Gathering detailed insights and metrics for @fastify/reply-from
Gathering detailed insights and metrics for @fastify/reply-from
Gathering detailed insights and metrics for @fastify/reply-from
Gathering detailed insights and metrics for @fastify/reply-from
fastify plugin to forward the current http request to another server
npm install @fastify/reply-from
Module System
Min. Node Version
Typescript Support
Node Version
NPM Version
152 Stars
418 Commits
77 Forks
20 Watching
4 Branches
79 Contributors
Updated on 28 Nov 2024
JavaScript (97.27%)
TypeScript (2.73%)
Cumulative downloads
Total Downloads
Last day
-19.3%
20,055
Compared to previous day
Last week
2.5%
126,979
Compared to previous week
Last month
-0%
537,922
Compared to previous month
Last year
243.9%
8,882,639
Compared to previous year
7
Fastify plugin to forward the current HTTP request to another server. HTTP2 to HTTP is supported too.
npm i @fastify/reply-from
@fastify/reply-from
and @fastify/multipart
should not be registered as sibling plugins nor should they be registered in plugins which have a parent-child relationship.<br>
The two plugins are incompatible, in the sense that the behavior of @fastify/reply-from
might not be the expected one when the above-mentioned conditions are not respected.<br>
This is due to the fact that @fastify/multipart
consumes the multipart content by parsing it, hence this content is not forwarded to the target service by @fastify/reply-from
.<br>
However, the two plugins may be used within the same fastify instance, at the condition that they belong to disjoint branches of the fastify plugins hierarchy tree.
The following example set up two Fastify servers and forward the request from one to the other:
1'use strict' 2 3const Fastify = require('fastify') 4 5const target = Fastify({ 6 logger: true 7}) 8 9target.get('/', (request, reply) => { 10 reply.send('hello world') 11}) 12 13const proxy = Fastify({ 14 logger: true 15}) 16 17proxy.register(require('@fastify/reply-from'), { 18 base: 'http://localhost:3001/' 19}) 20 21proxy.get('/', (request, reply) => { 22 reply.from('/') 23}) 24 25target.listen({ port: 3001 }, (err) => { 26 if (err) { 27 throw err 28 } 29 30 proxy.listen({ port: 3000 }, (err) => { 31 if (err) { 32 throw err 33 } 34 }) 35})
base
Set the base URL for all the forwarded requests. Will be required if http2
is set to true
Note that every path will be discarded.
Custom URL protocols unix+http:
and unix+https:
can be used to forward requests to a unix
socket server by using querystring.escape(socketPath)
as the hostname. This is not supported
for http2 nor undici. To illustrate:
1const socketPath = require('node:querystring').escape('/run/http-daemon.socket') 2proxy.register(require('@fastify/reply-from'), { 3 base: 'unix+http://${socketPath}/' 4});
undici
By default, undici will be used to perform the HTTP/1.1 requests. Enabling this flag should guarantee 20-50% more throughput.
This flag could controls the settings of the undici client, like so:
1proxy.register(require('@fastify/reply-from'), { 2 base: 'http://localhost:3001/', 3 // default settings 4 undici: { 5 connections: 128, 6 pipelining: 1, 7 keepAliveTimeout: 60 * 1000, 8 tls: { 9 rejectUnauthorized: false 10 } 11 } 12})
You can also include a proxy for the undici client:
1proxy.register(require('@fastify/reply-from'), { 2 base: 'http://localhost:3001/', 3 undici: { 4 proxy: 'http://my.proxy.server:8080', 5 } 6})
See undici own options for more configurations.
