674 lines
25 KiB
Rust
Raw Normal View History

Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
// Copyright 2020 Parity Technologies (UK) Ltd.
//
// Permission is hereby granted, free of charge, to any person obtaining a
// copy of this software and associated documentation files (the "Software"),
// to deal in the Software without restriction, including without limitation
// the rights to use, copy, modify, merge, publish, distribute, sublicense,
// and/or sell copies of the Software, and to permit persons to whom the
// Software is furnished to do so, subject to the following conditions:
//
// The above copyright notice and this permission notice shall be included in
// all copies or substantial portions of the Software.
//
// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
// OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
// FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
// DEALINGS IN THE SOFTWARE.
//! Generic request/response protocols.
//!
//! ## General Usage
//!
//! [`RequestResponse`] is a `NetworkBehaviour` that implements a generic
//! request/response protocol or protocol family, whereby each request is
//! sent over a new substream on a connection. `RequestResponse` is generic
//! over the actual messages being sent, which are defined in terms of a
//! [`RequestResponseCodec`]. Creating a request/response protocol thus amounts
//! to providing an implementation of this trait which can then be
//! given to [`RequestResponse::new`]. Further configuration options are
//! available via the [`RequestResponseConfig`].
//!
//! Requests are sent using [`RequestResponse::send_request`] and the
//! responses received as [`RequestResponseMessage::Response`] via
//! [`RequestResponseEvent::Message`].
//!
//! Responses are sent using [`RequestResponse::send_response`] upon
//! receiving a [`RequestResponseMessage::Request`] via
//! [`RequestResponseEvent::Message`].
//!
//! ## Protocol Families
//!
//! A single [`RequestResponse`] instance can be used with an entire
//! protocol family that share the same request and response types.
//! For that purpose, [`RequestResponseCodec::Protocol`] is typically
//! instantiated with a sum type.
//!
//! ## Limited Protocol Support
//!
//! It is possible to only support inbound or outbound requests for
//! a particular protocol. This is achieved by instantiating `RequestResponse`
//! with protocols using [`ProtocolSupport::Inbound`] or
//! [`ProtocolSupport::Outbound`]. Any subset of protocols of a protocol
//! family can be configured in this way. Such protocols will not be
//! advertised during inbound respectively outbound protocol negotiation
//! on the substreams.
pub mod codec;
pub mod handler;
Restore `RequestResponse::throttled`. (#1726) * Restore `RequestResponse::throttled`. In contrast to the existing "throttled" approach this PR adds back- pressure to the protocol without requiring pre-existing knowledge of all nodes about their limits. It adds small, CBOR-encoded headers to the actual payload data. Extra credit messages communicate back to the sender how many more requests it is allowed to send. * Remove some noise. * Resend credit grant after connection closed. Should an error in some lower layer cause a connection to be closed, our previously sent credit grant may not have reached the remote peer. Therefore, pessimistically, a credit grant is resent whenever a connection is closed. The remote ignores duplicate grants. * Remove inbound/outbound tracking per peer. * Send ACK as response to duplicate credit grants. * Simplify. * Fix grammar. * Incorporate review feedback. - Remove `ResponseSent` which was a leftover from previous attemps and issue a credit grant immediately in `send_response`. - Only resend credit grants after a connection is closed if we are still connected to this peer. * Move codec/header.rs to throttled/codec.rs. * More review suggestions. * Generalise `ProtocolWrapper` and use shorter prefix. * Update protocols/request-response/src/lib.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Minor comment changes. * Limit max. header size to 8KiB * Always construct initial limit with 1. Since honest senders always assume a send budget of 1 and wait for credit afterwards, setting the default limit to a higher value can only become effective after informing the peer about it which means leaving `max_recv` at 1 and setting `next_max` to the desired value. Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
2020-09-07 17:22:40 +02:00
pub mod throttled;
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
pub use codec::{RequestResponseCodec, ProtocolName};
pub use handler::ProtocolSupport;
Restore `RequestResponse::throttled`. (#1726) * Restore `RequestResponse::throttled`. In contrast to the existing "throttled" approach this PR adds back- pressure to the protocol without requiring pre-existing knowledge of all nodes about their limits. It adds small, CBOR-encoded headers to the actual payload data. Extra credit messages communicate back to the sender how many more requests it is allowed to send. * Remove some noise. * Resend credit grant after connection closed. Should an error in some lower layer cause a connection to be closed, our previously sent credit grant may not have reached the remote peer. Therefore, pessimistically, a credit grant is resent whenever a connection is closed. The remote ignores duplicate grants. * Remove inbound/outbound tracking per peer. * Send ACK as response to duplicate credit grants. * Simplify. * Fix grammar. * Incorporate review feedback. - Remove `ResponseSent` which was a leftover from previous attemps and issue a credit grant immediately in `send_response`. - Only resend credit grants after a connection is closed if we are still connected to this peer. * Move codec/header.rs to throttled/codec.rs. * More review suggestions. * Generalise `ProtocolWrapper` and use shorter prefix. * Update protocols/request-response/src/lib.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Minor comment changes. * Limit max. header size to 8KiB * Always construct initial limit with 1. Since honest senders always assume a send budget of 1 and wait for credit afterwards, setting the default limit to a higher value can only become effective after informing the peer about it which means leaving `max_recv` at 1 and setting `next_max` to the desired value. Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
2020-09-07 17:22:40 +02:00
pub use throttled::Throttled;
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
use futures::{
channel::oneshot,
};
use handler::{
RequestProtocol,
RequestResponseHandler,
RequestResponseHandlerEvent,
};
use libp2p_core::{
ConnectedPoint,
Multiaddr,
PeerId,
connection::ConnectionId,
};
use libp2p_swarm::{
DialPeerCondition,
NetworkBehaviour,
NetworkBehaviourAction,
NotifyHandler,
PollParameters,
};
use smallvec::SmallVec;
use std::{
collections::{VecDeque, HashMap},
Restore `RequestResponse::throttled`. (#1726) * Restore `RequestResponse::throttled`. In contrast to the existing "throttled" approach this PR adds back- pressure to the protocol without requiring pre-existing knowledge of all nodes about their limits. It adds small, CBOR-encoded headers to the actual payload data. Extra credit messages communicate back to the sender how many more requests it is allowed to send. * Remove some noise. * Resend credit grant after connection closed. Should an error in some lower layer cause a connection to be closed, our previously sent credit grant may not have reached the remote peer. Therefore, pessimistically, a credit grant is resent whenever a connection is closed. The remote ignores duplicate grants. * Remove inbound/outbound tracking per peer. * Send ACK as response to duplicate credit grants. * Simplify. * Fix grammar. * Incorporate review feedback. - Remove `ResponseSent` which was a leftover from previous attemps and issue a credit grant immediately in `send_response`. - Only resend credit grants after a connection is closed if we are still connected to this peer. * Move codec/header.rs to throttled/codec.rs. * More review suggestions. * Generalise `ProtocolWrapper` and use shorter prefix. * Update protocols/request-response/src/lib.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Minor comment changes. * Limit max. header size to 8KiB * Always construct initial limit with 1. Since honest senders always assume a send budget of 1 and wait for credit afterwards, setting the default limit to a higher value can only become effective after informing the peer about it which means leaving `max_recv` at 1 and setting `next_max` to the desired value. Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
2020-09-07 17:22:40 +02:00
fmt,
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
time::Duration,
Restore `RequestResponse::throttled`. (#1726) * Restore `RequestResponse::throttled`. In contrast to the existing "throttled" approach this PR adds back- pressure to the protocol without requiring pre-existing knowledge of all nodes about their limits. It adds small, CBOR-encoded headers to the actual payload data. Extra credit messages communicate back to the sender how many more requests it is allowed to send. * Remove some noise. * Resend credit grant after connection closed. Should an error in some lower layer cause a connection to be closed, our previously sent credit grant may not have reached the remote peer. Therefore, pessimistically, a credit grant is resent whenever a connection is closed. The remote ignores duplicate grants. * Remove inbound/outbound tracking per peer. * Send ACK as response to duplicate credit grants. * Simplify. * Fix grammar. * Incorporate review feedback. - Remove `ResponseSent` which was a leftover from previous attemps and issue a credit grant immediately in `send_response`. - Only resend credit grants after a connection is closed if we are still connected to this peer. * Move codec/header.rs to throttled/codec.rs. * More review suggestions. * Generalise `ProtocolWrapper` and use shorter prefix. * Update protocols/request-response/src/lib.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Minor comment changes. * Limit max. header size to 8KiB * Always construct initial limit with 1. Since honest senders always assume a send budget of 1 and wait for credit afterwards, setting the default limit to a higher value can only become effective after informing the peer about it which means leaving `max_recv` at 1 and setting `next_max` to the desired value. Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
2020-09-07 17:22:40 +02:00
sync::{atomic::AtomicU64, Arc},
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
task::{Context, Poll}
};
/// An inbound request or response.
#[derive(Debug)]
Restore `RequestResponse::throttled`. (#1726) * Restore `RequestResponse::throttled`. In contrast to the existing "throttled" approach this PR adds back- pressure to the protocol without requiring pre-existing knowledge of all nodes about their limits. It adds small, CBOR-encoded headers to the actual payload data. Extra credit messages communicate back to the sender how many more requests it is allowed to send. * Remove some noise. * Resend credit grant after connection closed. Should an error in some lower layer cause a connection to be closed, our previously sent credit grant may not have reached the remote peer. Therefore, pessimistically, a credit grant is resent whenever a connection is closed. The remote ignores duplicate grants. * Remove inbound/outbound tracking per peer. * Send ACK as response to duplicate credit grants. * Simplify. * Fix grammar. * Incorporate review feedback. - Remove `ResponseSent` which was a leftover from previous attemps and issue a credit grant immediately in `send_response`. - Only resend credit grants after a connection is closed if we are still connected to this peer. * Move codec/header.rs to throttled/codec.rs. * More review suggestions. * Generalise `ProtocolWrapper` and use shorter prefix. * Update protocols/request-response/src/lib.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Minor comment changes. * Limit max. header size to 8KiB * Always construct initial limit with 1. Since honest senders always assume a send budget of 1 and wait for credit afterwards, setting the default limit to a higher value can only become effective after informing the peer about it which means leaving `max_recv` at 1 and setting `next_max` to the desired value. Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
2020-09-07 17:22:40 +02:00
pub enum RequestResponseMessage<TRequest, TResponse, TChannelResponse = TResponse> {
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
/// A request message.
