This version is still in development and is not considered stable yet. For the latest stable version, please use Spring Framework 6.2.0! |
Reactive Core
The spring-web
module contains the following foundational support for reactive web
applications:
-
For server request processing there are two levels of support.
-
HttpHandler: Basic contract for HTTP request handling with non-blocking I/O and Reactive Streams back pressure, along with adapters for Reactor Netty, Undertow, Tomcat, Jetty, and any Servlet container.
-
WebHandler
API: Slightly higher level, general-purpose web API for request handling, on top of which concrete programming models such as annotated controllers and functional endpoints are built.
-
-
For the client side, there is a basic
ClientHttpConnector
contract to perform HTTP requests with non-blocking I/O and Reactive Streams back pressure, along with adapters for Reactor Netty, reactive Jetty HttpClient and Apache HttpComponents. The higher level WebClient used in applications builds on this basic contract. -
For client and server, codecs for serialization and deserialization of HTTP request and response content.
HttpHandler
HttpHandler is a simple contract with a single method to handle a request and a response. It is intentionally minimal, and its main and only purpose is to be a minimal abstraction over different HTTP server APIs.
The following table describes the supported server APIs:
Server name | Server API used | Reactive Streams support |
---|---|---|
Netty |
Netty API |
|
Undertow |
Undertow API |
spring-web: Undertow to Reactive Streams bridge |
Tomcat |
Servlet non-blocking I/O; Tomcat API to read and write ByteBuffers vs byte[] |
spring-web: Servlet non-blocking I/O to Reactive Streams bridge |
Jetty |
Servlet non-blocking I/O; Jetty API to write ByteBuffers vs byte[] |
spring-web: Servlet non-blocking I/O to Reactive Streams bridge |
Servlet container |
Servlet non-blocking I/O |
spring-web: Servlet non-blocking I/O to Reactive Streams bridge |
The following table describes server dependencies (also see supported versions):
Server name | Group id | Artifact name |
---|---|---|
Reactor Netty |
io.projectreactor.netty |
reactor-netty |
Undertow |
io.undertow |
undertow-core |
Tomcat |
org.apache.tomcat.embed |
tomcat-embed-core |
Jetty |
org.eclipse.jetty |
jetty-server, jetty-servlet |
The code snippets below show using the HttpHandler
adapters with each server API:
Reactor Netty
-
Java
-
Kotlin
HttpHandler handler = ...
ReactorHttpHandlerAdapter adapter = new ReactorHttpHandlerAdapter(handler);
HttpServer.create().host(host).port(port).handle(adapter).bindNow();
val handler: HttpHandler = ...
val adapter = ReactorHttpHandlerAdapter(handler)
HttpServer.create().host(host).port(port).handle(adapter).bindNow()
Undertow
-
Java
-
Kotlin
HttpHandler handler = ...
UndertowHttpHandlerAdapter adapter = new UndertowHttpHandlerAdapter(handler);
Undertow server = Undertow.builder().addHttpListener(port, host).setHandler(adapter).build();
server.start();
val handler: HttpHandler = ...
val adapter = UndertowHttpHandlerAdapter(handler)
val server = Undertow.builder().addHttpListener(port, host).setHandler(adapter).build()
server.start()
Tomcat
-
Java
-
Kotlin
HttpHandler handler = ...
Servlet servlet = new TomcatHttpHandlerAdapter(handler);
Tomcat server = new Tomcat();
File base = new File(System.getProperty("java.io.tmpdir"));
Context rootContext = server.addContext("", base.getAbsolutePath());
Tomcat.addServlet(rootContext, "main", servlet);
rootContext.addServletMappingDecoded("/", "main");
server.setHost(host);
server.setPort(port);
server.start();
val handler: HttpHandler = ...
val servlet = TomcatHttpHandlerAdapter(handler)
val server = Tomcat()
val base = File(System.getProperty("java.io.tmpdir"))
val rootContext = server.addContext("", base.absolutePath)
Tomcat.addServlet(rootContext, "main", servlet)
rootContext.addServletMappingDecoded("/", "main")
server.host = host
server.setPort(port)
server.start()
Jetty
-
Java
-
Kotlin
HttpHandler handler = ...
