llm-d · github-actions · Feb 6, 2026 · Dec 9, 2025 · Dec 9, 2025 · Dec 10, 2025
diff --git a/.gitignore b/.gitignore
@@ -29,6 +29,7 @@ go.work.sum
 # Environment Files
 .DS_Store
 .env
+CLAUDE.md
 
 # IDE files
 .idea

diff --git a/Makefile b/Makefile
@@ -25,7 +25,7 @@ export EPP_IMAGE ?= $(IMAGE_TAG_BASE):$(EPP_TAG)
 SIDECAR_TAG ?= dev
 SIDECAR_IMAGE_TAG_BASE ?= $(IMAGE_REGISTRY)/$(SIDECAR_IMAGE_NAME)
 export SIDECAR_IMAGE ?= $(SIDECAR_IMAGE_TAG_BASE):$(SIDECAR_TAG)
-VLLM_SIMULATOR_TAG ?= v0.6.1
+VLLM_SIMULATOR_TAG ?= latest
 VLLM_SIMULATOR_TAG_BASE ?= $(IMAGE_REGISTRY)/$(VLLM_SIMULATOR_IMAGE_NAME)
 export VLLM_SIMULATOR_IMAGE ?= $(VLLM_SIMULATOR_TAG_BASE):$(VLLM_SIMULATOR_TAG)
 NAMESPACE ?= hc4ai-operator

diff --git a/cmd/pd-sidecar/main.go b/cmd/pd-sidecar/main.go
@@ -35,7 +35,7 @@ var (
 	// supportedConnectors defines all valid P/D connector types
 	supportedConnectors = []string{
 		proxy.ConnectorNIXLV2,
-		proxy.ConnectorLMCache,
+		proxy.ConnectorSharedStorage,
 		proxy.ConnectorSGLang,
 	}
 )

diff --git a/pkg/sidecar/proxy/connector_lmcache.go b/pkg/sidecar/proxy/connector_lmcache.go
diff --git a/pkg/sidecar/proxy/connector_nixlv2.go b/pkg/sidecar/proxy/connector_nixlv2.go
@@ -107,7 +107,7 @@ func (s *Server) runNIXLProtocolV2(w http.ResponseWriter, r *http.Request, prefi
 	pw := &bufferedResponseWriter{}
 	prefillHandler.ServeHTTP(pw, preq)
 
-	if pw.statusCode < 200 || pw.statusCode >= 300 {
+	if isHTTPError(pw.statusCode) {
 		s.logger.Error(err, "request failed", "code", pw.statusCode)
 		w.WriteHeader(pw.statusCode)
 		return