Request {
Restore `RequestResponse::throttled`. (#1726) * Restore `RequestResponse::throttled`. In contrast to the existing "throttled" approach this PR adds back- pressure to the protocol without requiring pre-existing knowledge of all nodes about their limits. It adds small, CBOR-encoded headers to the actual payload data. Extra credit messages communicate back to the sender how many more requests it is allowed to send. * Remove some noise. * Resend credit grant after connection closed. Should an error in some lower layer cause a connection to be closed, our previously sent credit grant may not have reached the remote peer. Therefore, pessimistically, a credit grant is resent whenever a connection is closed. The remote ignores duplicate grants. * Remove inbound/outbound tracking per peer. * Send ACK as response to duplicate credit grants. * Simplify. * Fix grammar. * Incorporate review feedback. - Remove `ResponseSent` which was a leftover from previous attemps and issue a credit grant immediately in `send_response`. - Only resend credit grants after a connection is closed if we are still connected to this peer. * Move codec/header.rs to throttled/codec.rs. * More review suggestions. * Generalise `ProtocolWrapper` and use shorter prefix. * Update protocols/request-response/src/lib.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Minor comment changes. * Limit max. header size to 8KiB * Always construct initial limit with 1. Since honest senders always assume a send budget of 1 and wait for credit afterwards, setting the default limit to a higher value can only become effective after informing the peer about it which means leaving `max_recv` at 1 and setting `next_max` to the desired value. Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
2020-09-07 17:22:40 +02:00
/// The ID of this request.
request_id: RequestId,
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
/// The request message.
request: TRequest,
[request-response] Refine success & error reporting for inbound requests. (#1867) * Refine error reporting for inbound request handling. At the moment one can neither get confirmation when a response has been sent on the underlying transport, nor is one aware of response omissions. The latter was originally intended as a feature for support of one-way protocols, which seems like a bad idea in hindsight. The lack of notification for sent responses may prohibit implementation of some request-response protocols that need to ensure a happens-before relation between sending a response and a subsequent request, besides uses for collecting statistics. Even with these changes, there is no active notification for failed inbound requests as a result of connections unexpectedly closing, as is the case for outbound requests. Instead, for pending inbound requests this scenario can be identified if necessary by the absense of both `InboundFailure` and `ResponseSent` events for a particular previously received request. Interest in this situation is not expected to be common and would otherwise require explicitly tracking all inbound requests in the `RequestResponse` behaviour, which would be a pity. `RequestResponse::send_response` now also synchronously returns an error if the inbound upgrade handling the request has been aborted, due to timeout or closing of the connection, giving more options for graceful error handling for inbound requests. As an aside, the `Throttled` wrapper now no longer emits inbound or outbound error events occurring in the context of sending credit requests or responses. This is in addition to not emitting `ResponseSent` events for ACK responses of credit grants. * Update protocols/request-response/src/lib.rs Co-authored-by: Max Inden <mail@max-inden.de> * Address some minor clippy warnings. (#1868) * Track pending credit request IDs. In order to avoid emitting events relating to credit grants or acks on the public API. The public API should only emit events relating to the actual requests and responses sent by client code. * Small cleanup * Cleanup * Update versions and changelogs. * Unreleased Co-authored-by: Max Inden <mail@max-inden.de>
2020-12-07 13:07:47 +01:00
/// The channel waiting for the response.
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
///
[request-response] Refine success & error reporting for inbound requests. (#1867) * Refine error reporting for inbound request handling. At the moment one can neither get confirmation when a response has been sent on the underlying transport, nor is one aware of response omissions. The latter was originally intended as a feature for support of one-way protocols, which seems like a bad idea in hindsight. The lack of notification for sent responses may prohibit implementation of some request-response protocols that need to ensure a happens-before relation between sending a response and a subsequent request, besides uses for collecting statistics. Even with these changes, there is no active notification for failed inbound requests as a result of connections unexpectedly closing, as is the case for outbound requests. Instead, for pending inbound requests this scenario can be identified if necessary by the absense of both `InboundFailure` and `ResponseSent` events for a particular previously received request. Interest in this situation is not expected to be common and would otherwise require explicitly tracking all inbound requests in the `RequestResponse` behaviour, which would be a pity. `RequestResponse::send_response` now also synchronously returns an error if the inbound upgrade handling the request has been aborted, due to timeout or closing of the connection, giving more options for graceful error handling for inbound requests. As an aside, the `Throttled` wrapper now no longer emits inbound or outbound error events occurring in the context of sending credit requests or responses. This is in addition to not emitting `ResponseSent` events for ACK responses of credit grants. * Update protocols/request-response/src/lib.rs Co-authored-by: Max Inden <mail@max-inden.de> * Address some minor clippy warnings. (#1868) * Track pending credit request IDs. In order to avoid emitting events relating to credit grants or acks on the public API. The public API should only emit events relating to the actual requests and responses sent by client code. * Small cleanup * Cleanup * Update versions and changelogs. * Unreleased Co-authored-by: Max Inden <mail@max-inden.de>
2020-12-07 13:07:47 +01:00
/// If this channel is dropped instead of being used to send a response
/// via [`RequestResponse::send_response`], a [`RequestResponseEvent::InboundFailure`]
/// with [`InboundFailure::ResponseOmission`] is emitted.
Restore `RequestResponse::throttled`. (#1726) * Restore `RequestResponse::throttled`. In contrast to the existing "throttled" approach this PR adds back- pressure to the protocol without requiring pre-existing knowledge of all nodes about their limits. It adds small, CBOR-encoded headers to the actual payload data. Extra credit messages communicate back to the sender how many more requests it is allowed to send. * Remove some noise. * Resend credit grant after connection closed. Should an error in some lower layer cause a connection to be closed, our previously sent credit grant may not have reached the remote peer. Therefore, pessimistically, a credit grant is resent whenever a connection is closed. The remote ignores duplicate grants. * Remove inbound/outbound tracking per peer. * Send ACK as response to duplicate credit grants. * Simplify. * Fix grammar. * Incorporate review feedback. - Remove `ResponseSent` which was a leftover from previous attemps and issue a credit grant immediately in `send_response`. - Only resend credit grants after a connection is closed if we are still connected to this peer. * Move codec/header.rs to throttled/codec.rs. * More review suggestions. * Generalise `ProtocolWrapper` and use shorter prefix. * Update protocols/request-response/src/lib.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Minor comment changes. * Limit max. header size to 8KiB * Always construct initial limit with 1. Since honest senders always assume a send budget of 1 and wait for credit afterwards, setting the default limit to a higher value can only become effective after informing the peer about it which means leaving `max_recv` at 1 and setting `next_max` to the desired value. Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
2020-09-07 17:22:40 +02:00
channel: ResponseChannel<TChannelResponse>,
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
},
/// A response message.
Response {
/// The ID of the request that produced this response.
///
/// See [`RequestResponse::send_request`].
request_id: RequestId,
/// The response message.
response: TResponse
},
}
/// The events emitted by a [`RequestResponse`] protocol.
#[derive(Debug)]
Restore `RequestResponse::throttled`. (#1726) * Restore `RequestResponse::throttled`. In contrast to the existing "throttled" approach this PR adds back- pressure to the protocol without requiring pre-existing knowledge of all nodes about their limits. It adds small, CBOR-encoded headers to the actual payload data. Extra credit messages communicate back to the sender how many more requests it is allowed to send. * Remove some noise. * Resend credit grant after connection closed. Should an error in some lower layer cause a connection to be closed, our previously sent credit grant may not have reached the remote peer. Therefore, pessimistically, a credit grant is resent whenever a connection is closed. The remote ignores duplicate grants. * Remove inbound/outbound tracking per peer. * Send ACK as response to duplicate credit grants. * Simplify. * Fix grammar. * Incorporate review feedback. - Remove `ResponseSent` which was a leftover from previous attemps and issue a credit grant immediately in `send_response`. - Only resend credit grants after a connection is closed if we are still connected to this peer. * Move codec/header.rs to throttled/codec.rs. * More review suggestions. * Generalise `ProtocolWrapper` and use shorter prefix. * Update protocols/request-response/src/lib.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Minor comment changes. * Limit max. header size to 8KiB * Always construct initial limit with 1. Since honest senders always assume a send budget of 1 and wait for credit afterwards, setting the default limit to a higher value can only become effective after informing the peer about it which means leaving `max_recv` at 1 and setting `next_max` to the desired value. Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
2020-09-07 17:22:40 +02:00
pub enum RequestResponseEvent<TRequest, TResponse, TChannelResponse = TResponse> {
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
/// An incoming message (request or response).
Message {
/// The peer who sent the message.
peer: PeerId,
/// The incoming message.
Restore `RequestResponse::throttled`. (#1726) * Restore `RequestResponse::throttled`. In contrast to the existing "throttled" approach this PR adds back- pressure to the protocol without requiring pre-existing knowledge of all nodes about their limits. It adds small, CBOR-encoded headers to the actual payload data. Extra credit messages communicate back to the sender how many more requests it is allowed to send. * Remove some noise. * Resend credit grant after connection closed. Should an error in some lower layer cause a connection to be closed, our previously sent credit grant may not have reached the remote peer. Therefore, pessimistically, a credit grant is resent whenever a connection is closed. The remote ignores duplicate grants. * Remove inbound/outbound tracking per peer. * Send ACK as response to duplicate credit grants. * Simplify. * Fix grammar. * Incorporate review feedback. - Remove `ResponseSent` which was a leftover from previous attemps and issue a credit grant immediately in `send_response`. - Only resend credit grants after a connection is closed if we are still connected to this peer. * Move codec/header.rs to throttled/codec.rs. * More review suggestions. * Generalise `ProtocolWrapper` and use shorter prefix. * Update protocols/request-response/src/lib.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Minor comment changes. * Limit max. header size to 8KiB * Always construct initial limit with 1. Since honest senders always assume a send budget of 1 and wait for credit afterwards, setting the default limit to a higher value can only become effective after informing the peer about it which means leaving `max_recv` at 1 and setting `next_max` to the desired value. Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
2020-09-07 17:22:40 +02:00
message: RequestResponseMessage<TRequest, TResponse, TChannelResponse>
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
},
/// An outbound request failed.
OutboundFailure {
/// The peer to whom the request was sent.
peer: PeerId,
/// The (local) ID of the failed request.
request_id: RequestId,
/// The error that occurred.
error: OutboundFailure,
},
/// An inbound request failed.
