zias/stats
1
0
Fork 0
Code Issues Pull Requests 1 Actions Packages Projects Releases Wiki Activity

try simplified event buffer with duration calculation on the fly instead

Browse Source
This commit is contained in:
Carl-Gerhard Lindesvärd
2026-03-14 11:39:19 +01:00
parent 9c3c1458bb
commit 7a76b968ba
9 changed files with 269 additions and 789 deletions
Download Patch File Download Diff File Expand all files Collapse all files

337
packages/db/src/buffers/event-buffer.test.ts
View File

@@ -2,42 +2,8 @@ import { getRedisCache } from '@openpanel/redis';
import { afterAll, beforeEach, describe, expect, it, vi } from 'vitest';
import { ch } from '../clickhouse/client';
// Mock transformEvent to avoid circular dependency with buffers -> services -> buffers
vi.mock('../services/event.service', () => ({
transformEvent: (event: any) => ({
id: event.id ?? 'id',
name: event.name,
deviceId: event.device_id,
profileId: event.profile_id,
projectId: event.project_id,
sessionId: event.session_id,
properties: event.properties ?? {},
createdAt: new Date(event.created_at ?? Date.now()),
country: event.country,
city: event.city,
region: event.region,
longitude: event.longitude,
latitude: event.latitude,
os: event.os,
osVersion: event.os_version,
browser: event.browser,
browserVersion: event.browser_version,
device: event.device,
brand: event.brand,
model: event.model,
duration: event.duration ?? 0,
path: event.path ?? '',
origin: event.origin ?? '',
referrer: event.referrer,
referrerName: event.referrer_name,
referrerType: event.referrer_type,
meta: event.meta,
importedAt: undefined,
sdkName: event.sdk_name,
sdkVersion: event.sdk_version,
profile: event.profile,
}),
}));
// Break circular dep: event-buffer -> event.service -> buffers/index -> EventBuffer
vi.mock('../services/event.service', () => ({}));
import { EventBuffer } from './event-buffer';
@@ -68,19 +34,16 @@ describe('EventBuffer', () => {
created_at: new Date().toISOString(),
} as any;
// Get initial count
const initialCount = await eventBuffer.getBufferSize();
// Add event and flush (events are micro-batched)
eventBuffer.add(event);
await eventBuffer.flush();
// Buffer counter should increase by 1
const newCount = await eventBuffer.getBufferSize();
expect(newCount).toBe(initialCount + 1);
});
it('adds multiple screen_views - moves previous to buffer with duration', async () => {
it('adds screen_view directly to buffer queue', async () => {
const t0 = Date.now();
const sessionId = 'session_1';
@@ -100,63 +63,23 @@ describe('EventBuffer', () => {
created_at: new Date(t0 + 1000).toISOString(),
} as any;
const view3 = {
project_id: 'p1',
profile_id: 'u1',
session_id: sessionId,
name: 'screen_view',
created_at: new Date(t0 + 3000).toISOString(),
} as any;
// Add first screen_view
const count1 = await eventBuffer.getBufferSize();
eventBuffer.add(view1);
await eventBuffer.flush();
// Should be stored as "last" but NOT in queue yet
// screen_view goes directly to buffer
const count2 = await eventBuffer.getBufferSize();
expect(count2).toBe(count1); // No change in buffer
expect(count2).toBe(count1 + 1);
// Last screen_view should be retrievable
const last1 = await eventBuffer.getLastScreenView({
projectId: 'p1',
sessionId: sessionId,
});
expect(last1).not.toBeNull();
expect(last1!.createdAt.toISOString()).toBe(view1.created_at);
// Add second screen_view
eventBuffer.add(view2);
await eventBuffer.flush();
// Now view1 should be in buffer
const count3 = await eventBuffer.getBufferSize();
expect(count3).toBe(count1 + 1);
// view2 should now be the "last"
const last2 = await eventBuffer.getLastScreenView({
projectId: 'p1',
sessionId: sessionId,
});
expect(last2!.createdAt.toISOString()).toBe(view2.created_at);
// Add third screen_view
eventBuffer.add(view3);
await eventBuffer.flush();
// Now view2 should also be in buffer
const count4 = await eventBuffer.getBufferSize();
expect(count4).toBe(count1 + 2);
// view3 should now be the "last"
const last3 = await eventBuffer.getLastScreenView({
projectId: 'p1',
sessionId: sessionId,
});
expect(last3!.createdAt.toISOString()).toBe(view3.created_at);