Servlet servlet = new JettyHttpHandlerAdapter(handler);
Server server = new Server();
ServletContextHandler contextHandler = new ServletContextHandler(server, "");
contextHandler.addServlet(new ServletHolder(servlet), "/");
contextHandler.start();
ServerConnector connector = new ServerConnector(server);
connector.setHost(host);
connector.setPort(port);
server.addConnector(connector);
server.start();
val handler: HttpHandler = ...
val servlet = JettyHttpHandlerAdapter(handler)
val server = Server()
val contextHandler = ServletContextHandler(server, "")
contextHandler.addServlet(ServletHolder(servlet), "/")
contextHandler.start();
val connector = ServerConnector(server)
connector.host = host
connector.port = port
server.addConnector(connector)
server.start()
Servlet Container
To deploy as a WAR to any Servlet container, you can extend and include
AbstractReactiveWebInitializer
in the WAR. That class wraps an HttpHandler
with ServletHttpHandlerAdapter
and registers
that as a Servlet
.
WebHandler
API
The org.springframework.web.server
package builds on the HttpHandler
contract
to provide a general-purpose web API for processing requests through a chain of multiple
WebExceptionHandler
, multiple
WebFilter
, and a single
WebHandler
component. The chain can
be put together with WebHttpHandlerBuilder
by simply pointing to a Spring
ApplicationContext
where components are
auto-detected, and/or by registering components
with the builder.
While HttpHandler
has a simple goal to abstract the use of different HTTP servers, the
WebHandler
API aims to provide a broader set of features commonly used in web applications
such as:
-
User session with attributes.
-
Request attributes.
-
Resolved
Locale
orPrincipal
for the request. -
Access to parsed and cached form data.
-
Abstractions for multipart data.
-
and more..
Special bean types
The table below lists the components that WebHttpHandlerBuilder
can auto-detect in a
Spring ApplicationContext, or that can be registered directly with it:
Bean name | Bean type | Count | Description |
---|---|---|---|
<any> |
|
0..N |
Provide handling for exceptions from the chain of |
<any> |
|
0..N |
Apply interception style logic to before and after the rest of the filter chain and
the target |
|
|
1 |
The handler for the request. |
|
|
0..1 |
The manager for |
|
|
0..1 |
For access to |
|
|
0..1 |
The resolver for |
|
|
0..1 |
For processing forwarded type headers, either by extracting and removing them or by removing them only. Not used by default. |
Form Data
ServerWebExchange
exposes the following method for accessing form data:
-
Java
-
Kotlin
Mono<MultiValueMap<String, String>> getFormData();
suspend fun getFormData(): MultiValueMap<String, String>
The DefaultServerWebExchange
uses the configured HttpMessageReader
to parse form data
(application/x-www-form-urlencoded
) into a MultiValueMap
. By default,
FormHttpMessageReader
is configured for use by the ServerCodecConfigurer
bean
(see the Web Handler API).
Multipart Data
ServerWebExchange
exposes the following method for accessing multipart data:
-
Java
-
Kotlin
Mono<MultiValueMap<String, Part>> getMultipartData();
suspend fun getMultipartData(): MultiValueMap<String, Part>
The DefaultServerWebExchange
uses the configured
HttpMessageReader<MultiValueMap<String, Part>>
to parse multipart/form-data
,
multipart/mixed
, and multipart/related
content into a MultiValueMap
.
By default, this is the DefaultPartHttpMessageReader
, which does not have any third-party
dependencies.
Alternatively, the SynchronossPartHttpMessageReader
can be used, which is based on the
Synchronoss NIO Multipart library.
Both are configured through the ServerCodecConfigurer
bean
(see the Web Handler API).
To parse multipart data in streaming fashion, you can use the Flux<PartEvent>
returned from the
PartEventHttpMessageReader
instead of using @RequestPart
, as that implies Map
-like access
to individual parts by name and, hence, requires parsing multipart data in full.