[request-response] Refine success & error reporting for inbound requests. (#1867) * Refine error reporting for inbound request handling. At the moment one can neither get confirmation when a response has been sent on the underlying transport, nor is one aware of response omissions. The latter was originally intended as a feature for support of one-way protocols, which seems like a bad idea in hindsight. The lack of notification for sent responses may prohibit implementation of some request-response protocols that need to ensure a happens-before relation between sending a response and a subsequent request, besides uses for collecting statistics. Even with these changes, there is no active notification for failed inbound requests as a result of connections unexpectedly closing, as is the case for outbound requests. Instead, for pending inbound requests this scenario can be identified if necessary by the absense of both `InboundFailure` and `ResponseSent` events for a particular previously received request. Interest in this situation is not expected to be common and would otherwise require explicitly tracking all inbound requests in the `RequestResponse` behaviour, which would be a pity. `RequestResponse::send_response` now also synchronously returns an error if the inbound upgrade handling the request has been aborted, due to timeout or closing of the connection, giving more options for graceful error handling for inbound requests. As an aside, the `Throttled` wrapper now no longer emits inbound or outbound error events occurring in the context of sending credit requests or responses. This is in addition to not emitting `ResponseSent` events for ACK responses of credit grants. * Update protocols/request-response/src/lib.rs Co-authored-by: Max Inden <mail@max-inden.de> * Address some minor clippy warnings. (#1868) * Track pending credit request IDs. In order to avoid emitting events relating to credit grants or acks on the public API. The public API should only emit events relating to the actual requests and responses sent by client code. * Small cleanup * Cleanup * Update versions and changelogs. * Unreleased Co-authored-by: Max Inden <mail@max-inden.de>
2020-12-07 13:07:47 +01:00
///
/// > **Note**: The case whereby a connection on which a response is sent
/// > closes after [`RequestResponse::send_response`] already succeeded
/// > but before the response could be sent on the connection is reflected
/// > by there being no [`RequestResponseEvent::ResponseSent`] event.
/// > Code interested in ensuring a response has been successfully
/// > handed to the transport layer, e.g. before continuing with the next
/// > step in a multi-step protocol, should listen to these events.
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
InboundFailure {
/// The peer from whom the request was received.
peer: PeerId,
Restore `RequestResponse::throttled`. (#1726) * Restore `RequestResponse::throttled`. In contrast to the existing "throttled" approach this PR adds back- pressure to the protocol without requiring pre-existing knowledge of all nodes about their limits. It adds small, CBOR-encoded headers to the actual payload data. Extra credit messages communicate back to the sender how many more requests it is allowed to send. * Remove some noise. * Resend credit grant after connection closed. Should an error in some lower layer cause a connection to be closed, our previously sent credit grant may not have reached the remote peer. Therefore, pessimistically, a credit grant is resent whenever a connection is closed. The remote ignores duplicate grants. * Remove inbound/outbound tracking per peer. * Send ACK as response to duplicate credit grants. * Simplify. * Fix grammar. * Incorporate review feedback. - Remove `ResponseSent` which was a leftover from previous attemps and issue a credit grant immediately in `send_response`. - Only resend credit grants after a connection is closed if we are still connected to this peer. * Move codec/header.rs to throttled/codec.rs. * More review suggestions. * Generalise `ProtocolWrapper` and use shorter prefix. * Update protocols/request-response/src/lib.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Minor comment changes. * Limit max. header size to 8KiB * Always construct initial limit with 1. Since honest senders always assume a send budget of 1 and wait for credit afterwards, setting the default limit to a higher value can only become effective after informing the peer about it which means leaving `max_recv` at 1 and setting `next_max` to the desired value. Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
2020-09-07 17:22:40 +02:00
/// The ID of the failed inbound request.
request_id: RequestId,
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
/// The error that occurred.
error: InboundFailure,
},
[request-response] Refine success & error reporting for inbound requests. (#1867) * Refine error reporting for inbound request handling. At the moment one can neither get confirmation when a response has been sent on the underlying transport, nor is one aware of response omissions. The latter was originally intended as a feature for support of one-way protocols, which seems like a bad idea in hindsight. The lack of notification for sent responses may prohibit implementation of some request-response protocols that need to ensure a happens-before relation between sending a response and a subsequent request, besides uses for collecting statistics. Even with these changes, there is no active notification for failed inbound requests as a result of connections unexpectedly closing, as is the case for outbound requests. Instead, for pending inbound requests this scenario can be identified if necessary by the absense of both `InboundFailure` and `ResponseSent` events for a particular previously received request. Interest in this situation is not expected to be common and would otherwise require explicitly tracking all inbound requests in the `RequestResponse` behaviour, which would be a pity. `RequestResponse::send_response` now also synchronously returns an error if the inbound upgrade handling the request has been aborted, due to timeout or closing of the connection, giving more options for graceful error handling for inbound requests. As an aside, the `Throttled` wrapper now no longer emits inbound or outbound error events occurring in the context of sending credit requests or responses. This is in addition to not emitting `ResponseSent` events for ACK responses of credit grants. * Update protocols/request-response/src/lib.rs Co-authored-by: Max Inden <mail@max-inden.de> * Address some minor clippy warnings. (#1868) * Track pending credit request IDs. In order to avoid emitting events relating to credit grants or acks on the public API. The public API should only emit events relating to the actual requests and responses sent by client code. * Small cleanup * Cleanup * Update versions and changelogs. * Unreleased Co-authored-by: Max Inden <mail@max-inden.de>
2020-12-07 13:07:47 +01:00
/// A response to an inbound request has been sent.
///
/// When this event is received, the response has been flushed on
/// the underlying transport connection.
ResponseSent {
/// The peer to whom the response was sent.
peer: PeerId,
/// The ID of the inbound request whose response was sent.
request_id: RequestId,
},
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
}
/// Possible failures occurring in the context of sending
/// an outbound request and receiving the response.
#[derive(Debug)]
pub enum OutboundFailure {
/// The request could not be sent because a dialing attempt failed.
DialFailure,
/// The request timed out before a response was received.
///
/// It is not known whether the request may have been
/// received (and processed) by the remote peer.
Timeout,
/// The connection closed before a response was received.
///
/// It is not known whether the request may have been
/// received (and processed) by the remote peer.
ConnectionClosed,
/// The remote supports none of the requested protocols.
UnsupportedProtocols,
}
/// Possible failures occurring in the context of receiving an
/// inbound request and sending a response.
#[derive(Debug)]
pub enum InboundFailure {
/// The inbound request timed out, either while reading the
[request-response] Refine success & error reporting for inbound requests. (#1867) * Refine error reporting for inbound request handling. At the moment one can neither get confirmation when a response has been sent on the underlying transport, nor is one aware of response omissions. The latter was originally intended as a feature for support of one-way protocols, which seems like a bad idea in hindsight. The lack of notification for sent responses may prohibit implementation of some request-response protocols that need to ensure a happens-before relation between sending a response and a subsequent request, besides uses for collecting statistics. Even with these changes, there is no active notification for failed inbound requests as a result of connections unexpectedly closing, as is the case for outbound requests. Instead, for pending inbound requests this scenario can be identified if necessary by the absense of both `InboundFailure` and `ResponseSent` events for a particular previously received request. Interest in this situation is not expected to be common and would otherwise require explicitly tracking all inbound requests in the `RequestResponse` behaviour, which would be a pity. `RequestResponse::send_response` now also synchronously returns an error if the inbound upgrade handling the request has been aborted, due to timeout or closing of the connection, giving more options for graceful error handling for inbound requests. As an aside, the `Throttled` wrapper now no longer emits inbound or outbound error events occurring in the context of sending credit requests or responses. This is in addition to not emitting `ResponseSent` events for ACK responses of credit grants. * Update protocols/request-response/src/lib.rs Co-authored-by: Max Inden <mail@max-inden.de> * Address some minor clippy warnings. (#1868) * Track pending credit request IDs. In order to avoid emitting events relating to credit grants or acks on the public API. The public API should only emit events relating to the actual requests and responses sent by client code. * Small cleanup * Cleanup * Update versions and changelogs. * Unreleased Co-authored-by: Max Inden <mail@max-inden.de>
2020-12-07 13:07:47 +01:00
/// incoming request or before a response is sent, e.g. if
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
/// [`RequestResponse::send_response`] is not called in a
/// timely manner.
Timeout,
[request-response] Refine success & error reporting for inbound requests. (#1867) * Refine error reporting for inbound request handling. At the moment one can neither get confirmation when a response has been sent on the underlying transport, nor is one aware of response omissions. The latter was originally intended as a feature for support of one-way protocols, which seems like a bad idea in hindsight. The lack of notification for sent responses may prohibit implementation of some request-response protocols that need to ensure a happens-before relation between sending a response and a subsequent request, besides uses for collecting statistics. Even with these changes, there is no active notification for failed inbound requests as a result of connections unexpectedly closing, as is the case for outbound requests. Instead, for pending inbound requests this scenario can be identified if necessary by the absense of both `InboundFailure` and `ResponseSent` events for a particular previously received request. Interest in this situation is not expected to be common and would otherwise require explicitly tracking all inbound requests in the `RequestResponse` behaviour, which would be a pity. `RequestResponse::send_response` now also synchronously returns an error if the inbound upgrade handling the request has been aborted, due to timeout or closing of the connection, giving more options for graceful error handling for inbound requests. As an aside, the `Throttled` wrapper now no longer emits inbound or outbound error events occurring in the context of sending credit requests or responses. This is in addition to not emitting `ResponseSent` events for ACK responses of credit grants. * Update protocols/request-response/src/lib.rs Co-authored-by: Max Inden <mail@max-inden.de> * Address some minor clippy warnings. (#1868) * Track pending credit request IDs. In order to avoid emitting events relating to credit grants or acks on the public API. The public API should only emit events relating to the actual requests and responses sent by client code. * Small cleanup * Cleanup * Update versions and changelogs. * Unreleased Co-authored-by: Max Inden <mail@max-inden.de>
2020-12-07 13:07:47 +01:00
/// The local peer supports none of the protocols requested
/// by the remote.
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
UnsupportedProtocols,
[request-response] Refine success & error reporting for inbound requests. (#1867) * Refine error reporting for inbound request handling. At the moment one can neither get confirmation when a response has been sent on the underlying transport, nor is one aware of response omissions. The latter was originally intended as a feature for support of one-way protocols, which seems like a bad idea in hindsight. The lack of notification for sent responses may prohibit implementation of some request-response protocols that need to ensure a happens-before relation between sending a response and a subsequent request, besides uses for collecting statistics. Even with these changes, there is no active notification for failed inbound requests as a result of connections unexpectedly closing, as is the case for outbound requests. Instead, for pending inbound requests this scenario can be identified if necessary by the absense of both `InboundFailure` and `ResponseSent` events for a particular previously received request. Interest in this situation is not expected to be common and would otherwise require explicitly tracking all inbound requests in the `RequestResponse` behaviour, which would be a pity. `RequestResponse::send_response` now also synchronously returns an error if the inbound upgrade handling the request has been aborted, due to timeout or closing of the connection, giving more options for graceful error handling for inbound requests. As an aside, the `Throttled` wrapper now no longer emits inbound or outbound error events occurring in the context of sending credit requests or responses. This is in addition to not emitting `ResponseSent` events for ACK responses of credit grants. * Update protocols/request-response/src/lib.rs Co-authored-by: Max Inden <mail@max-inden.de> * Address some minor clippy warnings. (#1868) * Track pending credit request IDs. In order to avoid emitting events relating to credit grants or acks on the public API. The public API should only emit events relating to the actual requests and responses sent by client code. * Small cleanup * Cleanup * Update versions and changelogs. * Unreleased Co-authored-by: Max Inden <mail@max-inden.de>
2020-12-07 13:07:47 +01:00
/// The local peer failed to respond to an inbound request
/// due to the [`ResponseChannel`] being dropped instead of
/// being passed to [`RequestResponse::send_response`].
