Fetch cloud metadata

Based on the Python reference implementation:
elastic/apm-agent-python#826

By default we attempt to sniff metadata for all cloud
providers. This can be overridden by setting the
environment variable ELASTIC_APM_CLOUD_PROVIDER to one
of "none", "aws", "azure", or "gcp".

We set a short (100ms) socket connect timeout, and a
slightly longer overall timeout (1s) for fetching the
cloud metadata.

In tests we disable cloud metadata fetching by default.
There are currently no functional tests, as that would
rely on us running in a known cloud environment.

apmtest/env.go

@@ -0,0 +1,8 @@
+package apmtest
+
+import "os"
+
+func init() {
+	// Disable cloud metadata sniffing by default in tests.
+	os.Setenv("ELASTIC_APM_CLOUD_PROVIDER", "none")
+}

config.go

@@ -57,6 +57,7 @@ const (
 	envCentralConfig               = "ELASTIC_APM_CENTRAL_CONFIG"
 	envBreakdownMetrics            = "ELASTIC_APM_BREAKDOWN_METRICS"
 	envUseElasticTraceparentHeader = "ELASTIC_APM_USE_ELASTIC_TRACEPARENT_HEADER"
+	envCloudProvider               = "ELASTIC_APM_CLOUD_PROVIDER"
 
 	// NOTE(axw) profiling environment variables are experimental.
 	// They may be removed in a future minor version without being

docs/configuration.asciidoc

@@ -593,3 +593,20 @@ https://www.w3.org/TR/trace-context-1/[W3C Trace Context] specification.
 
 When this setting is `true`, the agent will also add the header `elastic-apm-traceparent`
 for backwards compatibility with older versions of Elastic APM agents.
+
+[float]
+[[config-cloud-provider]]
+==== `ELASTIC_APM_CLOUD_PROVIDER`
+
+[options="header"]
+|============
+| Environment                  | Default  | Example
+| `ELASTIC_APM_CLOUD_PROVIDER` | `"none"` | `"aws"`
+|============
+
+This config value allows you to specify which cloud provider should be assumed
+for metadata collection. By default, the agent will use trial and error to
+automatically collect the cloud metadata.
+
+Valid options are `"none"`, `"auto"`, `"aws"`, `"gcp"`, and `"azure"`
+If this config value is set to `"none"`, then no cloud metadata will be collected.

internal/apmcloudutil/aws.go

@@ -0,0 +1,83 @@
+package apmcloudutil
+
+import (
+	"context"
+	"encoding/json"
+	"errors"
+	"io/ioutil"
+	"net/http"
+
+	"go.elastic.co/apm/model"
+)
+
+const (
+	ec2TokenURL    = "http://169.254.169.254/latest/api/token"
+	ec2MetadataURL = "http://169.254.169.254/latest/dynamic/instance-identity/document"
+)
+
+// See: https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/configuring-instance-metadata-service.html
+func getAWSCloudMetadata(ctx context.Context, client *http.Client, out *model.Cloud) error {
+	token, err := getAWSToken(ctx, client)
+	if err != nil {
+		return err
+	}
+
+	req, err := http.NewRequest("GET", ec2MetadataURL, nil)
+	if err != nil {
+		return err
+	}
+	if token != "" {
+		req.Header.Set("X-aws-ec2-metadata-token", token)
+	}
+
+	resp, err := client.Do(req.WithContext(ctx))
+	if err != nil {
+		return err
+	}
+	defer resp.Body.Close()
+	if resp.StatusCode != http.StatusOK {
+		return errors.New(resp.Status)
+	}
+
+	var ec2Metadata struct {
+		AccountID        string `json:"accountId"`
+		AvailabilityZone string `json:"availabilityZone"`
+		Region           string `json:"region"`
+		InstanceID       string `json:"instanceId"`
+		InstanceType     string `json:"instanceType"`
+	}
+	if err := json.NewDecoder(resp.Body).Decode(&ec2Metadata); err != nil {
+		return err
+	}
+
+	out.Region = ec2Metadata.Region
+	out.AvailabilityZone = ec2Metadata.AvailabilityZone
+	if ec2Metadata.InstanceID != "" {
+		out.Instance = &model.CloudInstance{ID: ec2Metadata.InstanceID}
+	}
+	if ec2Metadata.InstanceType != "" {
+		out.Machine = &model.CloudMachine{Type: ec2Metadata.InstanceType}
+	}
+	if ec2Metadata.AccountID != "" {
+		out.Account = &model.CloudAccount{ID: ec2Metadata.AccountID}
+	}
+	return nil
+}
+
+func getAWSToken(ctx context.Context, client *http.Client) (string, error) {
+	req, err := http.NewRequest("PUT", ec2TokenURL, nil)
+	if err != nil {
+		return "", err
+	}
+	req.Header.Set("X-aws-ec2-metadata-token-ttl-seconds", "300")
+	resp, err := client.Do(req.WithContext(ctx))
+	if err != nil {
+		return "", err
+	}
+	defer resp.Body.Close()
+	token, err := ioutil.ReadAll(resp.Body)
+	if err != nil {
+		return "", err
+	}
+	return string(token), nil
+}

