Rate Limiter Module for NestJS
Documentation Map
- Description
- Installation
- Basic Usage
-
Options
- for
- type
- keyPrefix
- points
- pointsConsumed
- inmemoryBlockOnConsumed
- duration
- blockDuration
- inmemoryBlockDuration
- queueEnabled
- whiteList
- blackList
- storeClient
- insuranceLimiter
- storeType
- dbName
- tableName
- tableCreated
- clearExpiredByTimeout
- execEvenly
- execEvenlyMinDelayMs
- indexKeyPrefix
- maxQueueSize
- omitResponseHeaders
- errorMessage
- customResponseSchema
- Override Functions
- Benchmarks
- TODO List
Description
nestjs-rate-limiter
is a module which adds in configurable rate limiting for Nest applications.
Under the hood it uses rate-limiter-flexible.
Installation
npm i --save nestjs-rate-limiter
Or if you use Yarn:
yarn add nestjs-rate-limiter
Requirements
nestjs-rate-limiter
is built to work with Nest 6 and newer versions.
Basic Usage
Include Module
First you need to import this module into your main application module:
app.module.ts
import { RateLimiterModule } from 'nestjs-rate-limiter'
@Module({
imports: [RateLimiterModule],
})
export class ApplicationModule {}
Using Guard
Now you need to register the guard. You can do this only on some routes:
app.controller.ts
import { RateLimiterGuard } from 'nestjs-rate-limiter'
@UseGuards(RateLimiterGuard)
@Get('/login')
public async login() {
console.log('hello')
}
Or you can choose to register the guard globally:
app.module.ts
import { APP_GUARD } from '@nestjs/core'
import { RateLimiterModule, RateLimiterGuard } from 'nestjs-rate-limiter'
@Module({
imports: [RateLimiterModule],
providers: [
{
provide: APP_GUARD,
useClass: RateLimiterGuard,
},
],
})
export class ApplicationModule {}
With Decorator
You can use the @RateLimit
decorator to specify the points and duration for rate limiting on a per controller or per
route basis:
app.controller.ts
import { RateLimit } from 'nestjs-rate-limiter'
@RateLimit({ keyPrefix: 'sign-up', points: 1, duration: 60, errorMessage: 'Accounts cannot be created more than once in per minute' })
@Get('/signup')
public async signUp() {
console.log('hello')
}
Dynamic Keyprefix
import { RateLimit } from 'nestjs-rate-limiter'
@RateLimit({
keyPrefix: () => programmaticFuncThatReturnsValue(),
points: 1,
duration: 60,
customResponseSchema: () => { return { timestamp: '1611479696', message: 'Request has been blocked' }}
})
@Get('/example')
public async example() {
console.log('hello')
}
With All Options
The usage of the limiter options is as in the code block below. For an explanation of the each option, please see options
.
@Module({
imports: [
// All the values here are defaults.
RateLimiterModule.register({
for: 'Express',
type: 'Memory',
keyPrefix: 'global',
points: 4,
pointsConsumed: 1,
inmemoryBlockOnConsumed: 0,
duration: 1,
blockDuration: 0,
inmemoryBlockDuration: 0,
queueEnabled: false,
whiteList: [],
blackList: [],
storeClient: undefined,
insuranceLimiter: undefined,
storeType: undefined,
dbName: undefined,
tableName: undefined,
tableCreated: undefined,
clearExpiredByTimeout: undefined,
execEvenly: false,
execEvenlyMinDelayMs: undefined,
indexKeyPrefix: {},
maxQueueSize: 100,
omitResponseHeaders: false,
errorMessage: 'Rate limit exceeded',
logger: true,
customResponseSchema: undefined
}),
],
providers: [
{
provide: APP_GUARD,
useClass: RateLimiterGuard,
},
],
})
export class ApplicationModule {}
Fastify based Graphql
If you want to use this library on a fastify based graphql server, you need to override the graphql context in the app.module as shown below.
GraphQLModule.forRoot({
context: ({ request, reply }) => {
return { req: request, res: reply }
},
}),
Options
● for
Default: 'Express'
Type: 'Express' | 'Fastify' | 'Microservice' | 'ExpressGraphql' | 'FastifyGraphql'
In this option, you specify what the technology is running under the Nest application. The wrong value causes to limiter not working.
● type
Default: 'Memory'
Type: 'Memory' | 'Redis' | 'Memcache' | 'Postgres' | 'MySQL' | 'Mongo'
Here you define where the limiter data will be stored. Each option plays a different role in limiter performance, to see that please check benchmarks.
● keyPrefix
Default: 'global'
Type: string
For creating several limiters with different options to apply different modules/endpoints.
Set to empty string '', if keys should be stored without prefix.
Note: for some limiters it should correspond to Storage requirements for tables or collections name, as keyPrefix may be used as their name.
● points
Default: 4
Type: number
Maximum number of points can be consumed over duration.
● pointsConsumed
Default: 1
Type: number
You can consume more than 1 point per invocation of the rate limiter.
For instance if you have a limit of 100 points per 60 seconds, and pointsConsumed is set to 10, the user will effectively be able to make 10 requests per 60 seconds.
