vllm-router 0.1.9__tar.gz → 0.1.10__tar.gz

{vllm_router-0.1.9 → vllm_router-0.1.10}/Cargo.toml

@@ -22,7 +22,15 @@ path = "src/main.rs"
 clap = { version = "4", features = ["derive"] }
 axum = { version = "0.8.4", features = ["macros", "ws", "tracing"] }
 tower = { version = "0.5", features = ["full"] }
-tower-http = { version = "0.6", features = ["trace", "compression-gzip", "cors", "timeout", "limit", "request-id", "util"] }
+tower-http = { version = "0.6", features = [
+    "trace",
+    "compression-gzip",
+    "cors",
+    "timeout",
+    "limit",
+    "request-id",
+    "util",
+] }
 serde = { version = "1.0", features = ["derive"] }
 serde_json = "1.0"
 bytes = "1.8.0"
@@ -36,7 +44,11 @@ http = "1.1.0"
 tokio = { version = "1.42.0", features = ["full"] }
 async-trait = "0.1"
 tracing = "0.1"
-tracing-subscriber = { version = "0.3", features = ["env-filter", "json", "chrono"] }
+tracing-subscriber = { version = "0.3", features = [
+    "env-filter",
+    "json",
+    "chrono",
+] }
 tracing-log = "0.2"
 tracing-appender = "0.2.3"
 chrono = "0.4"
@@ -52,10 +64,14 @@ regex = "1.10"
 url = "2.5.4"
 tokio-stream = { version = "0.1", features = ["sync"] }
 anyhow = "1.0"
+dotenvy = "0.15"
 tokenizers = { version = "0.22.2" }
 tiktoken-rs = { version = "0.7.0" }
 minijinja = { version = "2.0" }
-rustls = { version = "0.23", default-features = false, features = ["ring", "std"] }
+rustls = { version = "0.23", default-features = false, features = [
+    "ring",
+    "std",
+] }
 hf-hub = { version = "0.4.3", features = ["tokio"] }
 
 # gRPC and Protobuf dependencies

{vllm_router-0.1.9/py_src/vllm_router.egg-info → vllm_router-0.1.10}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: vllm-router
-Version: 0.1.9
+Version: 0.1.10
 Summary: High-performance Rust-based load balancer for VLLM with multiple routing algorithms and prefill-decode disaggregation support
 Author-email: Byron Hsu <byronhsu1230@gmail.com>
 License: Apache-2.0
@@ -64,7 +64,12 @@ cargo build --release
 ```
 
 #### Python Package
+Install from PyPI
 ```bash
+pip install vllm-router                                                                                                                                                        ```
+
+To build from source:
+```bash    
 pip install setuptools-rust wheel build
 python -m build
 pip install dist/*.whl
@@ -129,6 +134,19 @@ cargo run --release -- \
 
 ## Configuration
 
+### Authentication
+
+Enable bearer-token validation by listing validation URLs (comma-separated) in `.env` via `API_KEY_VALIDATION_URLS` or passing `--api-key-validation-urls`.
+When set, all HTTP endpoints require `Authorization: Bearer <token>` and tokens are validated with HTTP 200 responses.
+
+```bash
+# .env
+API_KEY_VALIDATION_URLS=https://codebase.helmholtz.cloud/api/v4/user
+
+# CLI override
+vllm-router --api-key-validation-urls https://codebase.helmholtz.cloud/api/v4/user
+```
+
 ### Metrics
 
 Prometheus metrics endpoint available at `127.0.0.1:29000` by default.

{vllm_router-0.1.9 → vllm_router-0.1.10}/README.md

@@ -42,7 +42,12 @@ cargo build --release
 ```
 
 #### Python Package
+Install from PyPI
 ```bash
+pip install vllm-router                                                                                                                                                        ```
+
+To build from source:
+```bash    
 pip install setuptools-rust wheel build
 python -m build
 pip install dist/*.whl
@@ -107,6 +112,19 @@ cargo run --release -- \
 
 ## Configuration
 
+### Authentication
+
+Enable bearer-token validation by listing validation URLs (comma-separated) in `.env` via `API_KEY_VALIDATION_URLS` or passing `--api-key-validation-urls`.
+When set, all HTTP endpoints require `Authorization: Bearer <token>` and tokens are validated with HTTP 200 responses.
+
+```bash
+# .env
+API_KEY_VALIDATION_URLS=https://codebase.helmholtz.cloud/api/v4/user
+
+# CLI override
+vllm-router --api-key-validation-urls https://codebase.helmholtz.cloud/api/v4/user
+```
+
 ### Metrics
 
 Prometheus metrics endpoint available at `127.0.0.1:29000` by default.