ResponseOmission,
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
}
/// A channel for sending a response to an inbound request.
///
/// See [`RequestResponse::send_response`].
#[derive(Debug)]
pub struct ResponseChannel<TResponse> {
Restore `RequestResponse::throttled`. (#1726) * Restore `RequestResponse::throttled`. In contrast to the existing "throttled" approach this PR adds back- pressure to the protocol without requiring pre-existing knowledge of all nodes about their limits. It adds small, CBOR-encoded headers to the actual payload data. Extra credit messages communicate back to the sender how many more requests it is allowed to send. * Remove some noise. * Resend credit grant after connection closed. Should an error in some lower layer cause a connection to be closed, our previously sent credit grant may not have reached the remote peer. Therefore, pessimistically, a credit grant is resent whenever a connection is closed. The remote ignores duplicate grants. * Remove inbound/outbound tracking per peer. * Send ACK as response to duplicate credit grants. * Simplify. * Fix grammar. * Incorporate review feedback. - Remove `ResponseSent` which was a leftover from previous attemps and issue a credit grant immediately in `send_response`. - Only resend credit grants after a connection is closed if we are still connected to this peer. * Move codec/header.rs to throttled/codec.rs. * More review suggestions. * Generalise `ProtocolWrapper` and use shorter prefix. * Update protocols/request-response/src/lib.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Minor comment changes. * Limit max. header size to 8KiB * Always construct initial limit with 1. Since honest senders always assume a send budget of 1 and wait for credit afterwards, setting the default limit to a higher value can only become effective after informing the peer about it which means leaving `max_recv` at 1 and setting `next_max` to the desired value. Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
2020-09-07 17:22:40 +02:00
request_id: RequestId,
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
peer: PeerId,
sender: oneshot::Sender<TResponse>,
}
impl<TResponse> ResponseChannel<TResponse> {
/// Checks whether the response channel is still open, i.e.
/// the `RequestResponse` behaviour is still waiting for a
/// a response to be sent via [`RequestResponse::send_response`]
/// and this response channel.
///
/// If the response channel is no longer open then the inbound
/// request timed out waiting for the response.
pub fn is_open(&self) -> bool {
!self.sender.is_canceled()
}
Restore `RequestResponse::throttled`. (#1726) * Restore `RequestResponse::throttled`. In contrast to the existing "throttled" approach this PR adds back- pressure to the protocol without requiring pre-existing knowledge of all nodes about their limits. It adds small, CBOR-encoded headers to the actual payload data. Extra credit messages communicate back to the sender how many more requests it is allowed to send. * Remove some noise. * Resend credit grant after connection closed. Should an error in some lower layer cause a connection to be closed, our previously sent credit grant may not have reached the remote peer. Therefore, pessimistically, a credit grant is resent whenever a connection is closed. The remote ignores duplicate grants. * Remove inbound/outbound tracking per peer. * Send ACK as response to duplicate credit grants. * Simplify. * Fix grammar. * Incorporate review feedback. - Remove `ResponseSent` which was a leftover from previous attemps and issue a credit grant immediately in `send_response`. - Only resend credit grants after a connection is closed if we are still connected to this peer. * Move codec/header.rs to throttled/codec.rs. * More review suggestions. * Generalise `ProtocolWrapper` and use shorter prefix. * Update protocols/request-response/src/lib.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Minor comment changes. * Limit max. header size to 8KiB * Always construct initial limit with 1. Since honest senders always assume a send budget of 1 and wait for credit afterwards, setting the default limit to a higher value can only become effective after informing the peer about it which means leaving `max_recv` at 1 and setting `next_max` to the desired value. Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
2020-09-07 17:22:40 +02:00
/// Get the ID of the inbound request waiting for a response.
pub(crate) fn request_id(&self) -> RequestId {
self.request_id
}
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
}
Restore `RequestResponse::throttled`. (#1726) * Restore `RequestResponse::throttled`. In contrast to the existing "throttled" approach this PR adds back- pressure to the protocol without requiring pre-existing knowledge of all nodes about their limits. It adds small, CBOR-encoded headers to the actual payload data. Extra credit messages communicate back to the sender how many more requests it is allowed to send. * Remove some noise. * Resend credit grant after connection closed. Should an error in some lower layer cause a connection to be closed, our previously sent credit grant may not have reached the remote peer. Therefore, pessimistically, a credit grant is resent whenever a connection is closed. The remote ignores duplicate grants. * Remove inbound/outbound tracking per peer. * Send ACK as response to duplicate credit grants. * Simplify. * Fix grammar. * Incorporate review feedback. - Remove `ResponseSent` which was a leftover from previous attemps and issue a credit grant immediately in `send_response`. - Only resend credit grants after a connection is closed if we are still connected to this peer. * Move codec/header.rs to throttled/codec.rs. * More review suggestions. * Generalise `ProtocolWrapper` and use shorter prefix. * Update protocols/request-response/src/lib.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Minor comment changes. * Limit max. header size to 8KiB * Always construct initial limit with 1. Since honest senders always assume a send budget of 1 and wait for credit afterwards, setting the default limit to a higher value can only become effective after informing the peer about it which means leaving `max_recv` at 1 and setting `next_max` to the desired value. Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
2020-09-07 17:22:40 +02:00
/// The ID of an inbound or outbound request.
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
#[derive(Debug, Copy, Clone, PartialEq, Eq, Hash)]
pub struct RequestId(u64);
Restore `RequestResponse::throttled`. (#1726) * Restore `RequestResponse::throttled`. In contrast to the existing "throttled" approach this PR adds back- pressure to the protocol without requiring pre-existing knowledge of all nodes about their limits. It adds small, CBOR-encoded headers to the actual payload data. Extra credit messages communicate back to the sender how many more requests it is allowed to send. * Remove some noise. * Resend credit grant after connection closed. Should an error in some lower layer cause a connection to be closed, our previously sent credit grant may not have reached the remote peer. Therefore, pessimistically, a credit grant is resent whenever a connection is closed. The remote ignores duplicate grants. * Remove inbound/outbound tracking per peer. * Send ACK as response to duplicate credit grants. * Simplify. * Fix grammar. * Incorporate review feedback. - Remove `ResponseSent` which was a leftover from previous attemps and issue a credit grant immediately in `send_response`. - Only resend credit grants after a connection is closed if we are still connected to this peer. * Move codec/header.rs to throttled/codec.rs. * More review suggestions. * Generalise `ProtocolWrapper` and use shorter prefix. * Update protocols/request-response/src/lib.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Minor comment changes. * Limit max. header size to 8KiB * Always construct initial limit with 1. Since honest senders always assume a send budget of 1 and wait for credit afterwards, setting the default limit to a higher value can only become effective after informing the peer about it which means leaving `max_recv` at 1 and setting `next_max` to the desired value. Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
2020-09-07 17:22:40 +02:00
impl fmt::Display for RequestId {
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
write!(f, "{}", self.0)
}
}
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
/// The configuration for a `RequestResponse` protocol.
#[derive(Debug, Clone)]
pub struct RequestResponseConfig {
request_timeout: Duration,
connection_keep_alive: Duration,
}
impl Default for RequestResponseConfig {
fn default() -> Self {
Self {
connection_keep_alive: Duration::from_secs(10),
request_timeout: Duration::from_secs(10),
}
}
}
impl RequestResponseConfig {
/// Sets the keep-alive timeout of idle connections.
pub fn set_connection_keep_alive(&mut self, v: Duration) -> &mut Self {
self.connection_keep_alive = v;
self
}
/// Sets the timeout for inbound and outbound requests.
pub fn set_request_timeout(&mut self, v: Duration) -> &mut Self {
self.request_timeout = v;
self
}
}
/// A request/response protocol for some message codec.
pub struct RequestResponse<TCodec>
where
TCodec: RequestResponseCodec,
{
/// The supported inbound protocols.
inbound_protocols: SmallVec<[TCodec::Protocol; 2]>,
/// The supported outbound protocols.
outbound_protocols: SmallVec<[TCodec::Protocol; 2]>,
/// The next (local) request ID.
next_request_id: RequestId,
Restore `RequestResponse::throttled`. (#1726) * Restore `RequestResponse::throttled`. In contrast to the existing "throttled" approach this PR adds back- pressure to the protocol without requiring pre-existing knowledge of all nodes about their limits. It adds small, CBOR-encoded headers to the actual payload data. Extra credit messages communicate back to the sender how many more requests it is allowed to send. * Remove some noise. * Resend credit grant after connection closed. Should an error in some lower layer cause a connection to be closed, our previously sent credit grant may not have reached the remote peer. Therefore, pessimistically, a credit grant is resent whenever a connection is closed. The remote ignores duplicate grants. * Remove inbound/outbound tracking per peer. * Send ACK as response to duplicate credit grants. * Simplify. * Fix grammar. * Incorporate review feedback. - Remove `ResponseSent` which was a leftover from previous attemps and issue a credit grant immediately in `send_response`. - Only resend credit grants after a connection is closed if we are still connected to this peer. * Move codec/header.rs to throttled/codec.rs. * More review suggestions. * Generalise `ProtocolWrapper` and use shorter prefix. * Update protocols/request-response/src/lib.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Minor comment changes. * Limit max. header size to 8KiB * Always construct initial limit with 1. Since honest senders always assume a send budget of 1 and wait for credit afterwards, setting the default limit to a higher value can only become effective after informing the peer about it which means leaving `max_recv` at 1 and setting `next_max` to the desired value. Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
2020-09-07 17:22:40 +02:00
/// The next (inbound) request ID.
next_inbound_id: Arc<AtomicU64>,
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
/// The protocol configuration.
config: RequestResponseConfig,
/// The protocol codec for reading and writing requests and responses.
codec: TCodec,
/// Pending events to return from `poll`.
pending_events: VecDeque<
NetworkBehaviourAction<
RequestProtocol<TCodec>,
RequestResponseEvent<TCodec::Request, TCodec::Response>>>,
/// The currently connected peers and their known, reachable addresses, if any.
connected: HashMap<PeerId, SmallVec<[Connection; 2]>>,
/// Externally managed addresses via `add_address` and `remove_address`.
addresses: HashMap<PeerId, SmallVec<[Multiaddr; 6]>>,
/// Requests that have not yet been sent and are waiting for a connection
/// to be established.
pending_requests: HashMap<PeerId, SmallVec<[RequestProtocol<TCodec>; 10]>>,
/// Responses that have not yet been received.