You can also pass the plugin a custom instance:
1proxy.register(require('@fastify/reply-from'), { 2 base: 'http://localhost:3001/', 3 undici: new undici.Pool('http://localhost:3001') 4})
http
Set the http
option to an Object to use
Node's http.request
will be used if you do not enable http2
. To customize the request
,
you can pass in agentOptions
and
requestOptions
. To illustrate:
1proxy.register(require('@fastify/reply-from'), { 2 base: 'http://localhost:3001/', 3 http: { 4 agentOptions: { // pass in any options from https://nodejs.org/api/http.html#http_new_agent_options 5 keepAliveMsecs: 10 * 60 * 1000 6 }, 7 requestOptions: { // pass in any options from https://nodejs.org/api/http.html#http_http_request_options_callback 8 timeout: 5000 // timeout in msecs, defaults to 10000 (10 seconds) 9 } 10 } 11})
You can also pass custom HTTP agents. If you pass the agents, then the http.agentOptions will be ignored. To illustrate:
1proxy.register(require('@fastify/reply-from'), { 2 base: 'http://localhost:3001/', 3 http: { 4 agents: { 5 'http:': new http.Agent({ keepAliveMsecs: 10 * 60 * 1000 }), // pass in any options from https://nodejs.org/api/http.html#http_new_agent_options 6 'https:': new https.Agent({ keepAliveMsecs: 10 * 60 * 1000 }) 7 8 }, 9 requestOptions: { // pass in any options from https://nodejs.org/api/http.html#http_http_request_options_callback 10 timeout: 5000 // timeout in msecs, defaults to 10000 (10 seconds) 11 } 12 } 13})
http2
You can either set http2
to true
or set the settings object to connect to a HTTP/2 server.
The http2
settings object has the shape of:
1proxy.register(require('@fastify/reply-from'), { 2 base: 'http://localhost:3001/', 3 http2: { 4 sessionTimeout: 10000, // HTTP/2 session timeout in msecs, defaults to 60000 (1 minute) 5 requestTimeout: 5000, // HTTP/2 request timeout in msecs, defaults to 10000 (10 seconds) 6 sessionOptions: { // HTTP/2 session connect options, pass in any options from https://nodejs.org/api/http2.html#http2_http2_connect_authority_options_listener 7 rejectUnauthorized: true 8 }, 9 requestOptions: { // HTTP/2 request options, pass in any options from https://nodejs.org/api/http2.html#clienthttp2sessionrequestheaders-options 10 endStream: true 11 } 12 } 13})
disableRequestLogging
By default package will issue log messages when a request is received. By setting this option to true, these log messages will be disabled.
Default for disableRequestLogging
will be false
. To disable the log messages set disableRequestLogging
to true
.
1proxy.register(require('@fastify/reply-from'), { 2 base: 'http://localhost:3001/', 3 disableRequestLogging: true // request log messages will be disabled 4})
cacheURLs
The number of parsed URLs that will be cached. Default: 100
.
disableCache
This option will disable the URL caching. This cache is dedicated to reduce the amount of URL object generation. Generating URLs is a main bottleneck of this module, please disable this cache with caution.
contentTypesToEncode
An array of content types whose response body will be passed through JSON.stringify()
.
This only applies when a custom body
is not passed in. Defaults to:
1[ 2 'application/json' 3]
retryMethods
On which methods should the connection be retried in case of socket hang up. Be aware that setting here not idempotent method may lead to unexpected results on target.
By default: ['GET', 'HEAD', 'OPTIONS', 'TRACE']
This plugin will always retry on 503 errors, unless retryMethods
does not contain GET
.
globalAgent
Enables the possibility to explictly opt-in for global agents.
Usage for undici global agent:
1import { setGlobalDispatcher, ProxyAgent } from 'undici'
2
3const proxyAgent = new ProxyAgent('my.proxy.server')
4setGlobalDispatcher(proxyAgent)
5
6fastify.register(FastifyReplyFrom, {
7 base: 'http://localhost:3001/',
8 globalAgent: true
9})
Usage for http/https global agent:
1fastify.register(FastifyReplyFrom, { 2 base: 'http://localhost:3001/', 3 // http and https is allowed to use http.globalAgent or https.globalAgent 4 globalAgent: true, 5 http: { 6 } 7})
destroyAgent
If set to true
, it will destroy all agents when the Fastify is closed.
If set to false
, it will not destroy the agents.
