Capture second metrics sample to provide metrics delta for investigat…

…ing issues

CHANGELOG.md

@@ -18,6 +18,10 @@ Main (unreleased)
 
 - Add `otelcol.exporter.splunkhec` allowing to export otel data to Splunk HEC (@adlotsof)
 
+### Enhancements
+
+- Add second metrics sample to the support bundle to provide delta information (@dehaansa)
+
 ### Bugfixes
 
 - Fixed an issue in the `prometheus.exporter.postgres` component that would leak goroutines when the target was not reachable (@dehaansa)

docs/sources/troubleshoot/support_bundle.md

@@ -38,7 +38,8 @@ A support bundle contains the following data:
 `/api/v0/web/components` endpoint.
 * `alloy-logs.txt` contains the logs during the bundle generation.
 * `alloy-metadata.yaml` contains the {{< param "PRODUCT_NAME" >}} build version and the installation's operating system, architecture, and uptime.
-* `alloy-metrics.txt` contains a snapshot of the internal metrics for {{< param "PRODUCT_NAME" >}}.
+* `alloy-metrics-sample-1.txt` contains a snapshot of the internal metrics for {{< param "PRODUCT_NAME" >}} at the start of the bundle collection.
+* `alloy-metrics-sample-2.txt` contains a snapshot of the internal metrics for {{< param "PRODUCT_NAME" >}} at the end of the bundle collection.
 * `alloy-peers.json` contains information about the identified cluster peers of this {{< param "PRODUCT_NAME" >}} instance, generated by the 
 `/api/v0/web/peers` endpoint.
 * `alloy-runtime-flags.txt` contains the values of the runtime flags available in {{< param "PRODUCT_NAME" >}}.

internal/service/http/supportbundle.go

@@ -28,16 +28,17 @@ type SupportBundleContext struct {
 
 // Bundle collects all the data that is exposed as a support bundle.
 type Bundle struct {
-	meta         []byte
-	alloyMetrics []byte
-	components   []byte
-	peers        []byte
-	runtimeFlags []byte
-	heapBuf      *bytes.Buffer
-	goroutineBuf *bytes.Buffer
-	blockBuf     *bytes.Buffer
-	mutexBuf     *bytes.Buffer
-	cpuBuf       *bytes.Buffer
+	meta              []byte
+	alloyMetricsStart []byte
+	alloyMetricsEnd   []byte
+	components        []byte
+	peers             []byte
+	runtimeFlags      []byte
+	heapBuf           *bytes.Buffer
+	goroutineBuf      *bytes.Buffer
+	blockBuf          *bytes.Buffer
+	mutexBuf          *bytes.Buffer
+	cpuBuf            *bytes.Buffer
 }
 
 // Metadata contains general runtime information about the current Alloy environment.
@@ -50,6 +51,26 @@ type Metadata struct {
 
 // ExportSupportBundle gathers the information required for the support bundle.
 func ExportSupportBundle(ctx context.Context, runtimeFlags []string, srvAddress string, dialContext server.DialContextFunc) (*Bundle, error) {
+	var httpClient http.Client
+	httpClient.Transport = &http.Transport{DialContext: dialContext}
+
+	// Gather Alloy's own metrics.
+	alloyMetricsStart, err := retrieveAPIEndpoint(httpClient, srvAddress, "metrics")
+	if err != nil {
+		return nil, fmt.Errorf("failed to get internal Alloy metrics: %s", err)
+	}
+
+	// Gather running component configuration
+	components, err := retrieveAPIEndpoint(httpClient, srvAddress, "api/v0/web/components")
+	if err != nil {
+		return nil, fmt.Errorf("failed to get component details: %s", err)
+	}
+	// Gather cluster peers information
+	peers, err := retrieveAPIEndpoint(httpClient, srvAddress, "api/v0/web/peers")
+	if err != nil {
+		return nil, fmt.Errorf("failed to get peer details: %s", err)
+	}
+
 	// The block profiler is disabled by default. Temporarily enable recording
 	// of all blocking events. Also, temporarily record all mutex contentions,
 	// and defer restoring of earlier mutex profiling fraction.
@@ -76,24 +97,6 @@ func ExportSupportBundle(ctx context.Context, runtimeFlags []string, srvAddress
 		return nil, fmt.Errorf("failed to marshal support bundle metadata: %s", err)
 	}
 
