Add analytics to asset-worker

.changeset/eleven-rats-press.md

@@ -0,0 +1,5 @@
+---
+"@cloudflare/workers-shared": minor
+---
+
+Adds analytics to asset-worker

packages/workers-shared/asset-worker/src/analytics.ts

@@ -0,0 +1,77 @@
+import type { ReadyAnalytics } from "./types";
+
+// This will allow us to make breaking changes to the analytic schema
+const VERSION = 1;
+
+// When adding new columns please update the schema
+type Data = {
+	// -- Doubles --
+	// double1 - The time it takes for the whole request to complete in milliseconds
+	requestTime?: number;
+	// double2 - Colo ID
+	coloId?: number;
+	// double3 - Metal ID
+	metalId?: number;
+	// double4 - Colo tier (e.g. tier 1, tier 2, tier 3)
+	coloTier?: number;
+
+	// -- Blobs --
+	// blob1 - Hostname of the request
+	hostname?: string;
+	// blob2 - User agent making the request
+	userAgent?: string;
+	// blob3 - Html handling option
+	htmlHandling?: string;
+	// blob4 - Not found handling option
+	notFoundHandling?: string;
+	// blob5 - Error message
+	error?: string;
+	// blob6 - The current version UUID of asset-worker
+	version?: string;
+	// blob7 - Region of the colo (e.g. WEUR)
+	coloRegion?: string;
+};
+
+export class Analytics {
+	private data: Data = {};
+	private readyAnalytics?: ReadyAnalytics;
+
+	constructor(readyAnalytics?: ReadyAnalytics) {
+		this.readyAnalytics = readyAnalytics;
+	}
+
+	setData(newData: Partial<Data>) {
+		this.data = { ...this.data, ...newData };
+	}
+
+	getData(key: keyof Data) {
+		return this.data[key];
+	}
+
+	write(hostname?: string) {
+		if (!this.readyAnalytics) {
+			return;
+		}
+
+		this.readyAnalytics.logEvent({
+			version: VERSION,
+			accountId: 0, // TODO: need to plumb through
+			indexId: hostname,
+			doubles: [
+				this.data.requestTime ?? -1, // double1
+				this.data.coloId ?? -1, // double2
+				this.data.metalId ?? -1, // double3
+				this.data.coloTier ?? -1, // double4
+			],
+			blobs: [
+				this.data.hostname?.substring(0, 256), // blob1 - trim to 256 bytes
+				this.data.userAgent?.substring(0, 256), // blob2 - trim to 256 bytes
+				this.data.htmlHandling, // blob3
+				this.data.notFoundHandling, // blob4
+				this.data.error?.substring(0, 256), // blob5 - trim to 256 bytes
+				this.data.version, // blob6
+				this.data.coloRegion, // blob7
+			],
+		});
+	}
+}

packages/workers-shared/asset-worker/src/index.ts

@@ -1,5 +1,7 @@
 import { WorkerEntrypoint } from "cloudflare:workers";
+import { PerformanceTimer } from "../../utils/performance";
 import { setupSentry } from "../../utils/sentry";
+import { Analytics } from "./analytics";
 import { AssetsManifest } from "./assets-manifest";
 import { applyConfigurationDefaults } from "./configuration";
 import { decodePath, getIntent, handleRequest } from "./handler";
@@ -8,7 +10,8 @@ import {
 	MethodNotAllowedResponse,
 } from "./responses";
 import { getAssetWithMetadataFromKV } from "./utils/kv";
-import type { AssetConfig } from "../../utils/types";
+import type { AssetConfig, UnsafePerformanceTimer } from "../../utils/types";
+import type { ColoMetadata, Environment, ReadyAnalytics } from "./types";
 
 type Env = {
 	/*
@@ -29,6 +32,12 @@ type Env = {
 
 	SENTRY_ACCESS_CLIENT_ID: string;
 	SENTRY_ACCESS_CLIENT_SECRET: string;
+
+	ENVIRONMENT: Environment;
+	ANALYTICS: ReadyAnalytics;
+	COLO_METADATA: ColoMetadata;
+	UNSAFE_PERFORMANCE: UnsafePerformanceTimer;
+	VERSION_METADATA: WorkerVersionMetadata;
 };
 
 /*
@@ -45,6 +54,10 @@ type Env = {
 export default class extends WorkerEntrypoint<Env> {
 	async fetch(request: Request): Promise<Response> {
 		let sentry: ReturnType<typeof setupSentry> | undefined;
+		const analytics = new Analytics(this.env.ANALYTICS);
+		const performance = new PerformanceTimer(this.env.UNSAFE_PERFORMANCE);
+		const startTimeMs = performance.now();
+
 		try {
 			sentry = setupSentry(
 				request,
@@ -54,9 +67,34 @@ export default class extends WorkerEntrypoint<Env> {
 				this.env.SENTRY_ACCESS_CLIENT_SECRET
 			);
 