By Default: false
maxRetriesOn503
This plugin will always retry on GET
requests that returns 503 errors, unless retryMethods
does not contain GET
.
This option set the limit on how many times the plugin should retry the request, specifically for 503 errors.
By Default: 10
retryDelay
handler
. RequiredThis plugin gives the client an option to pass their own retry callback to allow the client to define what retryDelay they would like on any retries
outside the scope of what is handled by default in fastify-reply-from. To see the default please refer to index.js getDefaultDelay()
If a handler
is passed to the retryDelay
object the onus is on the client to invoke the default retry logic in their callback otherwise default cases such as 500 will not be handled
err
is the error thrown by making a request using whichever agent is configuredreq
is the raw request details sent to the underlying agent. Note: this object is not a Fastify request object, but instead the low-level request for the agent.res
is the raw response returned by the underlying agent (if available) Note: this object is not a Fastify response, but instead the low-level response from the agent. This property may be null if no response was obtained at all, like from a connection reset or timeout.attempt
in the object callback refers to the current retriesAttempt number. You are given the freedom to use this in concert with the retryCount property set to handle retriesgetDefaultRetry
refers to the default retry handler. If this callback returns not null and you wish to handle those case of errors simply invoke it as done below.retriesCount
refers to the retriesCount property a client passes to reply-from. Note if the client does not explicitly set this value it will default to 0. The objective value here is to avoid hard-coding and seeing the retriesCount set. It is your perogative to ensure that you ensure the value here is as you wish (and not 0
if not intended to be as a result of a lack of not setting it).Given example
1 const customRetryLogic = ({err, req, res, attempt, getDefaultRetry}) => { 2 //If this block is not included all non 500 errors will not be retried 3 const defaultDelay = getDefaultDelay(); 4 if (defaultDelay) return defaultDelay(); 5 6 //Custom retry logic 7 if (res && res.statusCode === 500 && req.method === 'GET') { 8 return 300 9 } 10 11 if (err && err.code == "UND_ERR_SOCKET"){ 12 return 600 13 } 14 15 return null 16 } 17 18....... 19 20fastify.register(FastifyReplyFrom, { 21 base: 'http://localhost:3001/', 22 retryDelay: customRetryLogic 23}) 24
Note the Typescript Equivalent
const customRetryLogic = ({req, res, err, getDefaultRetry}: RetryDetails) => {
...
}
...
reply.from(source, [opts])
The plugin decorates the
Reply
instance with a from
method, which will reply to the original request
from the desired source. The options allows to override any part of
the request or response being sent or received to/from the source.
Note: If base
is specified in plugin options, the source
here should not override the host/origin.
onResponse(request, reply, response)
Called when a HTTP response is received from the source. Passed the original source request
, the in-progress reply to the source as reply
, and the ongoing response
from the upstream server.
The default behavior is reply.send(response.stream)
, which will be disabled if the
option is specified.
When replying with a body of a different length it is necessary to remove
the content-length
header.
1{ 2 onResponse: (request, reply, res) => { 3 reply.removeHeader('content-length'); 4 reply.send('New body of different length'); 5 } 6}
Note: onResponse
is called after headers have already been sent. If you want to modify response headers, use the rewriteHeaders
hook.
onError(reply, error)
Called when a HTTP response is received with error from the source.
The default behavior is reply.send(error)
, which will be disabled if the
option is specified.
It must reply the error.
rewriteHeaders(headers, request)
Called to rewrite the headers of the response, before them being copied over to the outer response. Parameters are the original headers and the Fastify request. It must return the new headers object.
rewriteRequestHeaders(request, headers)
Called to rewrite the headers of the request, before them being sent to the other server. Parameters are the Fastify request and the original request headers. It must return the new headers object.
getUpstream(request, base)
Called to get upstream destination, before the request is being sent. Useful when you want to decide which target server to call based on the request data. Helpful for a gradual rollout of new services. Parameters are the Fastify request and the base string from the plugin options. It must return the upstream destination.
