convex-rate-limiter（凸速率限制器）

Convex Rate Limiter

@convex-dev/rate-limiter — Type-safe, transactional, application-level rate limiting for Convex.

Installation

npm install @convex-dev/rate-limiter

Register the component in convex/convex.config.ts:

import { defineApp } from "convex/server";
import rateLimiter from "@convex-dev/rate-limiter/convex.config.js";

const app = defineApp();
app.use(rateLimiter);
export default app;

Setup — Define Named Rate Limits

import { RateLimiter, MINUTE, HOUR, SECOND } from "@convex-dev/rate-limiter";
import { components } from "./_generated/api";

const rateLimiter = new RateLimiter(components.rateLimiter, {
  // Fixed window — hard quota that resets each period
  freeTrialSignUp: { kind: "fixed window", rate: 100, period: HOUR },

  // Token bucket — smooth traffic with burst allowance
  sendMessage: { kind: "token bucket", rate: 10, period: MINUTE, capacity: 3 },

  // Failed login throttle
  failedLogins: { kind: "token bucket", rate: 10, period: HOUR },
});

• period is in milliseconds (SECOND = 1000, MINUTE = 60000, HOUR = 3600000).
• Multiple RateLimiter instances allowed; config keys must not overlap.

Strategy Selection

Token bucket config:

{ kind: "token bucket", rate: 10, period: MINUTE, capacity: 3 }
// capacity = max burst tokens (optional, defaults to rate)

Fixed window config:

{ kind: "fixed window", rate: 100, period: HOUR }
// start: optional custom epoch; random by default to prevent thundering herd

Core API

limit() — Consume tokens

const { ok, retryAfter } = await rateLimiter.limit(ctx, "sendMessage", {
  key: userId,       // Per-entity key (omit for global limit)
  count: 1,          // Tokens to consume (default 1)
  throws: false,     // Set true to auto-throw ConvexError
});
if (!ok) throw new Error(`Rate limited. Retry in ${Math.ceil(retryAfter / 1000)}s`);

check() — Query without consuming (safe in queries)

const { ok, retryAfter } = await rateLimiter.check(ctx, "sendMessage", {
  key: userId,
});

reset() — Clear a rate limit

await rateLimiter.reset(ctx, "failedLogins", { key: email });

throws: true — Auto-throw pattern

import { isRateLimitError } from "@convex-dev/rate-limiter";

await rateLimiter.limit(ctx, "sendMessage", { key: userId, throws: true });
// Throws ConvexError with data: { kind, name, retryAfter }

Catch with isRateLimitError(error) to inspect .data.retryAfter.

Usage Patterns

Global rate limit (no key)

export const signUp = mutation({
  args: { email: v.string() },
  handler: async (ctx, args) => {
    await rateLimiter.limit(ctx, "freeTrialSignUp", { throws: true });
    await ctx.db.insert("users", { email: args.email });
  },
});

Per-user rate limit

export const sendMessage = mutation({
  args: { text: v.string() },
  handler: async (ctx, args) => {
    const user = await ctx.auth.getUserIdentity();
    await rateLimiter.limit(ctx, "sendMessage", {
      key: user?.subject,
      throws: true,
    });
    await ctx.db.insert("messages", { userId: user?.subject, text: args.text });
  },
});

Failed login throttle with reset on success

export const login = mutation({
  args: { email: v.string(), password: v.string() },
  handler: async (ctx, args) => {
    await rateLimiter.limit(ctx, "failedLogins", {
      key: args.email,
      throws: true,
    });
    const success = await verifyCredentials(ctx, args);
    if (success) {
      await rateLimiter.reset(ctx, "failedLogins", { key: args.email });
    }
    return success;
  },
});

Consume multiple tokens at once

await rateLimiter.limit(ctx, "llmTokens", {
  key: userId,
  count: estimateTokens(prompt),
  throws: true,
});

Critical Rules

1. Always pass key for per-entity limits. Omitting key makes it a global singleton.
2. Always surface retryAfter to the client — don't just say "rate limited".
3. Rate limit changes are transactional — they roll back if the mutation fails.
4. Use throws: true for cleaner code; catch with isRateLimitError().
5. Reset wisely — reset on success (e.g., login), never on every request.

Common Pitfalls

// BAD: Global limit when you want per-user
await rateLimiter.limit(ctx, "sendMessage");

// GOOD: Per-user limit
await rateLimiter.limit(ctx, "sendMessage", { key: userId });

// BAD: Ignoring retryAfter
const { ok } = await rateLimiter.limit(ctx, "action");
if (!ok) throw new Error("Rate limited");

// GOOD: Tell user when to retry
const { ok, retryAfter } = await rateLimiter.limit(ctx, "action");
if (!ok) throw new Error(`Wait ${Math.ceil(retryAfter / 1000)} seconds`);

Best Practices Summary

Advanced Topics

• Sharding for high throughput (OCC conflicts): See references/advanced.md
• Capacity reservation (prevent starvation): See references/advanced.md
• Direct value access & calculateRateLimit: See references/advanced.md
• Jitter patterns: See references/advanced.md
• Troubleshooting: See references/advanced.md

Common Patterns

• Tiered rate limits (free vs premium): See references/patterns.md
• Anonymous/IP-based limiting: See references/patterns.md
• Multiple rate limits in one transaction: See references/patterns.md
• React client integration (useRateLimit): See references/patterns.md
• Inline/dynamic config: See references/patterns.md
• Testing: See references/patterns.md