Restore `RequestResponse::throttled`. (#1726) * Restore `RequestResponse::throttled`. In contrast to the existing "throttled" approach this PR adds back- pressure to the protocol without requiring pre-existing knowledge of all nodes about their limits. It adds small, CBOR-encoded headers to the actual payload data. Extra credit messages communicate back to the sender how many more requests it is allowed to send. * Remove some noise. * Resend credit grant after connection closed. Should an error in some lower layer cause a connection to be closed, our previously sent credit grant may not have reached the remote peer. Therefore, pessimistically, a credit grant is resent whenever a connection is closed. The remote ignores duplicate grants. * Remove inbound/outbound tracking per peer. * Send ACK as response to duplicate credit grants. * Simplify. * Fix grammar. * Incorporate review feedback. - Remove `ResponseSent` which was a leftover from previous attemps and issue a credit grant immediately in `send_response`. - Only resend credit grants after a connection is closed if we are still connected to this peer. * Move codec/header.rs to throttled/codec.rs. * More review suggestions. * Generalise `ProtocolWrapper` and use shorter prefix. * Update protocols/request-response/src/lib.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Minor comment changes. * Limit max. header size to 8KiB * Always construct initial limit with 1. Since honest senders always assume a send budget of 1 and wait for credit afterwards, setting the default limit to a higher value can only become effective after informing the peer about it which means leaving `max_recv` at 1 and setting `next_max` to the desired value. Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
2020-09-07 17:22:40 +02:00
pending_responses: HashMap<RequestId, (PeerId, ConnectionId)>
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
}
impl<TCodec> RequestResponse<TCodec>
where
TCodec: RequestResponseCodec + Clone,
{
/// Creates a new `RequestResponse` behaviour for the given
/// protocols, codec and configuration.
pub fn new<I>(codec: TCodec, protocols: I, cfg: RequestResponseConfig) -> Self
where
I: IntoIterator<Item = (TCodec::Protocol, ProtocolSupport)>
{
let mut inbound_protocols = SmallVec::new();
let mut outbound_protocols = SmallVec::new();
for (p, s) in protocols {
if s.inbound() {
inbound_protocols.push(p.clone());
}
if s.outbound() {
outbound_protocols.push(p.clone());
}
}
RequestResponse {
inbound_protocols,
outbound_protocols,
next_request_id: RequestId(1),
Restore `RequestResponse::throttled`. (#1726) * Restore `RequestResponse::throttled`. In contrast to the existing "throttled" approach this PR adds back- pressure to the protocol without requiring pre-existing knowledge of all nodes about their limits. It adds small, CBOR-encoded headers to the actual payload data. Extra credit messages communicate back to the sender how many more requests it is allowed to send. * Remove some noise. * Resend credit grant after connection closed. Should an error in some lower layer cause a connection to be closed, our previously sent credit grant may not have reached the remote peer. Therefore, pessimistically, a credit grant is resent whenever a connection is closed. The remote ignores duplicate grants. * Remove inbound/outbound tracking per peer. * Send ACK as response to duplicate credit grants. * Simplify. * Fix grammar. * Incorporate review feedback. - Remove `ResponseSent` which was a leftover from previous attemps and issue a credit grant immediately in `send_response`. - Only resend credit grants after a connection is closed if we are still connected to this peer. * Move codec/header.rs to throttled/codec.rs. * More review suggestions. * Generalise `ProtocolWrapper` and use shorter prefix. * Update protocols/request-response/src/lib.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Minor comment changes. * Limit max. header size to 8KiB * Always construct initial limit with 1. Since honest senders always assume a send budget of 1 and wait for credit afterwards, setting the default limit to a higher value can only become effective after informing the peer about it which means leaving `max_recv` at 1 and setting `next_max` to the desired value. Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
2020-09-07 17:22:40 +02:00
next_inbound_id: Arc::new(AtomicU64::new(1)),
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
config: cfg,
codec,
pending_events: VecDeque::new(),
connected: HashMap::new(),
pending_requests: HashMap::new(),
pending_responses: HashMap::new(),
addresses: HashMap::new(),
}
}
Restore `RequestResponse::throttled`. (#1726) * Restore `RequestResponse::throttled`. In contrast to the existing "throttled" approach this PR adds back- pressure to the protocol without requiring pre-existing knowledge of all nodes about their limits. It adds small, CBOR-encoded headers to the actual payload data. Extra credit messages communicate back to the sender how many more requests it is allowed to send. * Remove some noise. * Resend credit grant after connection closed. Should an error in some lower layer cause a connection to be closed, our previously sent credit grant may not have reached the remote peer. Therefore, pessimistically, a credit grant is resent whenever a connection is closed. The remote ignores duplicate grants. * Remove inbound/outbound tracking per peer. * Send ACK as response to duplicate credit grants. * Simplify. * Fix grammar. * Incorporate review feedback. - Remove `ResponseSent` which was a leftover from previous attemps and issue a credit grant immediately in `send_response`. - Only resend credit grants after a connection is closed if we are still connected to this peer. * Move codec/header.rs to throttled/codec.rs. * More review suggestions. * Generalise `ProtocolWrapper` and use shorter prefix. * Update protocols/request-response/src/lib.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Minor comment changes. * Limit max. header size to 8KiB * Always construct initial limit with 1. Since honest senders always assume a send budget of 1 and wait for credit afterwards, setting the default limit to a higher value can only become effective after informing the peer about it which means leaving `max_recv` at 1 and setting `next_max` to the desired value. Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
2020-09-07 17:22:40 +02:00
/// Creates a `RequestResponse` which limits requests per peer.
///
/// The behaviour is wrapped in [`Throttled`] and detects the limits
/// per peer at runtime which are then enforced.
pub fn throttled<I>(c: TCodec, protos: I, cfg: RequestResponseConfig) -> Throttled<TCodec>
where
I: IntoIterator<Item = (TCodec::Protocol, ProtocolSupport)>,
TCodec: Send,
TCodec::Protocol: Sync
{
Throttled::new(c, protos, cfg)
}
Add `Throttled` to libp2p-request-response. (#1696) * Use a single exchange instead of two one_shots. * Add `Throttled` to libp2p-request-response. Wraps the existing `RequestResponse` behaviour and applies strict limits to the number of inbound and outbound requests per peer. The wrapper is opt-in and if not used, the protocol behaviour of `RequestResponse` does not change. This PR also does not introduce an extra protocol, hence the limits applied need to be known a priori for all nodes which is not always possible or desirable. As mentioned in #1687 I think that we should eventually augment the protocol with metadata which allows a more dynamic exchange of requests and responses. This PR also replaces the two oneshot channels with a single one from the scambio crate which saves one allocation per request/response. If not desirable because the crate has seen less testing the first commit could be reverted. * Fix rustdoc error. * Remove some leftovers from development. * Add docs to `NetworBehaviourAction::{map_in,map_out}`. * Apply suggestions from code review Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Add `ping_protocol_throttled` test. * Add another test. * Revert "Use a single exchange instead of two one_shots." This reverts commit e34e1297d411298f6c69e238aa6c96e0b795d989. # Conflicts: # protocols/request-response/Cargo.toml # protocols/request-response/src/handler/protocol.rs * Update CHANGELOG. * Update CHANGELOG. Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
2020-08-12 16:04:54 +02:00
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
/// Initiates sending a request.
///
/// If the targeted peer is currently not connected, a dialing
/// attempt is initiated and the request is sent as soon as a
/// connection is established.
///
/// > **Note**: In order for such a dialing attempt to succeed,
/// > the `RequestResonse` protocol must either be embedded
/// > in another `NetworkBehaviour` that provides peer and
/// > address discovery, or known addresses of peers must be
/// > managed via [`RequestResponse::add_address`] and
/// > [`RequestResponse::remove_address`].
pub fn send_request(&mut self, peer: &PeerId, request: TCodec::Request) -> RequestId {
let request_id = self.next_request_id();
let request = RequestProtocol {
request_id,
codec: self.codec.clone(),
protocols: self.outbound_protocols.clone(),
request,
};
if let Some(request) = self.try_send_request(peer, request) {
self.pending_events.push_back(NetworkBehaviourAction::DialPeer {
peer_id: peer.clone(),
condition: DialPeerCondition::Disconnected,
});
self.pending_requests.entry(peer.clone()).or_default().push(request);
}
request_id
}
/// Initiates sending a response to an inbound request.
///
[request-response] Refine success & error reporting for inbound requests. (#1867) * Refine error reporting for inbound request handling. At the moment one can neither get confirmation when a response has been sent on the underlying transport, nor is one aware of response omissions. The latter was originally intended as a feature for support of one-way protocols, which seems like a bad idea in hindsight. The lack of notification for sent responses may prohibit implementation of some request-response protocols that need to ensure a happens-before relation between sending a response and a subsequent request, besides uses for collecting statistics. Even with these changes, there is no active notification for failed inbound requests as a result of connections unexpectedly closing, as is the case for outbound requests. Instead, for pending inbound requests this scenario can be identified if necessary by the absense of both `InboundFailure` and `ResponseSent` events for a particular previously received request. Interest in this situation is not expected to be common and would otherwise require explicitly tracking all inbound requests in the `RequestResponse` behaviour, which would be a pity. `RequestResponse::send_response` now also synchronously returns an error if the inbound upgrade handling the request has been aborted, due to timeout or closing of the connection, giving more options for graceful error handling for inbound requests. As an aside, the `Throttled` wrapper now no longer emits inbound or outbound error events occurring in the context of sending credit requests or responses. This is in addition to not emitting `ResponseSent` events for ACK responses of credit grants. * Update protocols/request-response/src/lib.rs Co-authored-by: Max Inden <mail@max-inden.de> * Address some minor clippy warnings. (#1868) * Track pending credit request IDs. In order to avoid emitting events relating to credit grants or acks on the public API. The public API should only emit events relating to the actual requests and responses sent by client code. * Small cleanup * Cleanup * Update versions and changelogs. * Unreleased Co-authored-by: Max Inden <mail@max-inden.de>
2020-12-07 13:07:47 +01:00
/// If the `ResponseChannel` is already closed due to a timeout or
/// the connection being closed, the response is returned as an `Err`
/// for further handling. Once the response has been successfully sent
/// on the corresponding connection, [`RequestResponseEvent::ResponseSent`]
/// is emitted.
