[AWS Fargate] Add memory hard limit from container metadata and remov…

…e usage percentage (#37194)

CHANGELOG.next.asciidoc

@@ -124,6 +124,7 @@ https://github.com/elastic/beats/compare/v8.8.1\...main[Check the HEAD diff]
 - Fix memory leak on Windows {issue}37142[37142] {pull}37171[37171]
 - Fix unintended skip in metric collection on Azure Monitor {issue}37204[37204] {pull}37203[37203]
 - Fix the "api-version query parameter (?api-version=) is required for all requests" error in Azure Billing. {pull}37158[37158]
+- Add memory hard limit from container metadata and remove usage percentage in AWS Fargate. {pull}37194[37194]
 
 *Osquerybeat*
 

metricbeat/docs/fields.asciidoc

@@ -4986,6 +4986,16 @@ type: keyword
 
 --
 
+*`awsfargate.task_stats.memory_hard_limit`*::
++
+--
+The Hard Memory Limit for the task from Amazon ECS.
+
+
+type: scaled_float
+
+--
+
 [float]
 === cpu
 
@@ -5501,18 +5511,6 @@ format: bytes
 
 --
 
-*`awsfargate.task_stats.memory.usage.pct`*::
-+
---
-Memory usage percentage.
-
-
-type: scaled_float
-
-format: percent
-
---
-
 *`awsfargate.task_stats.memory.usage.total`*::
 +
 --

x-pack/metricbeat/module/awsfargate/task_stats/_meta/data.json

@@ -108,7 +108,6 @@
                 },
                 "usage": {
                     "max": 15294464,
-                    "pct": 0.003136136404770672,
                     "total": 12349440
                 }
             },

x-pack/metricbeat/module/awsfargate/task_stats/_meta/fields.yml

@@ -22,6 +22,10 @@
       type: keyword
       description: >
         The known status for the task from Amazon ECS.
+    - name: memory_hard_limit
+      type: scaled_float
+      description: >
+        The Hard Memory Limit for the task from Amazon ECS.
     - name: cpu
       type: group
       description: Runtime CPU metrics.
@@ -251,11 +255,6 @@
               format: bytes
               description: >
                 Max memory usage.
-            - name: pct
-              type: scaled_float
-              format: percent
-              description: >
-                Memory usage percentage.
             - name: total
               type: long
               format: bytes

x-pack/metricbeat/module/awsfargate/task_stats/_meta/testdata/task.json

@@ -5,6 +5,9 @@
     "Revision": "7",
     "DesiredStatus": "RUNNING",
     "KnownStatus": "ACTIVATING",
+    "Limits": {
+      "Memory": 7168
+    },
     "Containers": [{
         "DockerId": "1234",
         "Name": "query-metadata",
@@ -14,6 +17,9 @@
             "com.amazonaws.ecs.container-name": "query-metadata",
             "com.amazonaws.ecs.task-arn": "arn:aws:ecs:us-west-2:111122223333:task/default/febee046097849aba589d4435207c04a",
             "com.amazonaws.ecs.task-definition-family": "query-metadata",
-            "com.amazonaws.ecs.task-definition-version": "7"}
+            "com.amazonaws.ecs.task-definition-version": "7"},
+        "Limits": {
+            "Memory": 3328
+        }
     }]
-}
+}

x-pack/metricbeat/module/awsfargate/task_stats/container.go

@@ -15,6 +15,7 @@ type container struct {
 	Name     string
 	Image    string
 	Labels   map[string]string
+	Limits   Limits
 }
 
 func getContainerMetadata(c *container) *container {
@@ -23,6 +24,7 @@ func getContainerMetadata(c *container) *container {
 		Image:    c.Image,
 		Name:     helpers.ExtractContainerName([]string{c.Name}),
 		Labels:   deDotLabels(c.Labels),
+		Limits:   c.Limits,
 	}
 }
 

x-pack/metricbeat/module/awsfargate/task_stats/data.go

@@ -57,6 +57,11 @@ func createEvent(stats *Stats) mb.Event {
 		_, _ = e.MetricSetFields.Put("task_known_status", taskKnownStatus)
 	}
 
