feat: add cli args for example http service (#221)

Co-authored-by: Biswa Ranjan Panda <biswaranjanp@nvidia.com>

examples/python_rs/llm/vllm/README.md

@@ -59,7 +59,7 @@ Run the server logging (with debug level logging):
 ```bash
 TRD_LOG=DEBUG http
 ```
-By default the server will run on port 9992.
+By default the server will run on port 8080.
 
 Add model to the server:
 ```bash
@@ -116,7 +116,7 @@ The disaggregated deployment utilizes separate GPUs for prefill and decode opera
 ### 3. Client
 
 ```bash
-curl localhost:9992/v1/chat/completions \
+curl localhost:8080/v1/chat/completions \
   -H "Content-Type: application/json" \
   -d '{
     "model": "deepseek-ai/DeepSeek-R1-Distill-Llama-8B",

examples/rust/Cargo.lock

@@ -1127,6 +1127,7 @@ dependencies = [
 name = "http"
 version = "0.2.0"
 dependencies = [
+ "clap",
  "serde",
  "serde_json",
  "tokio",

examples/rust/http/Cargo.toml

@@ -26,6 +26,7 @@ repository.workspace = true
 [dependencies]
 triton-distributed = { workspace = true}
 triton-llm = { workspace = true}
+clap = { version = "4.5", features = ["derive"] }
 
 serde = { workspace = true }
 serde_json = { workspace = true }

examples/rust/http/src/main.rs

@@ -14,6 +14,8 @@
 // limitations under the License.
 
 use std::sync::Arc;
+use clap::Parser;
+use std::env;
 
 use triton_distributed::{logging, DistributedRuntime, Result, Runtime, Worker};
 use triton_llm::http::service::{
@@ -21,6 +23,26 @@ use triton_llm::http::service::{
     service_v2::HttpService,
 };
 
+#[derive(Parser)]
+#[command(author, version, about, long_about = None)]
+struct Args {
+    /// Host for the HTTP service
+    #[arg(long, default_value = "0.0.0.0")]
+    host: String,
+
+    /// Port number for the HTTP service
+    #[arg(short, long, default_value = "8080")]
+    port: u16,
+
+    /// Namespace for the distributed component
+    #[arg(long, default_value = "public")]
+    namespace: String,
+
+    /// Component name for the service
+    #[arg(long, default_value = "http")]
+    component: String,
+}
+
 fn main() -> Result<()> {
     logging::init();
     let worker = Worker::from_settings()?;
@@ -30,8 +52,13 @@ fn main() -> Result<()> {
 async fn app(runtime: Runtime) -> Result<()> {
     let distributed = DistributedRuntime::from_settings(runtime.clone()).await?;
 
+    let args = Args::parse();
+
     // create the http service and acquire the model manager
-    let http_service = HttpService::builder().port(9992).build()?;
+    let http_service = HttpService::builder()
+        .port(args.port)
+        .host(args.host)
+        .build()?;
     let manager = http_service.model_manager().clone();
 
     // todo - use the IntoComponent trait to register the component
@@ -42,7 +69,7 @@ async fn app(runtime: Runtime) -> Result<()> {
     // written to etcd
     // the cli when operating on an `http` component will validate the namespace.component is
     // registered with HttpServiceComponentDefinition
-    let component = distributed.namespace("public")?.component("http")?;
+    let component = distributed.namespace(&args.namespace)?.component(&args.component)?;
 
     let etcd_root = component.etcd_path();
     let etcd_path = format!("{}/models/chat/", etcd_root);

llm/rust/triton-llm/src/http/service/service_v2.rs

@@ -25,6 +25,7 @@ pub struct HttpService {
     models: ModelManager,
     router: axum::Router,
     port: u16,
+    host: String,
 }
 
 #[derive(Clone, Builder)]
@@ -33,6 +34,9 @@ pub struct HttpServiceConfig {
     #[builder(default = "8787")]
     port: u16,
 
+    #[builder(setter(into), default = "String::from(\"0.0.0.0\")")]
+    host: String,
+
     // #[builder(default)]
     // custom: Vec<axum::Router>
     #[builder(default = "true")]
@@ -57,7 +61,7 @@ impl HttpService {
     }
 
     pub async fn run(&self, cancel_token: CancellationToken) -> Result<()> {
-        let address = format!("0.0.0.0:{}", self.port);
+        let address = format!("{}:{}", self.host, self.port);
         tracing::info!(address, "Starting HTTP service on: {address}");
 
         let listener = tokio::net::TcpListener::bind(address.as_str())
@@ -122,6 +126,7 @@ impl HttpServiceConfigBuilder {
             models: model_manager,
             router,
             port: config.port,
+            host: config.host,
         })
     }
 }