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
///
[request-response] Refine success & error reporting for inbound requests. (#1867) * Refine error reporting for inbound request handling. At the moment one can neither get confirmation when a response has been sent on the underlying transport, nor is one aware of response omissions. The latter was originally intended as a feature for support of one-way protocols, which seems like a bad idea in hindsight. The lack of notification for sent responses may prohibit implementation of some request-response protocols that need to ensure a happens-before relation between sending a response and a subsequent request, besides uses for collecting statistics. Even with these changes, there is no active notification for failed inbound requests as a result of connections unexpectedly closing, as is the case for outbound requests. Instead, for pending inbound requests this scenario can be identified if necessary by the absense of both `InboundFailure` and `ResponseSent` events for a particular previously received request. Interest in this situation is not expected to be common and would otherwise require explicitly tracking all inbound requests in the `RequestResponse` behaviour, which would be a pity. `RequestResponse::send_response` now also synchronously returns an error if the inbound upgrade handling the request has been aborted, due to timeout or closing of the connection, giving more options for graceful error handling for inbound requests. As an aside, the `Throttled` wrapper now no longer emits inbound or outbound error events occurring in the context of sending credit requests or responses. This is in addition to not emitting `ResponseSent` events for ACK responses of credit grants. * Update protocols/request-response/src/lib.rs Co-authored-by: Max Inden <mail@max-inden.de> * Address some minor clippy warnings. (#1868) * Track pending credit request IDs. In order to avoid emitting events relating to credit grants or acks on the public API. The public API should only emit events relating to the actual requests and responses sent by client code. * Small cleanup * Cleanup * Update versions and changelogs. * Unreleased Co-authored-by: Max Inden <mail@max-inden.de>
2020-12-07 13:07:47 +01:00
/// The provided `ResponseChannel` is obtained from an inbound
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
/// [`RequestResponseMessage::Request`].
[request-response] Refine success & error reporting for inbound requests. (#1867) * Refine error reporting for inbound request handling. At the moment one can neither get confirmation when a response has been sent on the underlying transport, nor is one aware of response omissions. The latter was originally intended as a feature for support of one-way protocols, which seems like a bad idea in hindsight. The lack of notification for sent responses may prohibit implementation of some request-response protocols that need to ensure a happens-before relation between sending a response and a subsequent request, besides uses for collecting statistics. Even with these changes, there is no active notification for failed inbound requests as a result of connections unexpectedly closing, as is the case for outbound requests. Instead, for pending inbound requests this scenario can be identified if necessary by the absense of both `InboundFailure` and `ResponseSent` events for a particular previously received request. Interest in this situation is not expected to be common and would otherwise require explicitly tracking all inbound requests in the `RequestResponse` behaviour, which would be a pity. `RequestResponse::send_response` now also synchronously returns an error if the inbound upgrade handling the request has been aborted, due to timeout or closing of the connection, giving more options for graceful error handling for inbound requests. As an aside, the `Throttled` wrapper now no longer emits inbound or outbound error events occurring in the context of sending credit requests or responses. This is in addition to not emitting `ResponseSent` events for ACK responses of credit grants. * Update protocols/request-response/src/lib.rs Co-authored-by: Max Inden <mail@max-inden.de> * Address some minor clippy warnings. (#1868) * Track pending credit request IDs. In order to avoid emitting events relating to credit grants or acks on the public API. The public API should only emit events relating to the actual requests and responses sent by client code. * Small cleanup * Cleanup * Update versions and changelogs. * Unreleased Co-authored-by: Max Inden <mail@max-inden.de>
2020-12-07 13:07:47 +01:00
pub fn send_response(&mut self, ch: ResponseChannel<TCodec::Response>, rs: TCodec::Response)
-> Result<(), TCodec::Response>
{
ch.sender.send(rs)
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
}
/// Adds a known address for a peer that can be used for
/// dialing attempts by the `Swarm`, i.e. is returned
/// by [`NetworkBehaviour::addresses_of_peer`].
///
/// Addresses added in this way are only removed by `remove_address`.
pub fn add_address(&mut self, peer: &PeerId, address: Multiaddr) {
self.addresses.entry(peer.clone()).or_default().push(address);
}
/// Removes an address of a peer previously added via `add_address`.
pub fn remove_address(&mut self, peer: &PeerId, address: &Multiaddr) {
let mut last = false;
if let Some(addresses) = self.addresses.get_mut(peer) {
addresses.retain(|a| a != address);
last = addresses.is_empty();
}
if last {
self.addresses.remove(peer);
}
}
/// Checks whether a peer is currently connected.
pub fn is_connected(&self, peer: &PeerId) -> bool {
Restore `RequestResponse::throttled`. (#1726) * Restore `RequestResponse::throttled`. In contrast to the existing "throttled" approach this PR adds back- pressure to the protocol without requiring pre-existing knowledge of all nodes about their limits. It adds small, CBOR-encoded headers to the actual payload data. Extra credit messages communicate back to the sender how many more requests it is allowed to send. * Remove some noise. * Resend credit grant after connection closed. Should an error in some lower layer cause a connection to be closed, our previously sent credit grant may not have reached the remote peer. Therefore, pessimistically, a credit grant is resent whenever a connection is closed. The remote ignores duplicate grants. * Remove inbound/outbound tracking per peer. * Send ACK as response to duplicate credit grants. * Simplify. * Fix grammar. * Incorporate review feedback. - Remove `ResponseSent` which was a leftover from previous attemps and issue a credit grant immediately in `send_response`. - Only resend credit grants after a connection is closed if we are still connected to this peer. * Move codec/header.rs to throttled/codec.rs. * More review suggestions. * Generalise `ProtocolWrapper` and use shorter prefix. * Update protocols/request-response/src/lib.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Minor comment changes. * Limit max. header size to 8KiB * Always construct initial limit with 1. Since honest senders always assume a send budget of 1 and wait for credit afterwards, setting the default limit to a higher value can only become effective after informing the peer about it which means leaving `max_recv` at 1 and setting `next_max` to the desired value. Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
2020-09-07 17:22:40 +02:00
if let Some(connections) = self.connected.get(peer) {
!connections.is_empty()
} else {
false
}
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
}
/// Checks whether an outbound request initiated by
/// [`RequestResponse::send_request`] is still pending, i.e. waiting
/// for a response.
Restore `RequestResponse::throttled`. (#1726) * Restore `RequestResponse::throttled`. In contrast to the existing "throttled" approach this PR adds back- pressure to the protocol without requiring pre-existing knowledge of all nodes about their limits. It adds small, CBOR-encoded headers to the actual payload data. Extra credit messages communicate back to the sender how many more requests it is allowed to send. * Remove some noise. * Resend credit grant after connection closed. Should an error in some lower layer cause a connection to be closed, our previously sent credit grant may not have reached the remote peer. Therefore, pessimistically, a credit grant is resent whenever a connection is closed. The remote ignores duplicate grants. * Remove inbound/outbound tracking per peer. * Send ACK as response to duplicate credit grants. * Simplify. * Fix grammar. * Incorporate review feedback. - Remove `ResponseSent` which was a leftover from previous attemps and issue a credit grant immediately in `send_response`. - Only resend credit grants after a connection is closed if we are still connected to this peer. * Move codec/header.rs to throttled/codec.rs. * More review suggestions. * Generalise `ProtocolWrapper` and use shorter prefix. * Update protocols/request-response/src/lib.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Minor comment changes. * Limit max. header size to 8KiB * Always construct initial limit with 1. Since honest senders always assume a send budget of 1 and wait for credit afterwards, setting the default limit to a higher value can only become effective after informing the peer about it which means leaving `max_recv` at 1 and setting `next_max` to the desired value. Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
2020-09-07 17:22:40 +02:00
pub fn is_pending_outbound(&self, req_id: &RequestId) -> bool {
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
self.pending_responses.contains_key(req_id)
}
/// Returns the next request ID.
fn next_request_id(&mut self) -> RequestId {
let request_id = self.next_request_id;
self.next_request_id.0 += 1;
request_id
}
/// Tries to send a request by queueing an appropriate event to be
/// emitted to the `Swarm`. If the peer is not currently connected,
/// the given request is return unchanged.
