feat(grpc): Add tonic transport #2339

arjan-bal · 2025-07-15T07:25:36Z

This PR includes the following:

A transport trait that will be used by gRPC subchannels. Transports are expected to transfer serialized messages as bytes, but an in-memory transport may also transfer structs without serialization.
The runtime trait is extended to include a method for creating TCP streams. Adapters are added to convert a gRPC runtime to a Hyper runtime.
A transport implementation that uses tonic. To avoid a dependency on the code in tonic/src/transport, required code is copied over.
A tonic codec that sends/receives Bytes. This is a temporary workaround until tonic supports bypassing the codec and receiving bytes.
A test that uses the grpc tonic transport to create a bi-di stream with a tonic server.

arjan-bal · 2025-07-15T14:18:18Z

Hi @dfawley and @LucioFranco, could you please review this PR when you have a moment?

easwars · 2025-07-21T17:22:54Z

grpc/src/client/subchannel.rs

@@ -229,6 +235,7 @@ impl InternalSubchannel {
        transport: Arc<dyn Transport>,
        backoff: Arc<dyn Backoff>,
        unregister_fn: Box<dyn FnOnce(SubchannelKey) + Send + Sync>,
+        runtime: Arc<dyn Runtime>,


Curious:

Given that we will have the same runtime for all the different gRPC components that require a runtime, did we consider something like a singleton that is initialized at init time, and all the components can use a getter to retrieve and use the singleton instead of the runtime being passed to every component that needs it?

Different grpc channels could theoretically use different runtimes. Maybe that isn't something we need to support, but it's pretty easily attained - it just requires passing around the runtime a bit more than if it were global.

C++ passes the event engine through channel args. In my opinion passing the runtime through a function param allows for cleaner dependency injection. It also enforces that the runtime is set during channel creation, before RPCs are made.

Having a singleton runtime will force all gRPC channels in a binary to use the same runtime. I don't know if this is a con though. We can discuss this in the team meeting.

easwars · 2025-07-21T17:24:35Z

grpc/src/client/subchannel.rs

@@ -345,30 +353,34 @@ impl InternalSubchannel {
        let transport = self.transport.clone();
        let address = self.address().address;
        let state_machine_tx = self.state_machine_event_sender.clone();
-        let connect_task = tokio::task::spawn(async move {


Do we have some kind of vet equivalent to ensure that task spawning (and other features provided by the runtime) are always only used from the runtime and not from other places (like tokio or the standard library)?

Ultimately (pre-1.0) we want to not have any tokio runtime crates/features listed in Cargo.toml, except if you are using a tokio feature flag. That would prevent such a thing.

I spent some time looking into this. I found two approaches:

Use clippy disallowed_method, disallowed_macros, etc. to block tokio symbols like tokio::spawn, tokio::task::spawn, etc. The problem with this approach is that we need to list all the types we want to block, there's not glob (*) operator available. It's also easy to miss the clippy warnings since they don't block PR submission.

Introduce a separate crate, say grpc-runtime-tokio, for the default runtime implementation, and disable tokio's runtime features in the main grpc crate. If a function in the grpc crate tries to call tokio::spawn, it will fail to compile as the required feature will be disabled. The concern with this approach is that we need to export the runtime trait (and related types) which are unstable.

@LucioFranco would like to get your thoughts on this.

easwars · 2025-07-21T17:27:01Z

grpc/src/client/transport/registry.rs

        self.m
            .lock()
            .unwrap()
            .insert(address_type.to_string(), Arc::new(transport));
    }
+
    /// Retrieve a name resolver from the registry, or None if not found.


Nit: this comments needs updating.

Fixed the comment.

easwars · 2025-07-21T17:28:52Z

grpc/src/client/transport/registry.rs

@@ -26,20 +25,20 @@ impl std::fmt::Debug for TransportRegistry {

 impl TransportRegistry {
    /// Construct an empty name resolver registry.
-    pub fn new() -> Self {
+    pub(crate) fn new() -> Self {


Should we also implement Default trait for this type by inheriting it?

There is an explicit Default implementation below. I removed it and used the derive macro instead.

easwars · 2025-07-21T17:32:48Z

grpc/src/client/subchannel.rs

@@ -388,7 +400,9 @@ impl InternalSubchannel {
            // error string containing information about why the connection
            // terminated? But what can we do with that error other than logging
            // it, which the transport can do as well?
-            svc.disconnected().await;
+            if let Err(e) = closed_rx.await {


Should the above task spawn be on the runtime as well instead of directly using tokio?