By contrast, you can use @RequestBody
to decode the content to Flux<PartEvent>
without
collecting to a MultiValueMap
.
Forwarded Headers
As a request goes through proxies such as load balancers the host, port, and scheme may change, and that makes it a challenge to create links that point to the correct host, port, and scheme from a client perspective.
RFC 7239 defines the Forwarded
HTTP header
that proxies can use to provide information about the original request.
Non-standard Headers
There are other non-standard headers, too, including X-Forwarded-Host
, X-Forwarded-Port
,
X-Forwarded-Proto
, X-Forwarded-Ssl
, and X-Forwarded-Prefix
.
X-Forwarded-Host
While not standard, X-Forwarded-Host: <host>
is a de-facto standard header that is used to communicate the original host to a
downstream server. For example, if a request of example.com/resource
is sent to
a proxy which forwards the request to localhost:8080/resource
, then a header of
X-Forwarded-Host: example.com
can be sent to inform the server that the original host was example.com
.
X-Forwarded-Port
While not standard, X-Forwarded-Port: <port>
is a de-facto standard header that is used to
communicate the original port to a downstream server. For example, if a request of
example.com/resource
is sent to a proxy which forwards the request to
localhost:8080/resource
, then a header of X-Forwarded-Port: 443
can be sent
to inform the server that the original port was 443
.
X-Forwarded-Proto
While not standard, X-Forwarded-Proto: (https|http)
is a de-facto standard header that is used to communicate the original protocol (e.g. https / https)
to a downstream server. For example, if a request of example.com/resource
is sent to
a proxy which forwards the request to localhost:8080/resource
, then a header of
X-Forwarded-Proto: https
can be sent to inform the server that the original protocol was https
.
X-Forwarded-Ssl
While not standard, X-Forwarded-Ssl: (on|off)
is a de-facto standard header that is used to communicate the
original protocol (e.g. https / https) to a downstream server. For example, if a request of
example.com/resource
is sent to a proxy which forwards the request to
localhost:8080/resource
, then a header of X-Forwarded-Ssl: on
to inform the server that the
original protocol was https
.
X-Forwarded-Prefix
While not standard, X-Forwarded-Prefix: <prefix>
is a de-facto standard header that is used to communicate the original URL path prefix to a
downstream server.
Use of X-Forwarded-Prefix
can vary by deployment scenario, and needs to be flexible to
allow replacing, removing, or prepending the path prefix of the target server.
Scenario 1: Override path prefix
https://example.com/api/{path} -> http://localhost:8080/app1/{path}
The prefix is the start of the path before the capture group {path}
. For the proxy,
the prefix is /api
while for the server the prefix is /app1
. In this case, the proxy
can send X-Forwarded-Prefix: /api
to have the original prefix /api
override the
server prefix /app1
.
Scenario 2: Remove path prefix
At times, an application may want to have the prefix removed. For example, consider the following proxy to server mapping:
https://app1.example.com/{path} -> http://localhost:8080/app1/{path} https://app2.example.com/{path} -> http://localhost:8080/app2/{path}
The proxy has no prefix, while applications app1
and app2
have path prefixes
/app1
and /app2
respectively. The proxy can send X-Forwarded-Prefix:
to
have the empty prefix override server prefixes /app1
and /app2
.
A common case for this deployment scenario is where licenses are paid per production application server, and it is preferable to deploy multiple applications per server to reduce fees. Another reason is to run more applications on the same server in order to share the resources required by the server to run. In these scenarios, applications need a non-empty context root because there are multiple applications on the same server. However, this should not be visible in URL paths of the public API where applications may use different subdomains that provides benefits such as:
|
Scenario 3: Insert path prefix
In other cases, it may be necessary to prepend a prefix. For example, consider the following proxy to server mapping:
https://example.com/api/app1/{path} -> http://localhost:8080/app1/{path}
In this case, the proxy has a prefix of /api/app1
and the server has a prefix of
/app1
. The proxy can send X-Forwarded-Prefix: /api/app1
to have the original prefix
/api/app1
override the server prefix /app1
.
