ok

ingestion/tests/k6/cohort.js

@@ -0,0 +1,211 @@
+// k6 generator -- backfills events with timestamps spread over the last N days
+// so the analytics console can show a realistic cohort retention matrix.
+//
+// How it works
+//   1. setup() pre-computes a schedule: each of S shoppers gets a "first day"
+//      between [-N, 0], then revisits later days with a decay probability.
+//   2. The default function fires one scheduled event per iteration; we run
+//      `shared-iterations` so all VUs collaborate to drain the schedule.
+//
+// We deliberately omit `sentAt` from the payload. The ingest's 24h late-event
+// check is on sent_at (default = now when omitted), not on timestamp; so the
+// timestamp column lands in the past while the request itself looks live.
+//
+// Usage:
+//   k6 run tests/k6/cohort.js
+//   k6 run -e DAYS=14 -e SHOPPERS=80 -e VUS=10 tests/k6/cohort.js
+
+import http from 'k6/http';
+import { check } from 'k6';
+import encoding from 'k6/encoding';
+import { randomItem, randomIntBetween } from 'https://jslib.k6.io/k6-utils/1.4.0/index.js';
+
+const BASE      = __ENV.BASE      ?? 'http://localhost:3049';
+const WRITE_KEY = __ENV.WRITE_KEY ?? 'cdp_dev_writekey_1234567890';
+const SHOPPERS  = parseInt(__ENV.SHOPPERS ?? '50', 10);   // unique users
+const DAYS      = parseInt(__ENV.DAYS     ?? '7',  10);   // cohort window
+const VUS       = parseInt(__ENV.VUS      ?? '5',  10);
+
+const AUTH = 'Basic ' + encoding.b64encode(`${WRITE_KEY}:`);
+
+// ---------------------------------------------------------------------------
+// Fixtures
+// ---------------------------------------------------------------------------
+
+const PRODUCTS = [
+  { id: 'sku_alpha',  name: 'Alpha Hoodie',    category: 'apparel',    brand: 'CDP',     price: 49.0  },
+  { id: 'sku_beta',   name: 'Beta Mug',        category: 'drinkware',  brand: 'CDP',     price: 12.5  },
+  { id: 'sku_gamma',  name: 'Gamma Backpack',  category: 'bags',       brand: 'CDP',     price: 89.0  },
+  { id: 'sku_delta',  name: 'Delta Sneakers',  category: 'footwear',   brand: 'Athleta', price: 129.0 },
+  { id: 'sku_eps',    name: 'Epsilon Headset', category: 'electronics',brand: 'Sonix',   price: 199.0 },
+];
+
+const PLANS    = ['free', 'pro', 'team'];
+const COUNTRIES = ['US', 'VN', 'GB', 'SG', 'JP', 'DE', 'FR'];
+
+// ---------------------------------------------------------------------------
+// Pre-compute schedule (runs once before any VU starts)
+// ---------------------------------------------------------------------------
+
+function rand(min, max) { return Math.floor(Math.random() * (max - min + 1)) + min; }
+function pick(arr)      { return arr[Math.floor(Math.random() * arr.length)]; }
+
+function isoOnDay(daysAgo, hourSeed) {
+  // anchor at "today 12:00 UTC" minus N days, jitter a few hours so events
+  // spread across the day rather than clumping at the same minute.
+  const d = new Date();
+  d.setUTCHours(12, 0, 0, 0);
+  d.setUTCDate(d.getUTCDate() - daysAgo);
+  d.setUTCHours(d.getUTCHours() + (hourSeed % 9) - 4);
+  d.setUTCMinutes(d.getUTCMinutes() + rand(0, 59));
+  return d.toISOString();
+}
+
+export function setup() {
+  const events = [];
+  for (let i = 1; i <= SHOPPERS; i++) {
+    const userId   = `c_${String(i).padStart(3, '0')}`;
+    const traits   = {
+      email:   `cohort${i}@example.com`,
+      plan:    pick(PLANS),
+      country: pick(COUNTRIES),
+    };
+
+    // first day = day -N .. day -1 (no first-day-today so retention makes sense)
+    const firstDay = rand(1, DAYS - 1);
+
+    // Day 0 of this user's lifecycle = signup behavior.
+    push(events, userId, traits, firstDay, 'Product Viewed', null);
+    if (Math.random() < 0.7) push(events, userId, traits, firstDay, 'Product Added', null);
+    if (Math.random() < 0.4) push(events, userId, traits, firstDay, 'Order Completed', null);
+
+    // Later days: decaying return probability.