{vllm_router-0.1.9 → vllm_router-0.1.10/py_src/vllm_router.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: vllm-router
-Version: 0.1.9
+Version: 0.1.10
 Summary: High-performance Rust-based load balancer for VLLM with multiple routing algorithms and prefill-decode disaggregation support
 Author-email: Byron Hsu <byronhsu1230@gmail.com>
 License: Apache-2.0
@@ -64,7 +64,12 @@ cargo build --release
 ```
 
 #### Python Package
+Install from PyPI
 ```bash
+pip install vllm-router                                                                                                                                                        ```
+
+To build from source:
+```bash    
 pip install setuptools-rust wheel build
 python -m build
 pip install dist/*.whl
@@ -129,6 +134,19 @@ cargo run --release -- \
 
 ## Configuration
 
+### Authentication
+
+Enable bearer-token validation by listing validation URLs (comma-separated) in `.env` via `API_KEY_VALIDATION_URLS` or passing `--api-key-validation-urls`.
+When set, all HTTP endpoints require `Authorization: Bearer <token>` and tokens are validated with HTTP 200 responses.
+
+```bash
+# .env
+API_KEY_VALIDATION_URLS=https://codebase.helmholtz.cloud/api/v4/user
+
+# CLI override
+vllm-router --api-key-validation-urls https://codebase.helmholtz.cloud/api/v4/user
+```
+
 ### Metrics
 
 Prometheus metrics endpoint available at `127.0.0.1:29000` by default.

{vllm_router-0.1.9 → vllm_router-0.1.10}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "vllm-router"
-version = "0.1.9"
+version = "0.1.10"
 description = "High-performance Rust-based load balancer for VLLM with multiple routing algorithms and prefill-decode disaggregation support"
 authors = [{name = "Byron Hsu", email = "byronhsu1230@gmail.com"}]
 requires-python = ">=3.8"

{vllm_router-0.1.9 → vllm_router-0.1.10}/src/config/types.rs

@@ -1,4 +1,5 @@
 use super::ConfigResult;
+use crate::config::validation::ConfigValidator;
 use serde::{Deserialize, Serialize};
 use std::collections::HashMap;
 
@@ -29,6 +30,9 @@ pub struct RouterConfig {
     pub intra_node_data_parallel_size: usize,
     /// The api key used for the authorization with the worker
     pub api_key: Option<String>,
+    /// API key validation URLs (if set, incoming requests must validate against them)
+    #[serde(default)]
+    pub api_key_validation_urls: Vec<String>,
     /// Service discovery configuration (optional)
     pub discovery: Option<DiscoveryConfig>,
     /// Metrics configuration (optional)
@@ -417,6 +421,7 @@ impl Default for RouterConfig {
             worker_startup_check_interval_secs: 30,
             intra_node_data_parallel_size: 1,
             api_key: None,
+            api_key_validation_urls: vec![],
             discovery: None,
             metrics: None,
             log_dir: None,
@@ -455,7 +460,7 @@ impl RouterConfig {
 
     /// Validate the configuration
     pub fn validate(&self) -> ConfigResult<()> {
-        crate::config::validation::ConfigValidator::validate(self)
+        ConfigValidator::validate(self)
     }
 
     /// Get the routing mode type as a string
@@ -980,6 +985,7 @@ mod tests {
             worker_startup_check_interval_secs: 5,
             intra_node_data_parallel_size: 1,
             api_key: None,
+            api_key_validation_urls: vec![],
             discovery: Some(DiscoveryConfig {
                 enabled: true,
                 namespace: Some("vllm".to_string()),
@@ -1046,6 +1052,7 @@ mod tests {
             worker_startup_check_interval_secs: 15,
             intra_node_data_parallel_size: 1,
             api_key: None,
+            api_key_validation_urls: vec![],
             discovery: Some(DiscoveryConfig {
                 enabled: true,
                 namespace: None,
@@ -1103,6 +1110,7 @@ mod tests {
             worker_startup_check_interval_secs: 20,
             intra_node_data_parallel_size: 1,
             api_key: None,
+            api_key_validation_urls: vec![],
             discovery: Some(DiscoveryConfig {
                 enabled: true,
                 namespace: Some("production".to_string()),