fn try_send_request(&mut self, peer: &PeerId, request: RequestProtocol<TCodec>)
-> Option<RequestProtocol<TCodec>>
{
if let Some(connections) = self.connected.get(peer) {
Restore `RequestResponse::throttled`. (#1726) * Restore `RequestResponse::throttled`. In contrast to the existing "throttled" approach this PR adds back- pressure to the protocol without requiring pre-existing knowledge of all nodes about their limits. It adds small, CBOR-encoded headers to the actual payload data. Extra credit messages communicate back to the sender how many more requests it is allowed to send. * Remove some noise. * Resend credit grant after connection closed. Should an error in some lower layer cause a connection to be closed, our previously sent credit grant may not have reached the remote peer. Therefore, pessimistically, a credit grant is resent whenever a connection is closed. The remote ignores duplicate grants. * Remove inbound/outbound tracking per peer. * Send ACK as response to duplicate credit grants. * Simplify. * Fix grammar. * Incorporate review feedback. - Remove `ResponseSent` which was a leftover from previous attemps and issue a credit grant immediately in `send_response`. - Only resend credit grants after a connection is closed if we are still connected to this peer. * Move codec/header.rs to throttled/codec.rs. * More review suggestions. * Generalise `ProtocolWrapper` and use shorter prefix. * Update protocols/request-response/src/lib.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Minor comment changes. * Limit max. header size to 8KiB * Always construct initial limit with 1. Since honest senders always assume a send budget of 1 and wait for credit afterwards, setting the default limit to a higher value can only become effective after informing the peer about it which means leaving `max_recv` at 1 and setting `next_max` to the desired value. Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
2020-09-07 17:22:40 +02:00
if connections.is_empty() {
return Some(request)
}
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
let ix = (request.request_id.0 as usize) % connections.len();
let conn = connections[ix].id;
self.pending_responses.insert(request.request_id, (peer.clone(), conn));
self.pending_events.push_back(NetworkBehaviourAction::NotifyHandler {
peer_id: peer.clone(),
handler: NotifyHandler::One(conn),
event: request
});
None
} else {
Some(request)
}
}
}
impl<TCodec> NetworkBehaviour for RequestResponse<TCodec>
where
TCodec: RequestResponseCodec + Send + Clone + 'static,
{
type ProtocolsHandler = RequestResponseHandler<TCodec>;
type OutEvent = RequestResponseEvent<TCodec::Request, TCodec::Response>;
fn new_handler(&mut self) -> Self::ProtocolsHandler {
RequestResponseHandler::new(
self.inbound_protocols.clone(),
self.codec.clone(),
self.config.connection_keep_alive,
self.config.request_timeout,
Restore `RequestResponse::throttled`. (#1726) * Restore `RequestResponse::throttled`. In contrast to the existing "throttled" approach this PR adds back- pressure to the protocol without requiring pre-existing knowledge of all nodes about their limits. It adds small, CBOR-encoded headers to the actual payload data. Extra credit messages communicate back to the sender how many more requests it is allowed to send. * Remove some noise. * Resend credit grant after connection closed. Should an error in some lower layer cause a connection to be closed, our previously sent credit grant may not have reached the remote peer. Therefore, pessimistically, a credit grant is resent whenever a connection is closed. The remote ignores duplicate grants. * Remove inbound/outbound tracking per peer. * Send ACK as response to duplicate credit grants. * Simplify. * Fix grammar. * Incorporate review feedback. - Remove `ResponseSent` which was a leftover from previous attemps and issue a credit grant immediately in `send_response`. - Only resend credit grants after a connection is closed if we are still connected to this peer. * Move codec/header.rs to throttled/codec.rs. * More review suggestions. * Generalise `ProtocolWrapper` and use shorter prefix. * Update protocols/request-response/src/lib.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Minor comment changes. * Limit max. header size to 8KiB * Always construct initial limit with 1. Since honest senders always assume a send budget of 1 and wait for credit afterwards, setting the default limit to a higher value can only become effective after informing the peer about it which means leaving `max_recv` at 1 and setting `next_max` to the desired value. Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
2020-09-07 17:22:40 +02:00
self.next_inbound_id.clone()
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
)
}
fn addresses_of_peer(&mut self, peer: &PeerId) -> Vec<Multiaddr> {
let mut addresses = Vec::new();
if let Some(connections) = self.connected.get(peer) {
addresses.extend(connections.iter().filter_map(|c| c.address.clone()))
}
if let Some(more) = self.addresses.get(peer) {
addresses.extend(more.into_iter().cloned());
}
addresses
}
fn inject_connected(&mut self, peer: &PeerId) {
if let Some(pending) = self.pending_requests.remove(peer) {
for request in pending {
let request = self.try_send_request(peer, request);
assert!(request.is_none());
}
}
}
fn inject_connection_established(&mut self, peer: &PeerId, conn: &ConnectionId, endpoint: &ConnectedPoint) {
let address = match endpoint {
ConnectedPoint::Dialer { address } => Some(address.clone()),
ConnectedPoint::Listener { .. } => None
};
let connections = self.connected.entry(peer.clone()).or_default();
connections.push(Connection { id: *conn, address })
}
fn inject_connection_closed(&mut self, peer: &PeerId, conn: &ConnectionId, _: &ConnectedPoint) {
if let Some(connections) = self.connected.get_mut(peer) {
if let Some(pos) = connections.iter().position(|c| &c.id == conn) {
connections.remove(pos);
}
}
Restore `RequestResponse::throttled`. (#1726) * Restore `RequestResponse::throttled`. In contrast to the existing "throttled" approach this PR adds back- pressure to the protocol without requiring pre-existing knowledge of all nodes about their limits. It adds small, CBOR-encoded headers to the actual payload data. Extra credit messages communicate back to the sender how many more requests it is allowed to send. * Remove some noise. * Resend credit grant after connection closed. Should an error in some lower layer cause a connection to be closed, our previously sent credit grant may not have reached the remote peer. Therefore, pessimistically, a credit grant is resent whenever a connection is closed. The remote ignores duplicate grants. * Remove inbound/outbound tracking per peer. * Send ACK as response to duplicate credit grants. * Simplify. * Fix grammar. * Incorporate review feedback. - Remove `ResponseSent` which was a leftover from previous attemps and issue a credit grant immediately in `send_response`. - Only resend credit grants after a connection is closed if we are still connected to this peer. * Move codec/header.rs to throttled/codec.rs. * More review suggestions. * Generalise `ProtocolWrapper` and use shorter prefix. * Update protocols/request-response/src/lib.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Minor comment changes. * Limit max. header size to 8KiB * Always construct initial limit with 1. Since honest senders always assume a send budget of 1 and wait for credit afterwards, setting the default limit to a higher value can only become effective after informing the peer about it which means leaving `max_recv` at 1 and setting `next_max` to the desired value. Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
2020-09-07 17:22:40 +02:00
let pending_events = &mut self.pending_events;
// Any pending responses of requests sent over this connection must be considered failed.
self.pending_responses.retain(|rid, (peer, cid)| {
if conn != cid {
return true
}
pending_events.push_back(NetworkBehaviourAction::GenerateEvent(
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
RequestResponseEvent::OutboundFailure {
Restore `RequestResponse::throttled`. (#1726) * Restore `RequestResponse::throttled`. In contrast to the existing "throttled" approach this PR adds back- pressure to the protocol without requiring pre-existing knowledge of all nodes about their limits. It adds small, CBOR-encoded headers to the actual payload data. Extra credit messages communicate back to the sender how many more requests it is allowed to send. * Remove some noise. * Resend credit grant after connection closed. Should an error in some lower layer cause a connection to be closed, our previously sent credit grant may not have reached the remote peer. Therefore, pessimistically, a credit grant is resent whenever a connection is closed. The remote ignores duplicate grants. * Remove inbound/outbound tracking per peer. * Send ACK as response to duplicate credit grants. * Simplify. * Fix grammar. * Incorporate review feedback. - Remove `ResponseSent` which was a leftover from previous attemps and issue a credit grant immediately in `send_response`. - Only resend credit grants after a connection is closed if we are still connected to this peer. * Move codec/header.rs to throttled/codec.rs. * More review suggestions. * Generalise `ProtocolWrapper` and use shorter prefix. * Update protocols/request-response/src/lib.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Minor comment changes. * Limit max. header size to 8KiB * Always construct initial limit with 1. Since honest senders always assume a send budget of 1 and wait for credit afterwards, setting the default limit to a higher value can only become effective after informing the peer about it which means leaving `max_recv` at 1 and setting `next_max` to the desired value. Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
2020-09-07 17:22:40 +02:00
peer: peer.clone(),
request_id: *rid,
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
error: OutboundFailure::ConnectionClosed
}
));
Restore `RequestResponse::throttled`. (#1726) * Restore `RequestResponse::throttled`. In contrast to the existing "throttled" approach this PR adds back- pressure to the protocol without requiring pre-existing knowledge of all nodes about their limits. It adds small, CBOR-encoded headers to the actual payload data. Extra credit messages communicate back to the sender how many more requests it is allowed to send. * Remove some noise. * Resend credit grant after connection closed. Should an error in some lower layer cause a connection to be closed, our previously sent credit grant may not have reached the remote peer. Therefore, pessimistically, a credit grant is resent whenever a connection is closed. The remote ignores duplicate grants. * Remove inbound/outbound tracking per peer. * Send ACK as response to duplicate credit grants. * Simplify. * Fix grammar. * Incorporate review feedback. - Remove `ResponseSent` which was a leftover from previous attemps and issue a credit grant immediately in `send_response`. - Only resend credit grants after a connection is closed if we are still connected to this peer. * Move codec/header.rs to throttled/codec.rs. * More review suggestions. * Generalise `ProtocolWrapper` and use shorter prefix. * Update protocols/request-response/src/lib.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Minor comment changes. * Limit max. header size to 8KiB * Always construct initial limit with 1. Since honest senders always assume a send budget of 1 and wait for credit afterwards, setting the default limit to a higher value can only become effective after informing the peer about it which means leaving `max_recv` at 1 and setting `next_max` to the desired value. Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
2020-09-07 17:22:40 +02:00
false
});
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
}
fn inject_disconnected(&mut self, peer: &PeerId) {
self.connected.remove(peer);
}
fn inject_dial_failure(&mut self, peer: &PeerId) {
// If there are pending outgoing requests when a dial failure occurs,
// it is implied that we are not connected to the peer, since pending
// outgoing requests are drained when a connection is established and
// only created when a peer is not connected when a request is made.
// Thus these requests must be considered failed, even if there is
// another, concurrent dialing attempt ongoing.
