Merge branch 'main' into renovate/semver-1.x-lockfile

README.md

@@ -8,28 +8,6 @@
 [![Coverage Status](https://coveralls.io/repos/github/Unleash/unleash-edge/badge.svg?branch=main)](https://coveralls.io/github/Unleash/unleash-edge?branch=main)
 ![downloads](https://img.shields.io/crates/d/unleash-edge.svg)
 
-- [Overview](#overview)
-- [Quickstart](#quickstart)
-- [Edge behaviors](#edge-behaviors)
-  - [Strict behavior](#strict-behavior)
-  - [Dynamic behavior](#dynamic-behavior)
-- [Getting Unleash Edge](#getting-unleash-edge)
-- [Running Unleash Edge](#running-unleash-edge)
-- [Metrics](#metrics)
-  - [Prometheus integration](#prometheus-integration)
-- [Compatibility](#compatibility)
-- [Debugging](#debugging)
-- [Additional resources](#additional-resources)
-  - [Edge concepts](#edge-concepts)
-  - [CLI](#cli)
-  - [Deploying Edge](#deploying-edge)
-  - [Migrating from Unleash Proxy](#migrating-from-unleash-proxy)
-  - [Performance benchmarking](#performance-benchmarking)
-  - [Contribution and development guide](#contribution-and-development-guide)
-
-
-> Availability: Unleash v4.15+.
-
 ## Overview
 
 Unleash Edge is a fast and lightweight proxy layer between your Unleash API and SDKs. It acts as a read replica of your Unleash instance and is designed to help you scale Unleash. It allows you to support thousands of connected SDKs without increasing the number of requests you make to your Unleash instance.
@@ -64,6 +42,12 @@ For example:
 docker run -it -p 3063:3063 -e STRICT=true -e UPSTREAM_URL=https://app.unleash-hosted.com/testclient -e TOKENS='*:development.4a798ad11cde8c0e637ff19f3287683ebc21d23d607c641f2dd79daa54' unleashorg/unleash-edge:v19.6.2 edge
 ```
 
+## Versioning and availability
+
+Unleash Edge is versioned and released independently of [Unleash](https://github.com/Unleash/unleash). To use Unleash Edge, you need Unleash version 4.15 or later. We recommend using the latest versions of Unleash and Unleash Edge to ensure optimal performance and access to the latest features and security updates.
+
+Unleash Edge does not have full feature parity with Unleash. Some features, such as filtering feature flags by tags, are not supported.
+
 ## Edge behaviors
 
 > Availability: Unleash Edge v19.2+

server/README.md

@@ -8,28 +8,6 @@
 [![Coverage Status](https://coveralls.io/repos/github/Unleash/unleash-edge/badge.svg?branch=main)](https://coveralls.io/github/Unleash/unleash-edge?branch=main)
 ![downloads](https://img.shields.io/crates/d/unleash-edge.svg)
 
-- [Overview](#overview)
-- [Quickstart](#quickstart)
-- [Edge behaviors](#edge-behaviors)
-  - [Strict behavior](#strict-behavior)
-  - [Dynamic behavior](#dynamic-behavior)
-- [Getting Unleash Edge](#getting-unleash-edge)
-- [Running Unleash Edge](#running-unleash-edge)
-- [Metrics](#metrics)
-  - [Prometheus integration](#prometheus-integration)
-- [Compatibility](#compatibility)
-- [Debugging](#debugging)
-- [Additional resources](#additional-resources)
-  - [Edge concepts](#edge-concepts)
-  - [CLI](#cli)
-  - [Deploying Edge](#deploying-edge)
-  - [Migrating from Unleash Proxy](#migrating-from-unleash-proxy)
-  - [Performance benchmarking](#performance-benchmarking)
-  - [Contribution and development guide](#contribution-and-development-guide)
-
-
-> Availability: Unleash v4.15+.
-
 ## Overview
 
 Unleash Edge is a fast and lightweight proxy layer between your Unleash API and SDKs. It acts as a read replica of your Unleash instance and is designed to help you scale Unleash. It allows you to support thousands of connected SDKs without increasing the number of requests you make to your Unleash instance.
@@ -64,6 +42,12 @@ For example:
 docker run -it -p 3063:3063 -e STRICT=true -e UPSTREAM_URL=https://app.unleash-hosted.com/testclient -e TOKENS='*:development.4a798ad11cde8c0e637ff19f3287683ebc21d23d607c641f2dd79daa54' unleashorg/unleash-edge:v19.6.2 edge
 ```
 
+## Versioning and availability
+
+Unleash Edge is versioned and released independently of [Unleash](https://github.com/Unleash/unleash). To use Unleash Edge, you need Unleash version 4.15 or later. We recommend using the latest versions of Unleash and Unleash Edge to ensure optimal performance and access to the latest features and security updates.
+
+Unleash Edge does not have full feature parity with Unleash. Some features, such as filtering feature flags by tags, is not supported.
+
 ## Edge behaviors
 
 > Availability: Unleash Edge v19.2+

server/src/client_api.rs

@@ -43,20 +43,16 @@ pub async fn get_features(
 #[get("/streaming")]
 pub async fn stream_features(
     edge_token: EdgeToken,
+    broadcaster: Data<Broadcaster>,
     token_cache: Data<DashMap<String, EdgeToken>>,
     filter_query: Query<FeatureFilters>,
-    req: HttpRequest,
 ) -> EdgeResult<impl Responder> {
     let (validated_token, _filter_set, query) =
         get_feature_filter(&edge_token, &token_cache, filter_query.clone())?;
-    match req.app_data::<Data<Broadcaster>>() {
-        Some(broadcaster) => {
-            broadcaster
-                .connect(validated_token, filter_query, query)
-                .await
-        }
-        _ => Err(EdgeError::ClientCacheError),
-    }
+
+    broadcaster
+        .connect(validated_token, filter_query, query)
+        .await
 }
 