internal/apmcloudutil/aws_test.go

@@ -0,0 +1,80 @@
+package apmcloudutil
+
+import (
+	"context"
+	"net/http"
+	"net/http/httptest"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+
+	"go.elastic.co/apm/model"
+)
+
+func TestAWSCloudMetadata(t *testing.T) {
+	srv, client := newAWSMetadataServer()
+	defer srv.Close()
+
+	for _, provider := range []Provider{Auto, AWS} {
+		var out model.Cloud
+		var logger testLogger
+		assert.True(t, provider.getCloudMetadata(context.Background(), client, &logger, &out))
+		assert.Zero(t, logger)
+		assert.Equal(t, model.Cloud{
+			Provider:         "aws",
+			Region:           "us-east-2",
+			AvailabilityZone: "us-east-2a",
+			Instance: &model.CloudInstance{
+				ID: "i-0ae894a7c1c4f2a75",
+			},
+			Machine: &model.CloudMachine{
+				Type: "t2.medium",
+			},
+			Account: &model.CloudAccount{
+				ID: "946960629917",
+			},
+		}, out)
+	}
+}
+
+func newAWSMetadataServer() (*httptest.Server, *http.Client) {
+	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		switch r.URL.Path {
+		case "/latest/api/token":
+			w.Write([]byte("topsecret"))
+			return
+		case "/latest/dynamic/instance-identity/document":
+			token := r.Header.Get("X-Aws-Ec2-Metadata-Token")
+			if token != "topsecret" {
+				w.WriteHeader(http.StatusBadRequest)
+				w.Write([]byte("invalid token"))
+				return
+			}
+			break
+		default:
+			w.WriteHeader(http.StatusNotFound)
+			return
+		}
+
+		w.Write([]byte(`{
+    "accountId": "946960629917",
+    "architecture": "x86_64",
+    "availabilityZone": "us-east-2a",
+    "billingProducts": null,
+    "devpayProductCodes": null,
+    "marketplaceProductCodes": null,
+    "imageId": "ami-07c1207a9d40bc3bd",
+    "instanceId": "i-0ae894a7c1c4f2a75",
+    "instanceType": "t2.medium",
+    "kernelId": null,
+    "pendingTime": "2020-06-12T17:46:09Z",
+    "privateIp": "172.31.0.212",
+    "ramdiskId": null,
+    "region": "us-east-2",
+    "version": "2017-09-30"
+}`))
+	}))
+
+	client := &http.Client{Transport: newTargetedRoundTripper("169.254.169.254", srv.Listener.Addr().String())}
+	return srv, client
+}