expect(count3).toBe(count1 + 2);
});
it('adds session_end - moves last screen_view and session_end to buffer', async () => {
it('adds session_end directly to buffer queue', async () => {
const t0 = Date.now();
const sessionId = 'session_2';
@@ -176,134 +99,36 @@ describe('EventBuffer', () => {
created_at: new Date(t0 + 5000).toISOString(),
} as any;
// Add screen_view
const count1 = await eventBuffer.getBufferSize();
eventBuffer.add(view);
await eventBuffer.flush();
// Should be stored as "last", not in buffer yet
const count2 = await eventBuffer.getBufferSize();
expect(count2).toBe(count1);
// Add session_end
eventBuffer.add(sessionEnd);
await eventBuffer.flush();
// Both should now be in buffer (+2)
const count3 = await eventBuffer.getBufferSize();
expect(count3).toBe(count1 + 2);
// Last screen_view should be cleared
const last = await eventBuffer.getLastScreenView({
projectId: 'p2',
sessionId: sessionId,
});
expect(last).toBeNull();
});
it('session_end with no previous screen_view - only adds session_end to buffer', async () => {
const sessionId = 'session_3';
const sessionEnd = {
project_id: 'p3',
profile_id: 'u3',
session_id: sessionId,
name: 'session_end',
created_at: new Date().toISOString(),
} as any;
const count1 = await eventBuffer.getBufferSize();
eventBuffer.add(sessionEnd);
await eventBuffer.flush();
// Only session_end should be in buffer (+1)
const count2 = await eventBuffer.getBufferSize();
expect(count2).toBe(count1 + 1);
});
it('gets last screen_view by profileId', async () => {
const view = {
project_id: 'p4',
profile_id: 'u4',
session_id: 'session_4',
name: 'screen_view',
path: '/home',
created_at: new Date().toISOString(),
} as any;
eventBuffer.add(view);
await eventBuffer.flush();
// Query by profileId
const result = await eventBuffer.getLastScreenView({
projectId: 'p4',
profileId: 'u4',
});
expect(result).not.toBeNull();
expect(result!.name).toBe('screen_view');
expect(result!.path).toBe('/home');
});
it('gets last screen_view by sessionId', async () => {
const sessionId = 'session_5';
const view = {
project_id: 'p5',
profile_id: 'u5',
session_id: sessionId,
name: 'screen_view',
path: '/about',
created_at: new Date().toISOString(),
} as any;
eventBuffer.add(view);
await eventBuffer.flush();
// Query by sessionId
const result = await eventBuffer.getLastScreenView({
projectId: 'p5',
sessionId: sessionId,
});
expect(result).not.toBeNull();
expect(result!.name).toBe('screen_view');
expect(result!.path).toBe('/about');
});
it('returns null for non-existent last screen_view', async () => {
const result = await eventBuffer.getLastScreenView({
projectId: 'p_nonexistent',
profileId: 'u_nonexistent',
});
expect(result).toBeNull();
expect(count2).toBe(count1 + 2);
});
it('gets buffer count correctly', async () => {
// Initially 0
expect(await eventBuffer.getBufferSize()).toBe(0);
// Add regular event
eventBuffer.add({
project_id: 'p6',
name: 'event1',
created_at: new Date().toISOString(),
} as any);
await eventBuffer.flush();
expect(await eventBuffer.getBufferSize()).toBe(1);
// Add another regular event
eventBuffer.add({
project_id: 'p6',
name: 'event2',
created_at: new Date().toISOString(),
} as any);
await eventBuffer.flush();
expect(await eventBuffer.getBufferSize()).toBe(2);
// Add screen_view (not counted until flushed)
// screen_view also goes directly to buffer
eventBuffer.add({
project_id: 'p6',
profile_id: 'u6',
@@ -312,21 +137,6 @@ describe('EventBuffer', () => {
created_at: new Date().toISOString(),
} as any);
await eventBuffer.flush();
// Still 2 (screen_view is pending)
expect(await eventBuffer.getBufferSize()).toBe(2);
// Add another screen_view (first one gets flushed)
eventBuffer.add({
project_id: 'p6',
profile_id: 'u6',
session_id: 'session_6',
name: 'screen_view',
created_at: new Date(Date.now() + 1000).toISOString(),
} as any);
await eventBuffer.flush();
// Now 3 (2 regular + 1 flushed screen_view)
expect(await eventBuffer.getBufferSize()).toBe(3);
});
@@ -355,14 +165,12 @@ describe('EventBuffer', () => {
await eventBuffer.processBuffer();
// Should insert both events
