diff --git a/pingora-proxy/examples/ctx.rs b/pingora-proxy/examples/ctx.rs
index 3284b6f02..05e2bf4e8 100644
--- a/pingora-proxy/examples/ctx.rs
+++ b/pingora-proxy/examples/ctx.rs
@@ -17,7 +17,7 @@ use clap::Parser;
 use log::info;
 use std::sync::Mutex;
 
-use pingora_core::{protocols::http::ServerSession, server::configuration::Opt};
+use pingora_core::server::configuration::Opt;
 use pingora_core::server::Server;
 use pingora_core::upstreams::peer::HttpPeer;
 use pingora_core::Result;
@@ -43,7 +43,7 @@ fn check_beta_user(req: &pingora_http::RequestHeader) -> bool {
 #[async_trait]
 impl ProxyHttp for MyProxy {
     type CTX = MyCtx;
-    fn new_ctx(&self, session: &Box<ServerSession>) -> Self::CTX {
+    fn new_ctx(&self, session: &Session) -> Self::CTX {
         MyCtx { beta_user: false }
     }
 
diff --git a/pingora-proxy/examples/gateway.rs b/pingora-proxy/examples/gateway.rs
index 06b366cd4..581e53862 100644
--- a/pingora-proxy/examples/gateway.rs
+++ b/pingora-proxy/examples/gateway.rs
@@ -17,10 +17,10 @@ use clap::Parser;
 use log::info;
 use prometheus::register_int_counter;
 
+use pingora_core::server::configuration::Opt;
 use pingora_core::server::Server;
 use pingora_core::upstreams::peer::HttpPeer;
 use pingora_core::Result;
-use pingora_core::{protocols::http::ServerSession, server::configuration::Opt};
 use pingora_http::ResponseHeader;
 use pingora_proxy::{ProxyHttp, Session};
 
@@ -36,7 +36,7 @@ pub struct MyGateway {
 #[async_trait]
 impl ProxyHttp for MyGateway {
     type CTX = ();
-    fn new_ctx(&self, _session: &Box<ServerSession>) -> Self::CTX {}
+    fn new_ctx(&self, _session: &Session) -> Self::CTX {}
 
     async fn request_filter(&self, session: &mut Session, _ctx: &mut Self::CTX) -> Result<bool> {
         if session.req_header().uri.path().starts_with("/login")
diff --git a/pingora-proxy/examples/load_balancer.rs b/pingora-proxy/examples/load_balancer.rs
index 13c1b6572..0d599ecb7 100644
--- a/pingora-proxy/examples/load_balancer.rs
+++ b/pingora-proxy/examples/load_balancer.rs
@@ -15,7 +15,7 @@
 use async_trait::async_trait;
 use clap::Parser;
 use log::info;
-use pingora_core::{protocols::http::ServerSession, services::background::background_service};
+use pingora_core::services::background::background_service;
 use std::{sync::Arc, time::Duration};
 
 use pingora_core::server::configuration::Opt;
@@ -30,7 +30,7 @@ pub struct LB(Arc<LoadBalancer<RoundRobin>>);
 #[async_trait]
 impl ProxyHttp for LB {
     type CTX = ();
-    fn new_ctx(&self, _session: &Box<ServerSession>) -> Self::CTX {}
+    fn new_ctx(&self, _session: &Session) -> Self::CTX {}
 
     async fn upstream_peer(&self, _session: &mut Session, _ctx: &mut ()) -> Result<Box<HttpPeer>> {
         let upstream = self
diff --git a/pingora-proxy/examples/modify_response.rs b/pingora-proxy/examples/modify_response.rs
index 3d6355b3f..bea3fa535 100644
--- a/pingora-proxy/examples/modify_response.rs
+++ b/pingora-proxy/examples/modify_response.rs
@@ -18,10 +18,10 @@ use clap::Parser;
 use serde::{Deserialize, Serialize};
 use std::net::ToSocketAddrs;
 
+use pingora_core::server::configuration::Opt;
 use pingora_core::server::Server;
 use pingora_core::upstreams::peer::HttpPeer;
 use pingora_core::Result;
-use pingora_core::{protocols::http::ServerSession, server::configuration::Opt};
 use pingora_http::ResponseHeader;
 use pingora_proxy::{ProxyHttp, Session};
 
@@ -43,7 +43,7 @@ pub struct MyCtx {
 #[async_trait]
 impl ProxyHttp for Json2Yaml {
     type CTX = MyCtx;
-    fn new_ctx(&self, _session: &Box<ServerSession>) -> Self::CTX {
+    fn new_ctx(&self, _session: &Session) -> Self::CTX {
         MyCtx { buffer: vec![] }
     }
 
diff --git a/pingora-proxy/examples/multi_lb.rs b/pingora-proxy/examples/multi_lb.rs
index a8f5da5cc..ab9cef305 100644
--- a/pingora-proxy/examples/multi_lb.rs
+++ b/pingora-proxy/examples/multi_lb.rs
@@ -15,7 +15,7 @@
 use async_trait::async_trait;
 use std::sync::Arc;
 