● inmemoryBlockOnConsumed
Default: 0
Type: number
For Redis, Memcached, MongoDB, MySQL, PostgreSQL, etc.
Can be used against DDoS attacks. In-memory blocking works in current process memory and for consume method only.
It blocks a key in memory for msBeforeNext milliseconds from the last consume result, if inmemoryBlockDuration is not set. This helps to avoid extra requests. It is not necessary to increment counter on store, if all points are consumed already.
● duration
Default: 1
Type: number
Number of seconds before consumed points are reset.
Keys never expire, if duration is 0.
● blockDuration
Default: 0
Type: number
If positive number and consumed more than points in current duration, block for blockDuration seconds.
● inmemoryBlockDuration
Default: 0
Type: number
For Redis, Memcached, MongoDB, MySQL, PostgreSQL, etc.
Block key for inmemoryBlockDuration seconds, if inmemoryBlockOnConsumed or more points are consumed. Set it the same as blockDuration option for distributed application to have consistent result on all processes.
● queueEnabled
Default: false
Type: boolean
It activates the queue mechanism, and holds the incoming requests for duration
value.
● whiteList
Default: []
Type: string[]
If the IP is white listed, consume resolved no matter how many points consumed.
● blackList
Default: []
Type: string[]
If the IP is black listed, consume rejected anytime. Blacklisted IPs are blocked on code level not in store/memory. Think of it as of requests filter.
● storeClient
Default: undefined
Type: any
Required for Redis, Memcached, MongoDB, MySQL, PostgreSQL, etc.
Have to be redis, ioredis, memcached, mongodb, pg, mysql2, mysql or any other related pool or connection.
● insuranceLimiter
Default: undefined
Type: any
Default: undefined For Redis, Memcached, MongoDB, MySQL, PostgreSQL.
Instance of RateLimiterAbstract extended object to store limits, when database comes up with any error.
All data from insuranceLimiter is NOT copied to parent limiter, when error gone
Note: insuranceLimiter automatically setup blockDuration and execEvenly to same values as in parent to avoid unexpected behaviour.
● storeType
Default: storeClient.constructor.name
Type: any
For MySQL and PostgreSQL It is required only for Knex and have to be set to 'knex'
● dbName
Default for MySQL, Postgres & Mongo: 'rate-limiter'
Type: string
Database where limits are stored. It is created during creating a limiter. Doesn't work with Mongoose, as mongoose connection is established to exact database.
● tableName
Default: equals to 'keyPrefix' option
Type: string
For MongoDB, MySQL, PostgreSQL.
By default, limiter creates a table for each unique keyPrefix. tableName option sets table/collection name where values should be store.
● tableCreated
Default: false
Type: boolean
Does not create a table for rate limiter, if tableCreated is true
.
● clearExpiredByTimeout
Default for MySQL and PostgreSQL: true
Type: boolean
Rate limiter deletes data expired more than 1 hour ago every 5 minutes.
● execEvenly
Default: false
Type: boolean
Delay action to be executed evenly over duration First action in duration is executed without delay. All next allowed actions in current duration are delayed by formula msBeforeDurationEnd / (remainingPoints + 2) with minimum delay of duration * 1000 / points It allows to cut off load peaks similar way to Leaky Bucket.
Note: it isn't recommended to use it for long duration and few points, as it may delay action for too long with default execEvenlyMinDelayMs.
● execEvenlyMinDelayMs
Default: duration * 1000 / points
Type: number
Sets minimum delay in milliseconds, when action is delayed with execEvenly
● indexKeyPrefix
Default: {}
Type: {}
Object which is used to create combined index by {...indexKeyPrefix, key: 1} attributes.
● maxQueueSize
Default: 100
Type: number
Determines the maximum number of requests in the queue and returns 429
as response to requests that over of the maxQueueSize.
● omitResponseHeaders
Default: false
Type: boolean
Whether or not the rate limit headers (X-Retry-After
, X-RateLimit-Limit
, X-Retry-Remaining
, X-Retry-Reset
) should be omitted in the response.
● errorMessage
Default: 'Rate limit exceeded'
Type: string
errorMessage option can change the error message of rate limiter exception.
● logger
Default: true
Type: boolean
logger option allows to enable or disable logging from library.
● customResponseSchema
Default: undefined
Type: string
customResponseSchema option allows to provide customizable response schemas
Override Functions
getIpFromRequest
function by extending RateLimiterGuard
class.
It's possible to override import { RateLimiterGuard } from 'nestjs-rate-limiter'
import type { Request } from 'express'
class ExampleRateLimiterGuard extends RateLimiterGuard {
protected getIpFromRequest(request: Request): string {
return request.get('x-forwarded-for');
}
}
Benchmarks
1000 concurrent clients with maximum 2000 requests per sec during 30 seconds.
1. Memory 0.34 ms
2. Redis 2.45 ms
3. Memcached 3.89 ms
4. Mongo 4.75 ms
500 concurrent clients with maximum 1000 req per sec during 30 seconds
5. PostgreSQL 7.48 ms (with connection pool max 100)
6. MySQL 14.59 ms (with connection pool 100)
TODO
- [ ] Support Websocket
- [ ] Support Rpc