf.Gauge,
+				),
+			},
+		},
+		{
+			name: "set",
+			line: "my_set:1|s|#host:localhost|c:f76b5a1c03caa192580874b253c158010ade668cf03080a57aa8283919d56e75",
+			keep: true,
+			expected: []telegraf.Metric{
+				testutil.MustMetric(
+					"my_set",
+					map[string]string{
+						"host":        "localhost",
+						"metric_type": "set",
+						"container":   "f76b5a1c03caa192580874b253c158010ade668cf03080a57aa8283919d56e75",
+					},
+					map[string]interface{}{
+						"value": 1,
+					},
+					time.Now(),
+				),
+			},
+		},
+		{
+			name: "timer",
+			line: "my_timer:3|ms|@0.1|#live,host:localhost|c:f76b5a1c03caa192580874b253c158010ade668cf03080a57aa8283919d56e75",
+			keep: true,
+			expected: []telegraf.Metric{
+				testutil.MustMetric(
+					"my_timer",
+					map[string]string{
+						"host":        "localhost",
+						"live":        "true",
+						"metric_type": "timing",
+						"container":   "f76b5a1c03caa192580874b253c158010ade668cf03080a57aa8283919d56e75",
+					},
+					map[string]interface{}{
+						"count":  10,
+						"lower":  float64(3),
+						"mean":   float64(3),
+						"median": float64(3),
+						"stddev": float64(0),
+						"sum":    float64(30),
+						"upper":  float64(3),
+					},
+					time.Now(),
+				),
+			},
+		},
+		{
+			name: "empty tag set",
+			line: "cpu:42|c|#|c:f76b5a1c03caa192580874b253c158010ade668cf03080a57aa8283919d56e75",
+			keep: true,
+			expected: []telegraf.Metric{
+				testutil.MustMetric(
+					"cpu",
+					map[string]string{
+						"metric_type": "counter",
+						"container":   "f76b5a1c03caa192580874b253c158010ade668cf03080a57aa8283919d56e75",
+					},
+					map[string]interface{}{
+						"value": 42,
+					},
+					time.Now(),
+					telegraf.Counter,
+				),
+			},
+		},
+		{
+			name: "drop it",
+			line: "cpu:42|c|#live,host:localhost|c:f76b5a1c03caa192580874b253c158010ade668cf03080a57aa8283919d56e75",
+			keep: false,
+			expected: []telegraf.Metric{
+				testutil.MustMetric(
+					"cpu",
+					map[string]string{
+						"host":        "localhost",
+						"live":        "true",
+						"metric_type": "counter",
+					},
+					map[string]interface{}{
+						"value": 42,
+					},
+					time.Now(),
+					telegraf.Counter,
+				),
+			},
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			var acc testutil.Accumulator
+
+			s := NewTestStatsd()
+			s.DataDogExtensions = true
+			s.DataDogKeepContainerTag = tt.keep
+
+			require.NoError(t, s.parseStatsdLine(tt.line))
+			require.NoError(t, s.Gather(&acc))
+
+			testutil.RequireMetricsEqual(t, tt.expected, acc.GetTelegrafMetrics(),
+				testutil.SortMetrics(), testutil.IgnoreTime())
+		})
+	}
+}
+
 // Test that statsd buckets are parsed to measurement names properly
 func TestParseName(t *testing.T) {
 	s := NewTestStatsd()