expect(insertSpy).toHaveBeenCalled();
const callArgs = insertSpy.mock.calls[0]![0];
expect(callArgs.format).toBe('JSONEachRow');
expect(callArgs.table).toBe('events');
expect(Array.isArray(callArgs.values)).toBe(true);
// Buffer should be empty after processing
expect(await eventBuffer.getBufferSize()).toBe(0);
insertSpy.mockRestore();
@@ -373,7 +181,6 @@ describe('EventBuffer', () => {
process.env.EVENT_BUFFER_CHUNK_SIZE = '2';
const eb = new EventBuffer();
// Add 4 events
for (let i = 0; i < 4; i++) {
eb.add({
project_id: 'p8',
@@ -389,14 +196,12 @@ describe('EventBuffer', () => {
await eb.processBuffer();
// With chunk size 2 and 4 events, should be called twice
expect(insertSpy).toHaveBeenCalledTimes(2);
const call1Values = insertSpy.mock.calls[0]![0].values as any[];
const call2Values = insertSpy.mock.calls[1]![0].values as any[];
expect(call1Values.length).toBe(2);
expect(call2Values.length).toBe(2);
// Restore
if (prev === undefined) delete process.env.EVENT_BUFFER_CHUNK_SIZE;
else process.env.EVENT_BUFFER_CHUNK_SIZE = prev;
@@ -418,126 +223,54 @@ describe('EventBuffer', () => {
expect(count).toBeGreaterThanOrEqual(1);
});
it('handles multiple sessions independently', async () => {
it('handles multiple sessions independently — all events go to buffer', async () => {
const t0 = Date.now();
const count1 = await eventBuffer.getBufferSize();
// Session 1
const view1a = {
eventBuffer.add({
project_id: 'p10',
profile_id: 'u10',
session_id: 'session_10a',
name: 'screen_view',
created_at: new Date(t0).toISOString(),
} as any;
const view1b = {
project_id: 'p10',
profile_id: 'u10',
session_id: 'session_10a',
name: 'screen_view',
created_at: new Date(t0 + 1000).toISOString(),
} as any;
// Session 2
const view2a = {
} as any);
eventBuffer.add({
project_id: 'p10',
profile_id: 'u11',
session_id: 'session_10b',
name: 'screen_view',
created_at: new Date(t0).toISOString(),
} as any;
const view2b = {
} as any);
eventBuffer.add({
project_id: 'p10',
profile_id: 'u10',
session_id: 'session_10a',
name: 'screen_view',
created_at: new Date(t0 + 1000).toISOString(),
} as any);
eventBuffer.add({
project_id: 'p10',
profile_id: 'u11',
session_id: 'session_10b',
name: 'screen_view',
created_at: new Date(t0 + 2000).toISOString(),
} as any;
eventBuffer.add(view1a);
eventBuffer.add(view2a);
eventBuffer.add(view1b); // Flushes view1a
eventBuffer.add(view2b); // Flushes view2a
} as any);
await eventBuffer.flush();
// Should have 2 events in buffer (one from each session)
expect(await eventBuffer.getBufferSize()).toBe(2);
// Each session should have its own "last" screen_view
const last1 = await eventBuffer.getLastScreenView({
projectId: 'p10',
sessionId: 'session_10a',
});
expect(last1!.createdAt.toISOString()).toBe(view1b.created_at);
const last2 = await eventBuffer.getLastScreenView({
projectId: 'p10',
sessionId: 'session_10b',
});
expect(last2!.createdAt.toISOString()).toBe(view2b.created_at);
// All 4 events are in buffer directly
expect(await eventBuffer.getBufferSize()).toBe(count1 + 4);
});
it('screen_view without session_id goes directly to buffer', async () => {
const view = {
it('bulk adds events to buffer', async () => {
const events = Array.from({ length: 5 }, (_, i) => ({
project_id: 'p11',
profile_id: 'u11',
name: 'screen_view',
created_at: new Date().toISOString(),
} as any;
name: `event${i}`,
created_at: new Date(Date.now() + i).toISOString(),
})) as any[];
const count1 = await eventBuffer.getBufferSize();
eventBuffer.add(view);
eventBuffer.bulkAdd(events);
await eventBuffer.flush();
// Should go directly to buffer (no session_id)
const count2 = await eventBuffer.getBufferSize();
expect(count2).toBe(count1 + 1);
});
it('updates last screen_view when new one arrives from same profile but different session', async () => {
const t0 = Date.now();
const view1 = {
project_id: 'p12',
profile_id: 'u12',
session_id: 'session_12a',
name: 'screen_view',
path: '/page1',
created_at: new Date(t0).toISOString(),
} as any;
const view2 = {
project_id: 'p12',
profile_id: 'u12',
session_id: 'session_12b', // Different session!