ForwardedHeaderTransformer
ForwardedHeaderTransformer
is a component that modifies the host, port, and scheme of
the request, based on forwarded headers, and then removes those headers. If you declare
it as a bean with the name forwardedHeaderTransformer
, it will be
detected and used.
In 5.1 ForwardedHeaderFilter was deprecated and superseded by
ForwardedHeaderTransformer so forwarded headers can be processed earlier, before the
exchange is created. If the filter is configured anyway, it is taken out of the list of
filters, and ForwardedHeaderTransformer is used instead.
|
Security Considerations
There are security considerations for forwarded headers since an application cannot know
if the headers were added by a proxy, as intended, or by a malicious client. This is why
a proxy at the boundary of trust should be configured to remove untrusted forwarded traffic coming
from the outside. You can also configure the ForwardedHeaderTransformer
with
removeOnly=true
, in which case it removes but does not use the headers.
Filters
In the WebHandler
API, you can use a WebFilter
to apply interception-style
logic before and after the rest of the processing chain of filters and the target
WebHandler
. When using the WebFlux Config, registering a WebFilter
is as simple
as declaring it as a Spring bean and (optionally) expressing precedence by using @Order
on
the bean declaration or by implementing Ordered
.
CORS
Spring WebFlux provides fine-grained support for CORS configuration through annotations on
controllers. However, when you use it with Spring Security, we advise relying on the built-in
CorsFilter
, which must be ordered ahead of Spring Security’s chain of filters.
See the section on CORS and the CORS WebFilter
for more details.
Exceptions
In the WebHandler
API, you can use a WebExceptionHandler
to handle
exceptions from the chain of WebFilter
instances and the target WebHandler
. When using the
WebFlux Config, registering a WebExceptionHandler
is as simple as declaring it as a
Spring bean and (optionally) expressing precedence by using @Order
on the bean declaration or
by implementing Ordered
.
The following table describes the available WebExceptionHandler
implementations:
Exception Handler | Description |
---|---|
|
Provides handling for exceptions of type
|
|
Extension of This handler is declared in the WebFlux Config. |
Codecs
The spring-web
and spring-core
modules provide support for serializing and
deserializing byte content to and from higher level objects through non-blocking I/O with
Reactive Streams back pressure. The following describes this support:
-
Encoder
andDecoder
are low level contracts to encode and decode content independent of HTTP. -
HttpMessageReader
andHttpMessageWriter
are contracts to encode and decode HTTP message content. -
An
Encoder
can be wrapped withEncoderHttpMessageWriter
to adapt it for use in a web application, while aDecoder
can be wrapped withDecoderHttpMessageReader
. -
DataBuffer
abstracts different byte buffer representations (e.g. NettyByteBuf
,java.nio.ByteBuffer
, etc.) and is what all codecs work on. See Data Buffers and Codecs in the "Spring Core" section for more on this topic.
The spring-core
module provides byte[]
, ByteBuffer
, DataBuffer
, Resource
, and
String
encoder and decoder implementations. The spring-web
module provides Jackson
JSON, Jackson Smile, JAXB2, Protocol Buffers and other encoders and decoders along with
web-only HTTP message reader and writer implementations for form data, multipart content,
server-sent events, and others.
ClientCodecConfigurer
and ServerCodecConfigurer
are typically used to configure and
customize the codecs to use in an application. See the section on configuring
HTTP message codecs.
Jackson JSON
JSON and binary JSON (Smile) are both supported when the Jackson library is present.
The Jackson2Decoder
works as follows:
-
Jackson’s asynchronous, non-blocking parser is used to aggregate a stream of byte chunks into
TokenBuffer
's each representing a JSON object. -
Each
TokenBuffer
is passed to Jackson’sObjectMapper
to create a higher level object. -
When decoding to a single-value publisher (e.g.
Mono
), there is oneTokenBuffer
. -
When decoding to a multi-value publisher (e.g.