-use pingora_core::{prelude::*, protocols::http::ServerSession, services::background::GenBackgroundService};
+use pingora_core::{prelude::*, services::background::GenBackgroundService};
 use pingora_load_balancing::{
     health_check::TcpHealthCheck,
     selection::{BackendIter, BackendSelection, RoundRobin},
@@ -31,7 +31,7 @@ struct Router {
 #[async_trait]
 impl ProxyHttp for Router {
     type CTX = ();
-    fn new_ctx(&self, _session: &Box<ServerSession>) {}
+    fn new_ctx(&self, _session: &Session) {}
 
     async fn upstream_peer(&self, session: &mut Session, _ctx: &mut ()) -> Result<Box<HttpPeer>> {
         // determine LB cluster based on request uri
diff --git a/pingora-proxy/src/lib.rs b/pingora-proxy/src/lib.rs
index 6e5731a0c..49f70918a 100644
--- a/pingora-proxy/src/lib.rs
+++ b/pingora-proxy/src/lib.rs
@@ -680,17 +680,17 @@ where
         sub_req_ctx: Box<SubReqCtx>,
     ) {
         debug!("starting subrequest");
-        let ctx = self.inner.new_ctx(&session);
         let mut session = match self.handle_new_request(session).await {
             Some(downstream_session) => Session::new(downstream_session, &self.downstream_modules),
             None => return, // bad request
         };
-
+        
         // no real downstream to keepalive, but it doesn't matter what is set here because at the end
         // of this fn the dummy connection will be dropped
         session.set_keepalive(None);
-
+        
         session.subrequest_ctx.replace(sub_req_ctx);
+        let ctx = self.inner.new_ctx(&session);
         trace!("processing subrequest");
         self.process_request(session, ctx).await;
         trace!("subrequest done");
@@ -709,13 +709,12 @@ where
         shutdown: &ShutdownWatch,
     ) -> Option<Stream> {
         let session = Box::new(session);
-        let ctx = self.inner.new_ctx(&session);
         // TODO: keepalive pool, use stack
         let mut session = match self.handle_new_request(session).await {
             Some(downstream_session) => Session::new(downstream_session, &self.downstream_modules),
             None => return None, // bad request
         };
-
+        
         if *shutdown.borrow() {
             // stop downstream from reusing if this service is shutting down soon
             session.set_keepalive(None);
@@ -723,7 +722,8 @@ where
             // default 60s
             session.set_keepalive(Some(60));
         }
-
+        
+        let ctx = self.inner.new_ctx(&session);
         self.process_request(session, ctx).await
     }
 
diff --git a/pingora-proxy/src/proxy_trait.rs b/pingora-proxy/src/proxy_trait.rs
index 9feae715e..346d218a2 100644
--- a/pingora-proxy/src/proxy_trait.rs
+++ b/pingora-proxy/src/proxy_trait.rs
@@ -28,7 +28,7 @@ pub trait ProxyHttp {
     type CTX;
 
     /// Define how the `ctx` should be created.
-    fn new_ctx(&self, session: &Box<HttpSession>) -> Self::CTX;
+    fn new_ctx(&self, session: &Session) -> Self::CTX;
 
     /// Define where the proxy should send the request to.
     ///
diff --git a/pingora-proxy/tests/utils/server_utils.rs b/pingora-proxy/tests/utils/server_utils.rs
index c7977fca8..3375ecf7e 100644
--- a/pingora-proxy/tests/utils/server_utils.rs
+++ b/pingora-proxy/tests/utils/server_utils.rs
@@ -110,7 +110,7 @@ fn response_filter_common(
 #[async_trait]
 impl ProxyHttp for ExampleProxyHttps {
     type CTX = CTX;
-    fn new_ctx(&self, session: &Box<ServerSession>) -> Self::CTX {
+    fn new_ctx(&self, session: &Session) -> Self::CTX {
         CTX::default()
     }
 
@@ -207,7 +207,7 @@ pub struct ExampleProxyHttp {}
 #[async_trait]
 impl ProxyHttp for ExampleProxyHttp {
     type CTX = CTX;
-    fn new_ctx(&self, session: &Box<ServerSession>) -> Self::CTX {
+    fn new_ctx(&self, session: &Session) -> Self::CTX {
         CTX::default()
     }
 
@@ -340,7 +340,7 @@ pub struct ExampleProxyCache {}
 #[async_trait]
 impl ProxyHttp for ExampleProxyCache {
     type CTX = CacheCTX;
-    fn new_ctx(&self, session: &Box<ServerSession>) -> Self::CTX {
+    fn new_ctx(&self, session: &Session) -> Self::CTX {
         CacheCTX {
             upstream_status: None,
         }