internal/apmcloudutil/azure.go

@@ -0,0 +1,61 @@
+package apmcloudutil
+
+import (
+	"context"
+	"encoding/json"
+	"errors"
+	"net/http"
+
+	"go.elastic.co/apm/model"
+)
+
+const (
+	azureMetadataURL = "http://169.254.169.254/metadata/instance/compute?api-version=2019-08-15"
+)
+
+// See: https://docs.microsoft.com/en-us/azure/virtual-machines/windows/instance-metadata-service
+func getAzureCloudMetadata(ctx context.Context, client *http.Client, out *model.Cloud) error {
+	req, err := http.NewRequest("GET", azureMetadataURL, nil)
+	if err != nil {
+		return err
+	}
+	req.Header.Set("Metadata", "true")
+
+	resp, err := client.Do(req.WithContext(ctx))
+	if err != nil {
+		return err
+	}
+	defer resp.Body.Close()
+	if resp.StatusCode != http.StatusOK {
+		return errors.New(resp.Status)
+	}
+
+	var azureMetadata struct {
+		Location          string `json:"location"`
+		Name              string `json:"name"`
+		ResourceGroupName string `json:"resourceGroupName"`
+		SubscriptionID    string `json:"subscriptionId"`
+		VMID              string `json:"vmId"`
+		VMSize            string `json:"vmSize"`
+		Zone              string `json:"zone"`
+	}
+	if err := json.NewDecoder(resp.Body).Decode(&azureMetadata); err != nil {
+		return err
+	}
+
+	out.Region = azureMetadata.Location
+	out.AvailabilityZone = azureMetadata.Zone
+	if azureMetadata.VMID != "" || azureMetadata.Name != "" {
+		out.Instance = &model.CloudInstance{ID: azureMetadata.VMID, Name: azureMetadata.Name}
+	}
+	if azureMetadata.VMSize != "" {
+		out.Machine = &model.CloudMachine{Type: azureMetadata.VMSize}
+	}
+	if azureMetadata.ResourceGroupName != "" {
+		out.Project = &model.CloudProject{Name: azureMetadata.ResourceGroupName}
+	}
+	if azureMetadata.SubscriptionID != "" {
+		out.Account = &model.CloudAccount{ID: azureMetadata.SubscriptionID}
+	}
+	return nil
+}

internal/apmcloudutil/azure_test.go

@@ -0,0 +1,63 @@
+package apmcloudutil
+
+import (
+	"context"
+	"net/http"
+	"net/http/httptest"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+
+	"go.elastic.co/apm/model"
+)
+
+func TestAzureCloudMetadata(t *testing.T) {
+	srv, client := newAzureMetadataServer()
+	defer srv.Close()
+
+	for _, provider := range []Provider{Auto, Azure} {
+		var out model.Cloud
+		var logger testLogger
+		assert.True(t, provider.getCloudMetadata(context.Background(), client, &logger, &out))
+		assert.Zero(t, logger)
+		assert.Equal(t, model.Cloud{
+			Provider: "azure",
+			Region:   "westus2",
+			Instance: &model.CloudInstance{
+				ID:   "e11ebedc-019d-427f-84dd-56cd4388d3a8",
+				Name: "basepi-test",
+			},
+			Machine: &model.CloudMachine{
+				Type: "Standard_D2s_v3",
+			},
+			Project: &model.CloudProject{
+				Name: "basepi-testing",
+			},
+			Account: &model.CloudAccount{
+				ID: "7657426d-c4c3-44ac-88a2-3b2cd59e6dba",
+			},
+		}, out)
+	}
+}
+
+func newAzureMetadataServer() (*httptest.Server, *http.Client) {
+	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		if r.URL.Path != "/metadata/instance/compute" {
+			w.WriteHeader(http.StatusNotFound)
+			return
+		}
+		w.Write([]byte(`{
+    "location": "westus2",
+    "name": "basepi-test",
+    "resourceGroupName": "basepi-testing",
+    "subscriptionId": "7657426d-c4c3-44ac-88a2-3b2cd59e6dba",
+    "vmId": "e11ebedc-019d-427f-84dd-56cd4388d3a8",
+    "vmScaleSetName": "",
+    "vmSize": "Standard_D2s_v3",
+    "zone": ""
+}`))
+	}))
+
+	client := &http.Client{Transport: newTargetedRoundTripper("169.254.169.254", srv.Listener.Addr().String())}
+	return srv, client
+}