-	var httpClient http.Client
-	httpClient.Transport = &http.Transport{DialContext: dialContext}
-	// Gather Alloy's own metrics.
-	alloyMetrics, err := retrieveAPIEndpoint(httpClient, srvAddress, "metrics")
-	if err != nil {
-		return nil, fmt.Errorf("failed to get internal Alloy metrics: %s", err)
-	}
-	// Gather running component configuration
-	components, err := retrieveAPIEndpoint(httpClient, srvAddress, "api/v0/web/components")
-	if err != nil {
-		return nil, fmt.Errorf("failed to get component details: %s", err)
-	}
-	// Gather cluster peers information
-	peers, err := retrieveAPIEndpoint(httpClient, srvAddress, "api/v0/web/peers")
-	if err != nil {
-		return nil, fmt.Errorf("failed to get peer details: %s", err)
-	}
-
 	// Export pprof data.
 	var (
 		cpuBuf       bytes.Buffer
@@ -129,19 +132,26 @@ func ExportSupportBundle(ctx context.Context, runtimeFlags []string, srvAddress
 		return nil, err
 	}
 
+	// Gather Alloy's own metrics after the profile completes
+	alloyMetricsEnd, err := retrieveAPIEndpoint(httpClient, srvAddress, "metrics")
+	if err != nil {
+		return nil, fmt.Errorf("failed to get internal Alloy metrics: %s", err)
+	}
+
 	// Finally, bundle everything up to be served, either as a zip from
 	// memory, or exported to a directory.
 	bundle := &Bundle{
-		meta:         meta,
-		alloyMetrics: alloyMetrics,
-		components:   components,
-		peers:        peers,
-		runtimeFlags: []byte(strings.Join(runtimeFlags, "\n")),
-		heapBuf:      &heapBuf,
-		goroutineBuf: &goroutineBuf,
-		blockBuf:     &blockBuf,
-		mutexBuf:     &mutexBuf,
-		cpuBuf:       &cpuBuf,
+		meta:              meta,
+		alloyMetricsStart: alloyMetricsStart,
+		alloyMetricsEnd:   alloyMetricsEnd,
+		components:        components,
+		peers:             peers,
+		runtimeFlags:      []byte(strings.Join(runtimeFlags, "\n")),
+		heapBuf:           &heapBuf,
+		goroutineBuf:      &goroutineBuf,
+		blockBuf:          &blockBuf,
+		mutexBuf:          &mutexBuf,
+		cpuBuf:            &cpuBuf,
 	}
 
 	return bundle, nil
@@ -169,17 +179,18 @@ func ServeSupportBundle(rw http.ResponseWriter, b *Bundle, logsBuf *bytes.Buffer
 	rw.Header().Set("Content-Disposition", "attachment; filename=\"alloy-support-bundle.zip\"")
 
 	zipStructure := map[string][]byte{
-		"alloy-metadata.yaml":     b.meta,
-		"alloy-components.json":   b.components,
-		"alloy-peers.json":        b.peers,
-		"alloy-metrics.txt":       b.alloyMetrics,
-		"alloy-runtime-flags.txt": b.runtimeFlags,
-		"alloy-logs.txt":          logsBuf.Bytes(),
-		"pprof/cpu.pprof":         b.cpuBuf.Bytes(),
-		"pprof/heap.pprof":        b.heapBuf.Bytes(),
-		"pprof/goroutine.pprof":   b.goroutineBuf.Bytes(),
-		"pprof/mutex.pprof":       b.mutexBuf.Bytes(),
-		"pprof/block.pprof":       b.blockBuf.Bytes(),
+		"alloy-metadata.yaml":        b.meta,
+		"alloy-components.json":      b.components,
+		"alloy-peers.json":           b.peers,
+		"alloy-metrics-sample-1.txt": b.alloyMetricsStart,
+		"alloy-metrics-sample-2.txt": b.alloyMetricsEnd,
+		"alloy-runtime-flags.txt":    b.runtimeFlags,
+		"alloy-logs.txt":             logsBuf.Bytes(),
+		"pprof/cpu.pprof":            b.cpuBuf.Bytes(),
+		"pprof/heap.pprof":           b.heapBuf.Bytes(),
+		"pprof/goroutine.pprof":      b.goroutineBuf.Bytes(),
+		"pprof/mutex.pprof":          b.mutexBuf.Bytes(),
+		"pprof/block.pprof":          b.blockBuf.Bytes(),
 	}
 
 	for fn, b := range zipStructure {