{vllm_router-0.1.9 → vllm_router-0.1.10}/src/data_connector/response_memory_store.rs

@@ -158,7 +158,7 @@ impl ResponseStorage for MemoryResponseStorage {
                 .collect();
 
             // Sort by creation time (newest first)
-            responses_with_time.sort_by(|a, b| b.0.cmp(&a.0));
+            responses_with_time.sort_by_key(|b| std::cmp::Reverse(b.0));
 
             // Apply limit and collect the actual responses
             let limit = limit.unwrap_or(responses_with_time.len());

{vllm_router-0.1.9 → vllm_router-0.1.10}/src/lib.rs

@@ -44,7 +44,9 @@ struct Router {
     max_payload_size: usize,
     intra_node_data_parallel_size: usize,
     api_key: Option<String>,
+    api_key_validation_urls: Vec<String>,
     log_dir: Option<String>,
+
     log_level: Option<String>,
     service_discovery: bool,
     selector: HashMap<String, String>,
@@ -201,6 +203,7 @@ impl Router {
             worker_startup_check_interval_secs: self.worker_startup_check_interval,
             intra_node_data_parallel_size: self.intra_node_data_parallel_size,
             api_key: self.api_key.clone(),
+            api_key_validation_urls: self.api_key_validation_urls.clone(),
             discovery,
             metrics,
             log_dir: self.log_dir.clone(),
@@ -261,6 +264,7 @@ impl Router {
         max_payload_size = 512 * 1024 * 1024,  // 512MB default for large batches
         intra_node_data_parallel_size = 1,
         api_key = None,
+        api_key_validation_urls = vec![],
         log_dir = None,
         log_level = None,
         service_discovery = false,
@@ -326,6 +330,7 @@ impl Router {
         max_payload_size: usize,
         intra_node_data_parallel_size: usize,
         api_key: Option<String>,
+        api_key_validation_urls: Vec<String>,
         log_dir: Option<String>,
         log_level: Option<String>,
         service_discovery: bool,
@@ -402,6 +407,7 @@ impl Router {
             max_payload_size,
             intra_node_data_parallel_size,
             api_key,
+            api_key_validation_urls,
             log_dir,
             log_level,
             service_discovery,

{vllm_router-0.1.9 → vllm_router-0.1.10}/src/main.rs

@@ -191,6 +191,10 @@ struct CliArgs {
     #[arg(long)]
     api_key: Option<String>,
 
+    /// API key validation URLs (defaults to env file)
+    #[arg(long, num_args = 0..)]
+    api_key_validation_urls: Vec<String>,
+
     /// Backend to route requests to (vllm, trtllm, openai, anthropic)
     #[arg(long, value_enum, default_value_t = Backend::Vllm, alias = "runtime")]
     backend: Backend,
@@ -540,6 +544,18 @@ impl CliArgs {
             _ => Self::determine_connection_mode(&all_urls),
         };
 
+        let api_key_validation_urls = if !self.api_key_validation_urls.is_empty() {
+            self.api_key_validation_urls.clone()
+        } else if let Ok(raw_urls) = std::env::var("API_KEY_VALIDATION_URLS") {
+            raw_urls
+                .split(',')
+                .map(|url| url.trim().to_string())
+                .filter(|url| !url.is_empty())
+                .collect()
+        } else {
+            Vec::new()
+        };
+
         // Build RouterConfig
         Ok(RouterConfig {
             mode,
@@ -553,6 +569,7 @@ impl CliArgs {
             worker_startup_check_interval_secs: self.worker_startup_check_interval,
             intra_node_data_parallel_size: self.intra_node_data_parallel_size,
             api_key: self.api_key.clone(),
+            api_key_validation_urls,
             discovery,
             metrics,
             log_dir: self.log_dir.clone(),
@@ -646,6 +663,7 @@ impl CliArgs {
 }
 
 fn main() -> Result<(), Box<dyn std::error::Error>> {
+    dotenvy::dotenv().ok();
     println!("DEBUG: Main function started");
 
     // Parse prefill arguments manually before clap parsing