+	memoryHardLimit := stats.Container.Limits.Memory
+	if memoryHardLimit != 0 {
+		_, _ = e.MetricSetFields.Put("memory_hard_limit", memoryHardLimit)
+	}
+
 	_, _ = e.MetricSetFields.Put("identifier", generateIdentifier(stats.Container.Name, stats.Container.DockerId))
 	return e
 }
@@ -163,7 +168,6 @@ func createMemoryFields(stats *Stats) mapstr.M {
 			},
 			"usage": mapstr.M{
 				"total": stats.memoryStats.Usage,
-				"pct":   stats.memoryStats.UsageP,
 				"max":   stats.memoryStats.MaxUsage,
 			},
 		}

x-pack/metricbeat/module/awsfargate/task_stats/memory.go

@@ -13,7 +13,6 @@ type memoryStats struct {
 	TotalRss  uint64
 	TotalRssP float64
 	Usage     uint64
-	UsageP    float64
 	//Raw stats from the cgroup subsystem
 	Stats map[string]uint64
 	//Windows-only memory stats
@@ -30,7 +29,6 @@ func getMemoryStats(taskStats types.StatsJSON) memoryStats {
 		MaxUsage:  taskStats.Stats.MemoryStats.MaxUsage,
 		TotalRssP: float64(totalRSS) / float64(taskStats.Stats.MemoryStats.Limit),
 		Usage:     taskStats.Stats.MemoryStats.Usage,
-		UsageP:    float64(taskStats.Stats.MemoryStats.Usage) / float64(taskStats.Stats.MemoryStats.Limit),
 		Stats:     taskStats.Stats.MemoryStats.Stats,
 		//Windows memory statistics
 		Commit:            taskStats.Stats.MemoryStats.Commit,

x-pack/metricbeat/module/awsfargate/task_stats/task_stats.go

@@ -79,9 +79,15 @@ type TaskMetadata struct {
 	Revision      string       `json:"Revision"`
 	DesiredStatus string       `json:"DesiredStatus"`
 	KnownStatus   string       `json:"KnownStatus"`
+	Limit         Limits       `json:"Limits"`
 	Containers    []*container `json:"Containers"`
 }
 
+// Limits is a struct that represents the memory limit from ${ECS_CONTAINER_METADATA_URI_V4}/task, which is the Hard Memory Limit set in AWS ECS
+type Limits struct {
+	Memory uint64 `json:"Memory"`
+}
+
 // New creates a new instance of the MetricSet. New is responsible for unpacking
 // any MetricSet specific configuration options if there are any.
 func New(base mb.BaseMetricSet) (mb.MetricSet, error) {
@@ -131,6 +137,8 @@ func (m *MetricSet) queryTaskMetadataEndpoints() ([]Stats, error) {
 	if err != nil {
 		return nil, fmt.Errorf("http.Get failed: %w", err)
 	}
+	defer taskStatsResp.Body.Close()
+
 	taskStatsOutput, err := getTaskStats(taskStatsResp)
 	if err != nil {
 		return nil, fmt.Errorf("getTaskStats failed: %w", err)
@@ -145,6 +153,8 @@ func (m *MetricSet) queryTaskMetadataEndpoints() ([]Stats, error) {
 	if err != nil {
 		return nil, fmt.Errorf("http.Get failed: %w", err)
 	}
+	defer taskResp.Body.Close()
+
 	taskOutput, err := getTask(taskResp)
 	if err != nil {
 		return nil, fmt.Errorf("getTask failed: %w", err)

x-pack/metricbeat/module/awsfargate/task_stats/task_stats_integration_test.go

@@ -51,6 +51,11 @@ func TestFetch(t *testing.T) {
 	// The goal here is to make sure every element inside the
 	// event has a matching field ("no field left behind").
 	mbtest.TestMetricsetFieldsDocumented(t, metricSet, []mb.Event{event})
+
+	t.Cleanup(func() {
+		taskStatsResp.Body.Close()
+		byteTaskResp.Body.Close()
+	})
 }
 
 func TestData(t *testing.T) {
@@ -85,6 +90,11 @@ func TestData(t *testing.T) {
 	standardizeEvent := m.StandardizeEvent(event)
 
 	mbtest.WriteEventToDataJSON(t, standardizeEvent, "")
+
+	t.Cleanup(func() {
+		taskStatsResp.Body.Close()
+		byteTaskResp.Body.Close()
+	})
 }
 