 #[utoipa::path(

server/src/error.rs

@@ -97,27 +97,28 @@ pub enum EdgeError {
     ClientFeaturesParseError(String),
     ClientHydrationFailed(String),
     ClientRegisterError,
-    FrontendNotYetHydrated(FrontendHydrationMissing),
-    FrontendExpectedToBeHydrated(String),
-    FeatureNotFound(String),
-    PersistenceError(String),
+    ContextParseError,
     EdgeMetricsError,
     EdgeMetricsRequestError(reqwest::StatusCode, Option<UnleashBadRequest>),
     EdgeTokenError,
     EdgeTokenParseError,
+    FeatureNotFound(String),
+    FrontendExpectedToBeHydrated(String),
+    FrontendNotYetHydrated(FrontendHydrationMissing),
+    HealthCheckError(String),
     InvalidBackupFile(String, String),
     InvalidServerUrl(String),
     InvalidTokenWithStrictBehavior,
-    HealthCheckError(String),
     JsonParseError(String),
     NoFeaturesFile,
     NoTokenProvider,
     NoTokens(String),
     NotReady,
+    PersistenceError(String),
     ReadyCheckError(String),
+    SseError(String),
     TlsError,
     TokenParseError(String),
-    ContextParseError,
     TokenValidationError(reqwest::StatusCode),
 }
 
@@ -210,6 +211,7 @@ impl Display for EdgeError {
                 write!(f, "Edge is not ready to serve requests")
             }
             EdgeError::InvalidTokenWithStrictBehavior => write!(f, "Edge is running with strict behavior and the token is not subsumed by any registered tokens"),
+            EdgeError::SseError(message) => write!(f, "{}", message),
         }
     }
 }
@@ -250,6 +252,7 @@ impl ResponseError for EdgeError {
             EdgeError::FrontendExpectedToBeHydrated(_) => StatusCode::INTERNAL_SERVER_ERROR,
             EdgeError::NotReady => StatusCode::SERVICE_UNAVAILABLE,
             EdgeError::InvalidTokenWithStrictBehavior => StatusCode::FORBIDDEN,
+            EdgeError::SseError(_) => StatusCode::INTERNAL_SERVER_ERROR,
         }
     }
 
@@ -294,9 +297,8 @@ impl From<serde_json::Error> for EdgeError {
 }
 
 impl From<SendError<Event>> for EdgeError {
-    // todo: create better enum representation. use this is placeholder
-    fn from(_value: SendError<Event>) -> Self {
-        EdgeError::TlsError
+    fn from(value: SendError<Event>) -> Self {
+        EdgeError::SseError(value.to_string())
     }
 }
 

server/src/http/broadcaster.rs

@@ -14,6 +14,7 @@ use actix_web_lab::{
 };
 use dashmap::DashMap;
 use futures::future;
+use prometheus::{register_int_gauge, IntGauge};
 use serde::Serialize;
 use tokio::sync::mpsc;
 use tokio_stream::wrappers::ReceiverStream;
@@ -51,6 +52,14 @@ pub struct Broadcaster {
     features_cache: Arc<FeatureCache>,
 }
 
+lazy_static::lazy_static! {
+    pub static ref CONNECTED_STREAMING_CLIENTS: IntGauge = register_int_gauge!(
+        "connected_streaming_clients",
+        "Number of connected streaming clients",
+    )
+    .unwrap();
+}
+
 impl Broadcaster {
     /// Constructs new broadcaster and spawns ping loop.
     pub fn new(features: Arc<FeatureCache>) -> Arc<Self> {
@@ -90,6 +99,7 @@ impl Broadcaster {
 
     /// Removes all non-responsive clients from broadcast list.
     async fn heartbeat(&self) {
+        let mut active_connections = 0i64;
         for mut group in self.active_connections.iter_mut() {
             let mut ok_clients = Vec::new();
 
@@ -103,11 +113,10 @@ impl Broadcaster {
                 }
             }
 
-            // validate tokens here?
-            // ok_clients.iter().filter(|client| client.token_is_valid())
-
+            active_connections += ok_clients.len() as i64;
             group.clients = ok_clients;
         }
+        CONNECTED_STREAMING_CLIENTS.set(active_connections)
     }
 
     /// Registers client with broadcaster, returning an SSE response body.
@@ -140,7 +149,6 @@ impl Broadcaster {
                 filter_set,
                 token,
             });
-
         Ok(Sse::from_infallible_receiver(rx))
     }
 
@@ -181,7 +189,7 @@ impl Broadcaster {
             // 1. We'll only allow streaming in strict mode
             // 2. We'll check whether the token is subsumed *before* trying to add it to the broadcaster
             // If both of these are true, then we should never hit this case (if Thomas's understanding is correct).
-            None => Err(EdgeError::ClientCacheError),
+            None => Err(EdgeError::AuthorizationDenied),
         }
     }
 

server/src/prom_metrics.rs

@@ -143,6 +143,11 @@ fn register_custom_metrics(registry: &prometheus::Registry) {
             crate::metrics::client_metrics::FEATURE_TOGGLE_USAGE_TOTAL.clone(),
         ))
         .unwrap();
+    registry
+        .register(Box::new(
+            crate::http::broadcaster::CONNECTED_STREAMING_CLIENTS.clone(),
+        ))
+        .unwrap();
 }
 
 #[cfg(test)]