+			const config = applyConfigurationDefaults(this.env.CONFIG);
+			const userAgent = request.headers.get("user-agent") ?? "UA UNKNOWN";
+
+			if (sentry) {
+				const colo = this.env.COLO_METADATA.coloId;
+				sentry.setTag("colo", this.env.COLO_METADATA.coloId);
+				sentry.setTag("metal", this.env.COLO_METADATA.metalId);
+				sentry.setUser({ userAgent: userAgent, colo: colo });
+			}
+
+			if (this.env.COLO_METADATA && this.env.VERSION_METADATA) {
+				const url = new URL(request.url);
+				analytics.setData({
+					coloId: this.env.COLO_METADATA.coloId,
+					metalId: this.env.COLO_METADATA.metalId,
+					coloTier: this.env.COLO_METADATA.coloTier,
+					coloRegion: this.env.COLO_METADATA.coloRegion,
+					version: this.env.VERSION_METADATA.id,
+					hostname: url.hostname,
+					htmlHandling: config.html_handling,
+					notFoundHandling: config.not_found_handling,
+					userAgent: userAgent,
+				});
+			}
+
 			return handleRequest(
 				request,
-				applyConfigurationDefaults(this.env.CONFIG),
+				config,
 				this.unstable_exists.bind(this),
 				this.unstable_getByETag.bind(this)
 			);
@@ -68,7 +106,14 @@ export default class extends WorkerEntrypoint<Env> {
 				sentry.captureException(err);
 			}
 
+			if (err instanceof Error) {
+				analytics.setData({ error: err.message });
+			}
+
 			return response;
+		} finally {
+			analytics.setData({ requestTime: performance.now() - startTimeMs });
+			analytics.write();
 		}
 	}
 

packages/workers-shared/asset-worker/src/types.ts

@@ -0,0 +1,20 @@
+export type Environment = "production" | "staging";
+
+export interface ReadyAnalytics {
+	logEvent: (e: ReadyAnalyticsEvent) => void;
+}
+
+export interface ColoMetadata {
+	metalId: number;
+	coloId: number;
+	coloRegion: string;
+	coloTier: number;
+}
+
+export interface ReadyAnalyticsEvent {
+	accountId?: number;
+	indexId?: string;
+	version?: number;
+	doubles?: (number | undefined)[];
+	blobs?: (string | undefined)[];
+}

packages/workers-shared/asset-worker/wrangler.toml

@@ -31,4 +31,11 @@ type = "internal_assets"
 
 [unsafe.metadata.build_options]
 stable_id = "cloudflare/cf_asset_worker"
-networks = ["cf","jdc"]
+networks = ["cf","jdc"]
+
+[version_metadata]
+binding = "VERSION_METADATA"
+
+[[unsafe.bindings]]
+name = "workers-asset-worker"
+type = "internal_capability_grants"

packages/workers-shared/router-worker/src/index.ts

@@ -1,14 +1,9 @@
+import { PerformanceTimer } from "../../utils/performance";
 import { setupSentry } from "../../utils/sentry";
 import { Analytics, DISPATCH_TYPE } from "./analytics";
-import { PerformanceTimer } from "./performance";
 import type AssetWorker from "../../asset-worker/src/index";
-import type { RoutingConfig } from "../../utils/types";
-import type {
-	ColoMetadata,
-	Environment,
-	ReadyAnalytics,
-	UnsafePerformanceTimer,
-} from "./types";
+import type { RoutingConfig, UnsafePerformanceTimer } from "../../utils/types";
+import type { ColoMetadata, Environment, ReadyAnalytics } from "./types";
 
 interface Env {
 	ASSET_WORKER: Service<AssetWorker>;

packages/workers-shared/router-worker/src/types.ts

@@ -11,11 +11,6 @@ export interface ColoMetadata {
 	coloTier: number;
 }
 
-export interface UnsafePerformanceTimer {
-	readonly timeOrigin: number;
-	now: () => number;
-}
-
 export interface ReadyAnalyticsEvent {
 	accountId?: number;
 	indexId?: string;

packages/workers-shared/utils/types.ts

@@ -20,3 +20,8 @@ export const AssetConfigSchema = z.object({
 
 export type RoutingConfig = z.infer<typeof RoutingConfigSchema>;
 export type AssetConfig = z.infer<typeof AssetConfigSchema>;
+
+export interface UnsafePerformanceTimer {
+	readonly timeOrigin: number;
+	now: () => number;
+}