Only http1! As http2 uses one connection for the whole session only the base upstream is used. If you want to have different upstreams based on the request you can add multiple Fastify.register's with different ContraintStrategies.
e.g.:
Route grpc-web/http1 and grpc/http2 to different routes with a ContentType-ConstraintStrategy:
const contentTypeMatchContraintStrategy = {
// strategy name for referencing in the route handler `constraints` options
name: 'contentType',
// storage factory for storing routes in the find-my-way route tree
storage: function () {
let handlers = {}
return {
get: (type: any) => { return handlers[type] || null },
set: (type: any, store: any) => { handlers[type] = store }
}
},
// function to get the value of the constraint from each incoming request
deriveConstraint: (req: any, ctx: any) => {
return req.headers['content-type']
},
// optional flag marking if handlers without constraints can match requests that have a value for this constraint
mustMatchWhenDerived: true
}
server.addConstraintStrategy(contentTypeMatchContraintStrategy);
and then 2 different upstreams with different register's:
// grpc-web / http1
server.register(fastifyHttpProxy, {
// Although most browsers send with http2, nodejs cannot handle this http2 request
// therefore we have to transport to the grpc-web-proxy via http1
http2: false,
upstream: 'http://grpc-web-proxy',
constraints: { "contentType": "application/grpc-web+proto" }
});
// grpc / http2
server.register(fastifyHttpProxy, {
http2: true,
upstream: 'http://grpc.server',
constraints: { "contentType": "application/grpc+proto" }
});
queryString
or queryString(search, reqUrl, request)
Replaces the original querystring of the request with what is specified.
This will be passed to
querystring.stringify
.
object
: accepts an object that will be passed to querystring.stringify
function
: function that will return a string with the query parameters e.g. name=test&type=user
body
Replaces the original request body with what is specified. Unless
contentType
is specified, the content will be passed
through JSON.stringify()
.
Setting this option for GET, HEAD requests will throw an error "Rewriting the body when doing a {GET|HEAD} is not allowed".
Setting this option to null
will strip the body (and content-type
header) entirely from the proxied request.
method
Replaces the original request method with what is specified.
retriesCount
How many times it will try to pick another connection on socket hangup (ECONNRESET
error).
Useful when keeping the connection open (KeepAlive).
This number should be a function of the number of connections and the number of instances of a target.
By default: 0 (disabled)
contentType
Override the 'Content-Type'
header of the forwarded request, if we are
already overriding the body
.
formbody
expects the body to be returned as a string and not an object.
Use the contentTypesToEncode
option to pass in ['application/x-www-form-urlencoded']
This library has:
timeout
for http
set by default. The default value is 10 seconds (10000
).requestTimeout
& sessionTimeout
for http2
set by default.
requestTimeout
is 10 seconds (10000
), a value of 0 disables the timeout.sessionTimeout
is 60 seconds (60000
), a value of 0 disables the timeout.When a timeout happens, 504 Gateway Timeout
will be returned to the client.
req.id
unique
(see hyperid).MIT
The latest stable version of the package.
Stable Version
1
5.3/10
Summary
@fastify/reply-from JSON Content-Type parsing confusion
Affected Versions
< 9.6.0
Patched Versions
9.6.0
Reason
8 commit(s) and 4 issue activity found in the last 90 days -- score normalized to 10
Reason
no dangerous workflow patterns detected
Reason
no binaries found in the repo
Reason
0 existing vulnerabilities detected
Reason
license file detected
Details
Reason
security policy file detected
Details
Reason
SAST tool is not run on all commits -- score normalized to 6
Details
Reason
Found 11/22 approved changesets -- score normalized to 5
Reason
detected GitHub workflow tokens with excessive permissions
Details
Reason
no effort to earn an OpenSSF best practices badge detected
Reason
project is not fuzzed
Details
Score
Last Scanned on 2024-11-25
The Open Source Security Foundation is a cross-industry collaboration to improve the security of open source software (OSS). The Scorecard provides security health metrics for open source projects.
Learn More