+    for (let d = firstDay - 1; d >= 0; d--) {
+      const since   = firstDay - d;             // days since first
+      const pReturn = Math.max(0.05, 0.85 - 0.15 * since); // 85% → 5%
+      if (Math.random() < pReturn) {
+        push(events, userId, traits, d, 'Product Viewed', null);
+        if (Math.random() < 0.3) push(events, userId, traits, d, 'Product Added', null);
+        if (Math.random() < 0.15) push(events, userId, traits, d, 'Order Completed', null);
+      }
+    }
+  }
+  // Shuffle so partition keys spread evenly.
+  for (let i = events.length - 1; i > 0; i--) {
+    const j = Math.floor(Math.random() * (i + 1));
+    [events[i], events[j]] = [events[j], events[i]];
+  }
+  return { events };
+}
+
+function push(events, userId, traits, daysAgo, eventName, _) {
+  events.push({
+    userId,
+    traits,
+    daysAgo,
+    eventName,
+    seed: events.length,
+  });
+}
+
+// ---------------------------------------------------------------------------
+// Scenario
+// ---------------------------------------------------------------------------
+
+export const options = {
+  scenarios: {
+    backfill: {
+      executor: 'shared-iterations',
+      vus: VUS,
+      // Set this big enough; setup() decides the real cap. We just stop once
+      // we run out of events (returns early in default()).
+      iterations: 20_000,
+      maxDuration: '5m',
+    },
+  },
+  thresholds: {
+    http_req_failed:   ['rate<0.02'],
+    checks:            ['rate>0.98'],
+  },
+};
+
+export default function (data) {
+  const ev = data.events[__ITER];
+  if (!ev) return; // schedule exhausted
+
+  const ts        = isoOnDay(ev.daysAgo, ev.seed);
+  const product   = randomItem(PRODUCTS);
+  const messageId = `cohort_${ev.userId}_${ev.seed}_${Date.now()}`;
+
+  let properties;
+  switch (ev.eventName) {
+    case 'Product Viewed':
+      properties = productProps(product);
+      break;
+    case 'Product Added':
+      properties = { ...productProps(product), quantity: randomIntBetween(1, 3) };
+      break;
+    case 'Order Completed':
+      properties = orderProps();
+      break;
+  }
+
+  const payload = JSON.stringify({
+    type: 'track',
+    messageId,
+    userId: ev.userId,
+    anonymousId: `anon_${ev.userId}`,
+    event: ev.eventName,
+    properties,
+    traits: ev.traits,
+    // event time in the past...
+    timestamp: ts,
+    // ...request time intentionally omitted so the ingest's late-event guard
+    // (which checks sent_at, not timestamp) does not reject us.
+    context: {
+      library_name: 'k6-cohort-sim',
+      library_version: '0.1.0',
+      ip: '127.0.0.1',
+      userAgent: 'k6/cohort',
+      locale: 'en-US',
+    },
+  });
+
+  const res = http.post(`${BASE}/v1/track`, payload, {
+    headers: { 'Content-Type': 'application/json', Authorization: AUTH },
+    tags:    { event: ev.eventName },
+  });
+
+  check(res, {
+    'status 200': (r) => r.status === 200,
+    'body ok':    (r) => r.json('ok') === true,
+  });
+}
+
+function productProps(p) {
+  return {
+    product_id: p.id, sku: p.id, name: p.name, category: p.category,
+    brand: p.brand, price: p.price, currency: 'USD',
+  };
+}
+
+function orderProps() {
+  const lines = [];
+  const n = randomIntBetween(1, 3);
+  let total = 0;
+  for (let i = 0; i < n; i++) {
+    const p = randomItem(PRODUCTS);
+    const qty = randomIntBetween(1, 2);
+    total += p.price * qty;
+    lines.push({ product_id: p.id, sku: p.id, name: p.name, price: p.price, quantity: qty });
+  }
+  return {
+    order_id: `ord_${Date.now()}_${randomIntBetween(1000, 9999)}`,
+    revenue:  Number(total.toFixed(2)),
+    currency: 'USD',
+    tax:      Number((total * 0.08).toFixed(2)),
+    shipping: 5,
+    total:    Number((total + total * 0.08 + 5).toFixed(2)),
+    products: lines,
+  };
+}