Yes, used the runtime here.

dfawley · 2025-07-22T20:51:08Z

grpc/Cargo.toml

+futures = "0.3.31"
+tower = { version = "0.5.2", features = ["buffer", "limit", "util"] }
+tower-service = "0.3.3"
+socket2 = "0.5.10"


Could you please keep these sorted so the diffs are easier to read? E.g. socket2 and tower-service are already present at the same number.

dfawley · 2025-07-22T20:53:41Z

grpc/src/client/subchannel.rs

@@ -229,6 +235,7 @@ impl InternalSubchannel {
        transport: Arc<dyn Transport>,
        backoff: Arc<dyn Backoff>,
        unregister_fn: Box<dyn FnOnce(SubchannelKey) + Send + Sync>,
+        runtime: Arc<dyn Runtime>,


Different grpc channels could theoretically use different runtimes. Maybe that isn't something we need to support, but it's pretty easily attained - it just requires passing around the runtime a bit more than if it were global.

dfawley · 2025-07-22T20:56:41Z

grpc/src/client/subchannel.rs

@@ -345,30 +353,34 @@ impl InternalSubchannel {
        let transport = self.transport.clone();
        let address = self.address().address;
        let state_machine_tx = self.state_machine_event_sender.clone();
-        let connect_task = tokio::task::spawn(async move {


Ultimately (pre-1.0) we want to not have any tokio runtime crates/features listed in Cargo.toml, except if you are using a tokio feature flag. That would prevent such a thing.

dfawley · 2025-07-22T21:00:25Z

grpc/src/client/transport/mod.rs

+// TODO: The following options are specific to HTTP/2. We should
+// instead pass an `Attribute` like struct to the connect method instead which
+// can hold config relevant to a particular transport.


FWIW I think it's good to eventually have a "common http2 configuration" struct that multiple h2 transports can share for configuration knobs common across any/most h2-based transport implementations. We can also have transport-specific configuration for tonic. And all of those should be in a generic transport configuration, attribute-like struct.

dfawley · 2025-07-22T21:01:21Z

grpc/src/client/transport/mod.rs

+    pub rate_limit: Option<(u64, Duration)>,
+    pub tcp_keepalive: Option<Duration>,
+    pub tcp_nodelay: bool,
+    pub connect_timeout: Option<Duration>,


deadline (Instant) instead?

grpc/src/codec.rs

dfawley · 2025-07-22T21:13:07Z

grpc/src/service.rs


 #[async_trait]
 pub trait Service: Send + Sync {
    async fn call(&self, method: String, request: Request) -> Response;
 }

 // TODO: define methods that will allow serialization/deserialization.
-pub trait Message: Any + Send + Sync {}
+pub trait Message: Any + Send + Sync {
+    fn as_any(self: Box<Self>) -> Box<dyn Any>;


I think these can be removed again, right? Because we're using trait upcasting elsewhere now.

Co-authored-by: Doug Fawley <[email protected]>

Add tonic transport

66e6c10

arjan-bal force-pushed the grpc-tonic-transport-1 branch from 1417e9d to 66e6c10 Compare July 15, 2025 07:27

Fix udeps and msrv

b113136

Merge remote-tracking branch 'source/master' into grpc-tonic-transport-1

957377c

arjan-bal force-pushed the grpc-tonic-transport-1 branch from fe1436e to 957377c Compare July 21, 2025 09:30

easwars reviewed Jul 21, 2025

View reviewed changes

arjan-bal added 3 commits July 23, 2025 20:38

Clippy fixes

13f1628

Address review

e69cb49

Use trait down casting

b01eca5

dfawley reviewed Jul 23, 2025

View reviewed changes

Update grpc/src/codec.rs

13bb667

Co-authored-by: Doug Fawley <[email protected]>

feat(grpc): Add tonic transport #2339

Are you sure you want to change the base?

feat(grpc): Add tonic transport #2339

Uh oh!

Conversation

arjan-bal commented Jul 15, 2025

Uh oh!

arjan-bal commented Jul 15, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

arjan-bal Jul 23, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

arjan-bal Jul 23, 2025 •

edited

Loading