name: 'screen_view',
path: '/page2',
created_at: new Date(t0 + 1000).toISOString(),
} as any;
eventBuffer.add(view1);
eventBuffer.add(view2);
await eventBuffer.flush();
// Both sessions should have their own "last"
const lastSession1 = await eventBuffer.getLastScreenView({
projectId: 'p12',
sessionId: 'session_12a',
});
expect(lastSession1!.path).toBe('/page1');
const lastSession2 = await eventBuffer.getLastScreenView({
projectId: 'p12',
sessionId: 'session_12b',
});
expect(lastSession2!.path).toBe('/page2');
// Profile should have the latest one
const lastProfile = await eventBuffer.getLastScreenView({
projectId: 'p12',
profileId: 'u12',
});
expect(lastProfile!.path).toBe('/page2');
expect(await eventBuffer.getBufferSize()).toBe(5);
});
});

302
packages/db/src/buffers/event-buffer.ts
View File

@@ -5,32 +5,9 @@ import {
publishEvent,
} from '@openpanel/redis';
import { ch } from '../clickhouse/client';
import {
type IClickhouseEvent,
type IServiceEvent,
transformEvent,
} from '../services/event.service';
import { type IClickhouseEvent } from '../services/event.service';
import { BaseBuffer } from './base-buffer';
/**
* Event Buffer
*
* 1. All events go into a single list buffer (event_buffer:queue)
* 2. screen_view events are handled specially:
* - Store current screen_view as "last" for the session
* - When a new screen_view arrives, flush the previous one with calculated duration
* 3. session_end events:
* - Retrieve the last screen_view (don't modify it)
* - Push both screen_view and session_end to buffer
* 4. Flush: Process all events from the list buffer
*/
interface PendingEvent {
event: IClickhouseEvent;
eventJson: string;
eventWithTimestamp?: string;
type: 'regular' | 'screen_view' | 'session_end';
}
export class EventBuffer extends BaseBuffer {
private batchSize = process.env.EVENT_BUFFER_BATCH_SIZE
? Number.parseInt(process.env.EVENT_BUFFER_BATCH_SIZE, 10)
@@ -46,7 +23,7 @@ export class EventBuffer extends BaseBuffer {
? Number.parseInt(process.env.EVENT_BUFFER_MICRO_BATCH_SIZE, 10)
: 100;
private pendingEvents: PendingEvent[] = [];
private pendingEvents: IClickhouseEvent[] = [];
private flushTimer: ReturnType<typeof setTimeout> | null = null;
private isFlushing = false;
/** Tracks consecutive flush failures for observability; reset on success. */
@@ -59,100 +36,6 @@ export class EventBuffer extends BaseBuffer {
private queueKey = 'event_buffer:queue';
protected bufferCounterKey = 'event_buffer:total_count';
private scriptShas: {
addScreenView?: string;
addSessionEnd?: string;
} = {};
private getLastScreenViewKeyBySession(sessionId: string) {
return `event_buffer:last_screen_view:session:${sessionId}`;
}
private getLastScreenViewKeyByProfile(projectId: string, profileId: string) {
return `event_buffer:last_screen_view:profile:${projectId}:${profileId}`;
}
/**
* Lua script for screen_view addition.
* Uses GETDEL for atomic get-and-delete to prevent race conditions.