 // buildResponse is a test helper that loads the content of `filename` and returns

x-pack/metricbeat/module/awsfargate/task_stats/task_stats_test.go

@@ -6,7 +6,7 @@ package task_stats
 
 import (
 	"bytes"
-	"io/ioutil"
+	"io"
 	"net/http"
 	"testing"
 
@@ -48,6 +48,9 @@ var (
 		"Revision": "7",
         "DesiredStatus": "RUNNING",
         "KnownStatus": "ACTIVATING",
+		"Limits": {
+		  "Memory": 7168
+		},
 		"Containers": [{
 			"DockerId": "query-metadata-1",
 			"Name": "query-metadata",
@@ -57,14 +60,17 @@ var (
 				"com.amazonaws.ecs.container-name": "query-metadata",
 				"com.amazonaws.ecs.task-arn": "arn:aws:ecs:us-west-2:111122223333:task/default/febee046097849aba589d4435207c04a",
 				"com.amazonaws.ecs.task-definition-family": "query-metadata",
-				"com.amazonaws.ecs.task-definition-version": "7"}
-			}]
-		}`
+				"com.amazonaws.ecs.task-definition-version": "7"},
+			"Limits": {
+				"Memory": 3328
+			}
+		}]
+	}`
 )
 
 func TestGetTaskStats(t *testing.T) {
 	taskStatsResp := &http.Response{
-		Body: ioutil.NopCloser(bytes.NewReader([]byte(taskStatsJson))),
+		Body: io.NopCloser(bytes.NewReader([]byte(taskStatsJson))),
 	}
 
 	taskStatsOutput, err := getTaskStats(taskStatsResp)
@@ -74,7 +80,7 @@ func TestGetTaskStats(t *testing.T) {
 
 func TestGetTask(t *testing.T) {
 	taskResp := &http.Response{
-		Body: ioutil.NopCloser(bytes.NewReader([]byte(taskRespJson))),
+		Body: io.NopCloser(bytes.NewReader([]byte(taskRespJson))),
 	}
 
 	taskOutput, err := getTask(taskResp)
@@ -87,23 +93,25 @@ func TestGetTask(t *testing.T) {
 	assert.Equal(t, "RUNNING", taskOutput.DesiredStatus)
 	assert.Equal(t, "ACTIVATING", taskOutput.KnownStatus)
 
+	assert.Equal(t, uint64(7168), taskOutput.Limit.Memory)
 	assert.Equal(t, 1, len(taskOutput.Containers))
 	assert.Equal(t, "query-metadata-1", taskOutput.Containers[0].DockerId)
 	assert.Equal(t, "query-metadata", taskOutput.Containers[0].Name)
 	assert.Equal(t, "mreferre/eksutils", taskOutput.Containers[0].Image)
 	assert.Equal(t, 5, len(taskOutput.Containers[0].Labels))
+	assert.Equal(t, uint64(3328), taskOutput.Containers[0].Limits.Memory)
 }
 
 func TestGetStatsList(t *testing.T) {
 	taskStatsResp := &http.Response{
-		Body: ioutil.NopCloser(bytes.NewReader([]byte(taskStatsJson))),
+		Body: io.NopCloser(bytes.NewReader([]byte(taskStatsJson))),
 	}
 
 	taskStatsOutput, err := getTaskStats(taskStatsResp)
 	assert.NoError(t, err)
 
 	taskResp := &http.Response{
-		Body: ioutil.NopCloser(bytes.NewReader([]byte(taskRespJson))),
+		Body: io.NopCloser(bytes.NewReader([]byte(taskRespJson))),
 	}
 
 	taskOutput, err := getTask(taskResp)
@@ -115,7 +123,7 @@ func TestGetStatsList(t *testing.T) {
 
 func TestGetCPUStats(t *testing.T) {
 	taskStatsResp := &http.Response{
-		Body: ioutil.NopCloser(bytes.NewReader([]byte(taskStatsJson))),
+		Body: io.NopCloser(bytes.NewReader([]byte(taskStatsJson))),
 	}
 
 	taskStatsOutput, err := getTaskStats(taskStatsResp)