if let Some(pending) = self.pending_requests.remove(peer) {
for request in pending {
self.pending_events.push_back(NetworkBehaviourAction::GenerateEvent(
RequestResponseEvent::OutboundFailure {
peer: peer.clone(),
request_id: request.request_id,
error: OutboundFailure::DialFailure
}
));
}
}
}
fn inject_event(
&mut self,
peer: PeerId,
_: ConnectionId,
event: RequestResponseHandlerEvent<TCodec>,
) {
match event {
RequestResponseHandlerEvent::Response { request_id, response } => {
self.pending_responses.remove(&request_id);
let message = RequestResponseMessage::Response { request_id, response };
self.pending_events.push_back(
NetworkBehaviourAction::GenerateEvent(
RequestResponseEvent::Message { peer, message }));
}
Restore `RequestResponse::throttled`. (#1726) * Restore `RequestResponse::throttled`. In contrast to the existing "throttled" approach this PR adds back- pressure to the protocol without requiring pre-existing knowledge of all nodes about their limits. It adds small, CBOR-encoded headers to the actual payload data. Extra credit messages communicate back to the sender how many more requests it is allowed to send. * Remove some noise. * Resend credit grant after connection closed. Should an error in some lower layer cause a connection to be closed, our previously sent credit grant may not have reached the remote peer. Therefore, pessimistically, a credit grant is resent whenever a connection is closed. The remote ignores duplicate grants. * Remove inbound/outbound tracking per peer. * Send ACK as response to duplicate credit grants. * Simplify. * Fix grammar. * Incorporate review feedback. - Remove `ResponseSent` which was a leftover from previous attemps and issue a credit grant immediately in `send_response`. - Only resend credit grants after a connection is closed if we are still connected to this peer. * Move codec/header.rs to throttled/codec.rs. * More review suggestions. * Generalise `ProtocolWrapper` and use shorter prefix. * Update protocols/request-response/src/lib.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Minor comment changes. * Limit max. header size to 8KiB * Always construct initial limit with 1. Since honest senders always assume a send budget of 1 and wait for credit afterwards, setting the default limit to a higher value can only become effective after informing the peer about it which means leaving `max_recv` at 1 and setting `next_max` to the desired value. Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
2020-09-07 17:22:40 +02:00
RequestResponseHandlerEvent::Request { request_id, request, sender } => {
let channel = ResponseChannel { request_id, peer: peer.clone(), sender };
let message = RequestResponseMessage::Request { request_id, request, channel };
self.pending_events.push_back(NetworkBehaviourAction::GenerateEvent(
RequestResponseEvent::Message { peer, message }
));
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
}
[request-response] Refine success & error reporting for inbound requests. (#1867) * Refine error reporting for inbound request handling. At the moment one can neither get confirmation when a response has been sent on the underlying transport, nor is one aware of response omissions. The latter was originally intended as a feature for support of one-way protocols, which seems like a bad idea in hindsight. The lack of notification for sent responses may prohibit implementation of some request-response protocols that need to ensure a happens-before relation between sending a response and a subsequent request, besides uses for collecting statistics. Even with these changes, there is no active notification for failed inbound requests as a result of connections unexpectedly closing, as is the case for outbound requests. Instead, for pending inbound requests this scenario can be identified if necessary by the absense of both `InboundFailure` and `ResponseSent` events for a particular previously received request. Interest in this situation is not expected to be common and would otherwise require explicitly tracking all inbound requests in the `RequestResponse` behaviour, which would be a pity. `RequestResponse::send_response` now also synchronously returns an error if the inbound upgrade handling the request has been aborted, due to timeout or closing of the connection, giving more options for graceful error handling for inbound requests. As an aside, the `Throttled` wrapper now no longer emits inbound or outbound error events occurring in the context of sending credit requests or responses. This is in addition to not emitting `ResponseSent` events for ACK responses of credit grants. * Update protocols/request-response/src/lib.rs Co-authored-by: Max Inden <mail@max-inden.de> * Address some minor clippy warnings. (#1868) * Track pending credit request IDs. In order to avoid emitting events relating to credit grants or acks on the public API. The public API should only emit events relating to the actual requests and responses sent by client code. * Small cleanup * Cleanup * Update versions and changelogs. * Unreleased Co-authored-by: Max Inden <mail@max-inden.de>
2020-12-07 13:07:47 +01:00
RequestResponseHandlerEvent::ResponseSent(request_id) => {
self.pending_events.push_back(
NetworkBehaviourAction::GenerateEvent(
RequestResponseEvent::ResponseSent { peer, request_id }));
}
RequestResponseHandlerEvent::ResponseOmission(request_id) => {
self.pending_events.push_back(
NetworkBehaviourAction::GenerateEvent(
RequestResponseEvent::InboundFailure {
peer,
request_id,
error: InboundFailure::ResponseOmission
}));
}
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
RequestResponseHandlerEvent::OutboundTimeout(request_id) => {
if let Some((peer, _conn)) = self.pending_responses.remove(&request_id) {
self.pending_events.push_back(
NetworkBehaviourAction::GenerateEvent(
RequestResponseEvent::OutboundFailure {
peer,
request_id,
error: OutboundFailure::Timeout,
}));
}
}
Restore `RequestResponse::throttled`. (#1726) * Restore `RequestResponse::throttled`. In contrast to the existing "throttled" approach this PR adds back- pressure to the protocol without requiring pre-existing knowledge of all nodes about their limits. It adds small, CBOR-encoded headers to the actual payload data. Extra credit messages communicate back to the sender how many more requests it is allowed to send. * Remove some noise. * Resend credit grant after connection closed. Should an error in some lower layer cause a connection to be closed, our previously sent credit grant may not have reached the remote peer. Therefore, pessimistically, a credit grant is resent whenever a connection is closed. The remote ignores duplicate grants. * Remove inbound/outbound tracking per peer. * Send ACK as response to duplicate credit grants. * Simplify. * Fix grammar. * Incorporate review feedback. - Remove `ResponseSent` which was a leftover from previous attemps and issue a credit grant immediately in `send_response`. - Only resend credit grants after a connection is closed if we are still connected to this peer. * Move codec/header.rs to throttled/codec.rs. * More review suggestions. * Generalise `ProtocolWrapper` and use shorter prefix. * Update protocols/request-response/src/lib.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Minor comment changes. * Limit max. header size to 8KiB * Always construct initial limit with 1. Since honest senders always assume a send budget of 1 and wait for credit afterwards, setting the default limit to a higher value can only become effective after informing the peer about it which means leaving `max_recv` at 1 and setting `next_max` to the desired value. Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
2020-09-07 17:22:40 +02:00
RequestResponseHandlerEvent::InboundTimeout(request_id) => {
self.pending_events.push_back(
NetworkBehaviourAction::GenerateEvent(
RequestResponseEvent::InboundFailure {
peer,
request_id,
error: InboundFailure::Timeout,
}));
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
}
RequestResponseHandlerEvent::OutboundUnsupportedProtocols(request_id) => {
self.pending_events.push_back(
NetworkBehaviourAction::GenerateEvent(
RequestResponseEvent::OutboundFailure {
peer,
request_id,
error: OutboundFailure::UnsupportedProtocols,
}));
}
Restore `RequestResponse::throttled`. (#1726) * Restore `RequestResponse::throttled`. In contrast to the existing "throttled" approach this PR adds back- pressure to the protocol without requiring pre-existing knowledge of all nodes about their limits. It adds small, CBOR-encoded headers to the actual payload data. Extra credit messages communicate back to the sender how many more requests it is allowed to send. * Remove some noise. * Resend credit grant after connection closed. Should an error in some lower layer cause a connection to be closed, our previously sent credit grant may not have reached the remote peer. Therefore, pessimistically, a credit grant is resent whenever a connection is closed. The remote ignores duplicate grants. * Remove inbound/outbound tracking per peer. * Send ACK as response to duplicate credit grants. * Simplify. * Fix grammar. * Incorporate review feedback. - Remove `ResponseSent` which was a leftover from previous attemps and issue a credit grant immediately in `send_response`. - Only resend credit grants after a connection is closed if we are still connected to this peer. * Move codec/header.rs to throttled/codec.rs. * More review suggestions. * Generalise `ProtocolWrapper` and use shorter prefix. * Update protocols/request-response/src/lib.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Minor comment changes. * Limit max. header size to 8KiB * Always construct initial limit with 1. Since honest senders always assume a send budget of 1 and wait for credit afterwards, setting the default limit to a higher value can only become effective after informing the peer about it which means leaving `max_recv` at 1 and setting `next_max` to the desired value. Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
2020-09-07 17:22:40 +02:00
RequestResponseHandlerEvent::InboundUnsupportedProtocols(request_id) => {
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
self.pending_events.push_back(
NetworkBehaviourAction::GenerateEvent(
RequestResponseEvent::InboundFailure {
peer,
Restore `RequestResponse::throttled`. (#1726) * Restore `RequestResponse::throttled`. In contrast to the existing "throttled" approach this PR adds back- pressure to the protocol without requiring pre-existing knowledge of all nodes about their limits. It adds small, CBOR-encoded headers to the actual payload data. Extra credit messages communicate back to the sender how many more requests it is allowed to send. * Remove some noise. * Resend credit grant after connection closed. Should an error in some lower layer cause a connection to be closed, our previously sent credit grant may not have reached the remote peer. Therefore, pessimistically, a credit grant is resent whenever a connection is closed. The remote ignores duplicate grants. * Remove inbound/outbound tracking per peer. * Send ACK as response to duplicate credit grants. * Simplify. * Fix grammar. * Incorporate review feedback. - Remove `ResponseSent` which was a leftover from previous attemps and issue a credit grant immediately in `send_response`. - Only resend credit grants after a connection is closed if we are still connected to this peer. * Move codec/header.rs to throttled/codec.rs. * More review suggestions. * Generalise `ProtocolWrapper` and use shorter prefix. * Update protocols/request-response/src/lib.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Update protocols/request-response/src/throttled.rs Co-authored-by: Roman Borschel <romanb@users.noreply.github.com> * Minor comment changes. * Limit max. header size to 8KiB * Always construct initial limit with 1. Since honest senders always assume a send budget of 1 and wait for credit afterwards, setting the default limit to a higher value can only become effective after informing the peer about it which means leaving `max_recv` at 1 and setting `next_max` to the desired value. Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
2020-09-07 17:22:40 +02:00
request_id,
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
error: InboundFailure::UnsupportedProtocols,
}));
}
}
}
2020-07-27 20:27:33 +00:00
fn poll(&mut self, _: &mut Context<'_>, _: &mut impl PollParameters)
Add libp2p-request-response protocol. (#1596) * Add the libp2p-request-response protocol. This crate provides a generic implementation for request/response protocols, whereby each request is sent on a new substream. * Fix OneShotHandler usage in floodsub. * Custom ProtocolsHandler and multiple protocols. 1. Implement a custom ProtocolsHandler instead of using the OneShotHandler for better control and error handling. In particular, all request/response sending/receiving is kept in the substreams upgrades and thus the background task of a connection. 2. Support multiple protocols (usually protocol versions) with a single `RequestResponse` instance, with configurable inbound/outbound support. * Small doc clarification. * Remove unnecessary Sync bounds. * Remove redundant Clone constraint. * Update protocols/request-response/Cargo.toml Co-authored-by: Toralf Wittner <tw@dtex.org> * Update dev-dependencies. * Update Cargo.tomls. * Add changelog. * Remove Sync bound from RequestResponseCodec::Protocol. Apparently the compiler just needs some help with the scope of borrows, which is unfortunate. * Try async-trait. * Allow checking whether a ResponseChannel is still open. Also expand the commentary on `send_response` to indicate that responses may be discard if they come in too late. * Add `RequestResponse::is_pending`. As an analogue of `ResponseChannel::is_open` for outbound requests. * Revert now unnecessary changes to the OneShotHandler. Since `libp2p-request-response` is no longer using it. * Update CHANGELOG for libp2p-swarm. Co-authored-by: Toralf Wittner <tw@dtex.org>
2020-06-29 17:08:40 +02:00
-> Poll<NetworkBehaviourAction<
RequestProtocol<TCodec>,
RequestResponseEvent<TCodec::Request, TCodec::Response>
>>
{
if let Some(ev) = self.pending_events.pop_front() {
return Poll::Ready(ev);
} else if self.pending_events.capacity() > EMPTY_QUEUE_SHRINK_THRESHOLD {
self.pending_events.shrink_to_fit();
}
Poll::Pending
}
}
/// Internal threshold for when to shrink the capacity
/// of empty queues. If the capacity of an empty queue
/// exceeds this threshold, the associated memory is
/// released.
const EMPTY_QUEUE_SHRINK_THRESHOLD: usize = 100;
/// Internal information tracked for an established connection.
struct Connection {
id: ConnectionId,
address: Option<Multiaddr>,
}