*
* KEYS[1] = last screen_view key (by session)
* KEYS[2] = last screen_view key (by profile, may be empty)
* KEYS[3] = queue key
* KEYS[4] = buffer counter key
* ARGV[1] = new event with timestamp as JSON: {"event": {...}, "ts": 123456}
* ARGV[2] = TTL for last screen_view (1 hour)
*/
private readonly addScreenViewScript = `
local sessionKey = KEYS[1]
local profileKey = KEYS[2]
local queueKey = KEYS[3]
local counterKey = KEYS[4]
local newEventData = ARGV[1]
local ttl = tonumber(ARGV[2])
local previousEventData = redis.call("GETDEL", sessionKey)
redis.call("SET", sessionKey, newEventData, "EX", ttl)
if profileKey and profileKey ~= "" then
redis.call("SET", profileKey, newEventData, "EX", ttl)
end
if previousEventData then
local prev = cjson.decode(previousEventData)
local curr = cjson.decode(newEventData)
if prev.ts and curr.ts then
prev.event.duration = math.max(0, curr.ts - prev.ts)
end
redis.call("RPUSH", queueKey, cjson.encode(prev.event))
redis.call("INCR", counterKey)
return 1
end
return 0
`;
/**
* Lua script for session_end.
* Uses GETDEL to atomically retrieve and delete the last screen_view.
*
* KEYS[1] = last screen_view key (by session)
* KEYS[2] = last screen_view key (by profile, may be empty)
* KEYS[3] = queue key
* KEYS[4] = buffer counter key
* ARGV[1] = session_end event JSON
*/
private readonly addSessionEndScript = `
local sessionKey = KEYS[1]
local profileKey = KEYS[2]
local queueKey = KEYS[3]
local counterKey = KEYS[4]
local sessionEndJson = ARGV[1]
local previousEventData = redis.call("GETDEL", sessionKey)
local added = 0
if previousEventData then
local prev = cjson.decode(previousEventData)
redis.call("RPUSH", queueKey, cjson.encode(prev.event))
redis.call("INCR", counterKey)
added = added + 1
end
redis.call("RPUSH", queueKey, sessionEndJson)
redis.call("INCR", counterKey)
added = added + 1
if profileKey and profileKey ~= "" then
redis.call("DEL", profileKey)
end
return added
`;
constructor() {
super({
name: 'event',
@@ -160,27 +43,6 @@ return added
await this.processBuffer();
},
});
this.loadScripts();
}
private async loadScripts() {
try {
const redis = getRedisCache();
const [screenViewSha, sessionEndSha] = await Promise.all([
redis.script('LOAD', this.addScreenViewScript),
redis.script('LOAD', this.addSessionEndScript),
]);
this.scriptShas.addScreenView = screenViewSha as string;
this.scriptShas.addSessionEnd = sessionEndSha as string;
this.logger.info('Loaded Lua scripts into Redis', {
addScreenView: this.scriptShas.addScreenView,
addSessionEnd: this.scriptShas.addSessionEnd,
});
} catch (error) {
this.logger.error('Failed to load Lua scripts', { error });
}
}
bulkAdd(events: IClickhouseEvent[]) {
@@ -190,30 +52,7 @@ return added
}
add(event: IClickhouseEvent) {
const eventJson = JSON.stringify(event);
let type: PendingEvent['type'] = 'regular';
let eventWithTimestamp: string | undefined;
if (event.session_id && event.name === 'screen_view') {
type = 'screen_view';
const timestamp = new Date(event.created_at || Date.now()).getTime();
eventWithTimestamp = JSON.stringify({
event: event,
ts: timestamp,
});
} else if (event.session_id && event.name === 'session_end') {
type = 'session_end';
}
const pendingEvent: PendingEvent = {
event,
eventJson,
eventWithTimestamp,
type,
};
this.pendingEvents.push(pendingEvent);
this.pendingEvents.push(event);
if (this.pendingEvents.length >= this.microBatchMaxSize) {
this.flushLocalBuffer();
@@ -228,57 +67,6 @@ return added
}
}
private addToMulti(multi: ReturnType<Redis['multi']>, pending: PendingEvent) {
const { event, eventJson, eventWithTimestamp, type } = pending;
if (type === 'screen_view' && event.session_id) {
const sessionKey = this.getLastScreenViewKeyBySession(event.session_id);