Flux
), eachTokenBuffer
is passed to theObjectMapper
as soon as enough bytes are received for a fully formed object. The input content can be a JSON array, or any line-delimited JSON format such as NDJSON, JSON Lines, or JSON Text Sequences.
The Jackson2Encoder
works as follows:
-
For a single value publisher (e.g.
Mono
), simply serialize it through theObjectMapper
. -
For a multi-value publisher with
application/json
, by default collect the values withFlux#collectToList()
and then serialize the resulting collection. -
For a multi-value publisher with a streaming media type such as
application/x-ndjson
orapplication/stream+x-jackson-smile
, encode, write, and flush each value individually using a line-delimited JSON format. Other streaming media types may be registered with the encoder. -
For SSE the
Jackson2Encoder
is invoked per event and the output is flushed to ensure delivery without delay.
By default both |
Form Data
FormHttpMessageReader
and FormHttpMessageWriter
support decoding and encoding
application/x-www-form-urlencoded
content.
On the server side where form content often needs to be accessed from multiple places,
ServerWebExchange
provides a dedicated getFormData()
method that parses the content
through FormHttpMessageReader
and then caches the result for repeated access.
See Form Data in the WebHandler
API section.
Once getFormData()
is used, the original raw content can no longer be read from the
request body. For this reason, applications are expected to go through ServerWebExchange
consistently for access to the cached form data versus reading from the raw request body.
Multipart
MultipartHttpMessageReader
and MultipartHttpMessageWriter
support decoding and
encoding "multipart/form-data", "multipart/mixed", and "multipart/related" content.
In turn MultipartHttpMessageReader
delegates to another HttpMessageReader
for the actual parsing to a Flux<Part>
and then simply collects the parts into a MultiValueMap
.
By default, the DefaultPartHttpMessageReader
is used, but this can be changed through the
ServerCodecConfigurer
.
For more information about the DefaultPartHttpMessageReader
, refer to the
javadoc of DefaultPartHttpMessageReader
.
On the server side where multipart form content may need to be accessed from multiple
places, ServerWebExchange
provides a dedicated getMultipartData()
method that parses
the content through MultipartHttpMessageReader
and then caches the result for repeated access.
See Multipart Data in the WebHandler
API section.
Once getMultipartData()
is used, the original raw content can no longer be read from the
request body. For this reason applications have to consistently use getMultipartData()
for repeated, map-like access to parts, or otherwise rely on the
SynchronossPartHttpMessageReader
for a one-time access to Flux<Part>
.
Limits
Decoder
and HttpMessageReader
implementations that buffer some or all of the input
stream can be configured with a limit on the maximum number of bytes to buffer in memory.
In some cases buffering occurs because input is aggregated and represented as a single
object — for example, a controller method with @RequestBody byte[]
,
x-www-form-urlencoded
data, and so on. Buffering can also occur with streaming, when
splitting the input stream — for example, delimited text, a stream of JSON objects, and
so on. For those streaming cases, the limit applies to the number of bytes associated
with one object in the stream.
To configure buffer sizes, you can check if a given Decoder
or HttpMessageReader
exposes a maxInMemorySize
property and if so the Javadoc will have details about default
values. On the server side, ServerCodecConfigurer
provides a single place from where to
set all codecs, see HTTP message codecs. On the client side, the limit for
all codecs can be changed in
WebClient.Builder.
For Multipart parsing the maxInMemorySize
property limits
the size of non-file parts. For file parts, it determines the threshold at which the part
is written to disk. For file parts written to disk, there is an additional
maxDiskUsagePerPart
property to limit the amount of disk space per part. There is also
a maxParts
property to limit the overall number of parts in a multipart request.
To configure all three in WebFlux, you’ll need to supply a pre-configured instance of
MultipartHttpMessageReader
to ServerCodecConfigurer
.
Streaming
When streaming to the HTTP response (for example, text/event-stream
,
application/x-ndjson
), it is important to send data periodically, in order to
reliably detect a disconnected client sooner rather than later. Such a send could be a
comment-only, empty SSE event or any other "no-op" data that would effectively serve as
a heartbeat.
