io.servicetalk.traffic.resilience.http

Class ServiceRejectionPolicy.Builder

Object

ServiceRejectionPolicy.Builder

Enclosing class:

ServiceRejectionPolicy

public static final class ServiceRejectionPolicy.Builder
extends Object

A ServiceRejectionPolicy builder to support a custom policy.

Constructor Summary

Constructors

Constructor and Description
Builder()

Method Summary

Modifier and Type	Method and Description
ServiceRejectionPolicy	build() Return a custom ServiceRejectionPolicy based on the options of this builder.
ServiceRejectionPolicy.Builder	onLimitResponseBuilder(BiFunction<HttpRequestMetaData,StreamingHttpResponseFactory,Single<StreamingHttpResponse>> onLimitResponseBuilder) Determines the StreamingHttpResponse when a capacity limit is met.
ServiceRejectionPolicy.Builder	onLimitRetryAfter(Consumer<HttpResponseMetaData> onLimitRetryAfter) Determines a retry-after header in the StreamingHttpResponse when a capacity limit is met.
ServiceRejectionPolicy.Builder	onLimitStopAcceptingConnections(boolean stopAccepting) When a certain CapacityLimiter rejects a request due to the active limit, (e.g., no CapacityLimiter.Ticket is returned) influence the server to also stop accepting new connections until the capacity is under healthy conditions again.
ServiceRejectionPolicy.Builder	onOpenCircuitResponseBuilder(BiFunction<HttpRequestMetaData,StreamingHttpResponseFactory,Single<StreamingHttpResponse>> onOpenCircuitResponseBuilder) Determines the StreamingHttpResponse when a circuit-breaker limit is met.
ServiceRejectionPolicy.Builder	onOpenCircuitRetryAfter(BiConsumer<HttpResponseMetaData,StateContext> onOpenCircuitRetryAfter) Determines a retry-after header in the StreamingHttpResponse when a capacity limit is met.

Methods inherited from class Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

Builder

public Builder()

Method Detail

onLimitResponseBuilder

public ServiceRejectionPolicy.Builder onLimitResponseBuilder(BiFunction<HttpRequestMetaData,StreamingHttpResponseFactory,Single<StreamingHttpResponse>> onLimitResponseBuilder)

Determines the StreamingHttpResponse when a capacity limit is met.

Parameters:

onLimitResponseBuilder - A factory function used to generate a StreamingHttpResponse based on the request when a capacity limit is observed.

Returns:

this.

onLimitRetryAfter

public ServiceRejectionPolicy.Builder onLimitRetryAfter(Consumer<HttpResponseMetaData> onLimitRetryAfter)

Determines a retry-after header in the StreamingHttpResponse when a capacity limit is met.

Parameters:

onLimitRetryAfter - A HttpResponseMetaData consumer, that can allow response decoration with additional headers to hint the peer (upon capacity limits) about a possible wait-time before a retry could be issued.

Returns:

this.

onLimitStopAcceptingConnections

public ServiceRejectionPolicy.Builder onLimitStopAcceptingConnections(boolean stopAccepting)

When a certain CapacityLimiter rejects a request due to the active limit, (e.g., no CapacityLimiter.Ticket is returned) influence the server to also stop accepting new connections until the capacity is under healthy conditions again. This setting only works when a CapacityLimiter matches the incoming request, in cases this doesn't hold (see. Builder(Supplier, boolean) Builder's rejectedNotMatched argument}) this won't be effective.

When a server socket stops accepting new connections (see. ServerListenContext.acceptConnections(boolean)) due to capacity concerns, the state will be toggled back when the ticket's terminal callback (dropped, failed, completed, ignored) returns a positive or negative value, demonstrating available capacity or not_supported respectively. When the returned value is 0 that means no-capacity available, which will keep the server in the not-accepting mode.

When enabling this feature, it's recommended for clients using this service to configure timeouts for their opening connection time and connection idleness time. For example, a client without connection-timeout or idle-timeout on the outgoing connections towards this service, won't be able to detect on time the connection delays. Likewise, on the server side you can configure the server backlog to a very small number or even disable it completely, to avoid holding established connections in the OS.

Worth noting that established connections that stay in the OS backlog, usually have a First In First Out behavior, which depending on the size of that queue, may result in extending latencies on newer requests because older ones are served first. Disabling the server backlog will give a better behavior.

Parameters:

stopAccepting - true will allow this filter to control the connection acceptance of the overall server socket.

Returns:

this.

onOpenCircuitResponseBuilder

public ServiceRejectionPolicy.Builder onOpenCircuitResponseBuilder(BiFunction<HttpRequestMetaData,StreamingHttpResponseFactory,Single<StreamingHttpResponse>> onOpenCircuitResponseBuilder)

Determines the StreamingHttpResponse when a circuit-breaker limit is met.

Parameters:

onOpenCircuitResponseBuilder - A factory function used to generate a StreamingHttpResponse based on the request when an open breaker is observed.

Returns:

this.

onOpenCircuitRetryAfter

public ServiceRejectionPolicy.Builder onOpenCircuitRetryAfter(BiConsumer<HttpResponseMetaData,StateContext> onOpenCircuitRetryAfter)

Determines a retry-after header in the StreamingHttpResponse when a capacity limit is met.

Parameters:

onOpenCircuitRetryAfter - A HttpResponseMetaData consumer, that can allow response decoration with additional headers to hint the peer (upon open breaker) about a possible wait-time before a retry could be issued.

Returns:

this.

build

public ServiceRejectionPolicy build()

Return a custom ServiceRejectionPolicy based on the options of this builder.

Returns:

A custom ServiceRejectionPolicy based on the options of this builder.