const profileKey = event.profile_id
? this.getLastScreenViewKeyByProfile(event.project_id, event.profile_id)
: '';
this.evalScript(
multi,
'addScreenView',
this.addScreenViewScript,
4,
sessionKey,
profileKey,
this.queueKey,
this.bufferCounterKey,
eventWithTimestamp!,
'3600',
);
} else if (type === 'session_end' && event.session_id) {
const sessionKey = this.getLastScreenViewKeyBySession(event.session_id);
const profileKey = event.profile_id
? this.getLastScreenViewKeyByProfile(event.project_id, event.profile_id)
: '';
this.evalScript(
multi,
'addSessionEnd',
this.addSessionEndScript,
4,
sessionKey,
profileKey,
this.queueKey,
this.bufferCounterKey,
eventJson,
);
} else {
multi.rpush(this.queueKey, eventJson).incr(this.bufferCounterKey);
}
if (event.profile_id) {
this.incrementActiveVisitorCount(
multi,
event.project_id,
event.profile_id,
);
}
}
public async flush() {
if (this.flushTimer) {
clearTimeout(this.flushTimer);
@@ -301,9 +89,17 @@ return added
const redis = getRedisCache();
const multi = redis.multi();
for (const pending of eventsToFlush) {
this.addToMulti(multi, pending);
for (const event of eventsToFlush) {
multi.rpush(this.queueKey, JSON.stringify(event));
if (event.profile_id) {
this.incrementActiveVisitorCount(
multi,
event.project_id,
event.profile_id,
);
}
}
multi.incrby(this.bufferCounterKey, eventsToFlush.length);
await multi.exec();
@@ -314,11 +110,14 @@ return added
this.pendingEvents = eventsToFlush.concat(this.pendingEvents);
this.flushRetryCount += 1;
this.logger.warn('Failed to flush local buffer to Redis; events re-queued', {
error,
eventCount: eventsToFlush.length,
flushRetryCount: this.flushRetryCount,
});
this.logger.warn(
'Failed to flush local buffer to Redis; events re-queued',
{
error,
eventCount: eventsToFlush.length,
flushRetryCount: this.flushRetryCount,
},
);
} finally {
this.isFlushing = false;
// Events may have accumulated while we were flushing; schedule another flush if needed
@@ -331,24 +130,6 @@ return added
}
}
private evalScript(
multi: ReturnType<Redis['multi']>,
scriptName: keyof typeof this.scriptShas,
scriptContent: string,
numKeys: number,
...args: (string | number)[]
) {
const sha = this.scriptShas[scriptName];
if (sha) {
multi.evalsha(sha, numKeys, ...args);
} else {
multi.eval(scriptContent, numKeys, ...args);
this.logger.warn(`Script ${scriptName} not loaded, using EVAL fallback`);
this.loadScripts();
}
}
async processBuffer() {
const redis = getRedisCache();
@@ -398,7 +179,10 @@ return added
const countByProject = new Map<string, number>();
for (const event of eventsToClickhouse) {
countByProject.set(event.project_id, (countByProject.get(event.project_id) ?? 0) + 1);
countByProject.set(
event.project_id,
(countByProject.get(event.project_id) ?? 0) + 1,
);
}
for (const [projectId, count] of countByProject) {
publishEvent('events', 'batch', { projectId, count });
@@ -419,42 +203,6 @@ return added
}
}
public async getLastScreenView(
params:
| {
sessionId: string;
}
| {
projectId: string;
profileId: string;
},
): Promise<IServiceEvent | null> {
const redis = getRedisCache();
let lastScreenViewKey: string;
if ('sessionId' in params) {
lastScreenViewKey = this.getLastScreenViewKeyBySession(params.sessionId);
} else {
lastScreenViewKey = this.getLastScreenViewKeyByProfile(
params.projectId,
params.profileId,
);
}
const eventDataStr = await redis.get(lastScreenViewKey);
if (eventDataStr) {
const eventData = getSafeJson<{ event: IClickhouseEvent; ts: number }>(
eventDataStr,
);
if (eventData?.event) {
return transformEvent(eventData.event);
}
}
return null;
}
public async getBufferSize() {
return this.getBufferSizeWithCounter(async () => {
const redis = getRedisCache();