DataBuffer
DataBuffer
is the representation for a byte buffer in WebFlux. The Spring Core part of
this reference has more on that in the section on
Data Buffers and Codecs. The key point to understand is that on some
servers like Netty, byte buffers are pooled and reference counted, and must be released
when consumed to avoid memory leaks.
WebFlux applications generally do not need to be concerned with such issues, unless they consume or produce data buffers directly, as opposed to relying on codecs to convert to and from higher level objects, or unless they choose to create custom codecs. For such cases please review the information in Data Buffers and Codecs, especially the section on Using DataBuffer.
Logging
DEBUG
level logging in Spring WebFlux is designed to be compact, minimal, and
human-friendly. It focuses on high value bits of information that are useful over and
over again vs others that are useful only when debugging a specific issue.
TRACE
level logging generally follows the same principles as DEBUG
(and for example also
should not be a firehose) but can be used for debugging any issue. In addition, some log
messages may show a different level of detail at TRACE
vs DEBUG
.
Good logging comes from the experience of using the logs. If you spot anything that does not meet the stated goals, please let us know.
Log Id
In WebFlux, a single request can be run over multiple threads and the thread ID is not useful for correlating log messages that belong to a specific request. This is why WebFlux log messages are prefixed with a request-specific ID by default.
On the server side, the log ID is stored in the ServerWebExchange
attribute
(LOG_ID_ATTRIBUTE
),
while a fully formatted prefix based on that ID is available from
ServerWebExchange#getLogPrefix()
. On the WebClient
side, the log ID is stored in the
ClientRequest
attribute
(LOG_ID_ATTRIBUTE
)
,while a fully formatted prefix is available from ClientRequest#logPrefix()
.
Sensitive Data
DEBUG
and TRACE
logging can log sensitive information. This is why form parameters and
headers are masked by default and you must explicitly enable their logging in full.
The following example shows how to do so for server-side requests:
-
Java
-
Kotlin
@Configuration
@EnableWebFlux
class MyConfig implements WebFluxConfigurer {
@Override
public void configureHttpMessageCodecs(ServerCodecConfigurer configurer) {
configurer.defaultCodecs().enableLoggingRequestDetails(true);
}
}
@Configuration
@EnableWebFlux
class MyConfig : WebFluxConfigurer {
override fun configureHttpMessageCodecs(configurer: ServerCodecConfigurer) {
configurer.defaultCodecs().enableLoggingRequestDetails(true)
}
}
The following example shows how to do so for client-side requests:
-
Java
-
Kotlin
Consumer<ClientCodecConfigurer> consumer = configurer ->
configurer.defaultCodecs().enableLoggingRequestDetails(true);
WebClient webClient = WebClient.builder()
.exchangeStrategies(strategies -> strategies.codecs(consumer))
.build();
val consumer: (ClientCodecConfigurer) -> Unit = { configurer -> configurer.defaultCodecs().enableLoggingRequestDetails(true) }
val webClient = WebClient.builder()
.exchangeStrategies({ strategies -> strategies.codecs(consumer) })
.build()
Appenders
Logging libraries such as SLF4J and Log4J 2 provide asynchronous loggers that avoid blocking. While those have their own drawbacks such as potentially dropping messages that could not be queued for logging, they are the best available options currently for use in a reactive, non-blocking application.
Custom codecs
Applications can register custom codecs for supporting additional media types, or specific behaviors that are not supported by the default codecs.
Some configuration options expressed by developers are enforced on default codecs. Custom codecs might want to get a chance to align with those preferences, like enforcing buffering limits or logging sensitive data.
The following example shows how to do so for client-side requests:
-
Java
-
Kotlin
WebClient webClient = WebClient.builder()
.codecs(configurer -> {
CustomDecoder decoder = new CustomDecoder();
configurer.customCodecs().registerWithDefaultConfig(decoder);
})
.build();
val webClient = WebClient.builder()
.codecs({ configurer ->
val decoder = CustomDecoder()
configurer.customCodecs().registerWithDefaultConfig(decoder)
})
.build()