From 0e7e63a027ffa7bec200c692a46a839fcbd239fc Mon Sep 17 00:00:00 2001 From: roytman Date: Thu, 26 Feb 2026 15:56:15 +0200 Subject: [PATCH 1/2] move routing-sidecar from initContainer to a regular container Signed-off-by: roytman --- deploy/components/vllm-sim-pd/deployments.yaml | 3 +-- deploy/components/vllm-sim/deployments.yaml | 3 +-- test/e2e/yaml/vllm-sim-dp.yaml | 3 +-- test/e2e/yaml/vllm-sim-pd.yaml | 3 +-- test/sidecar/config/base/pod-qwen.yaml | 3 +-- test/sidecar/config/nixl/qwen-decoder-pod.yaml | 3 +-- 6 files changed, 6 insertions(+), 12 deletions(-) diff --git a/deploy/components/vllm-sim-pd/deployments.yaml b/deploy/components/vllm-sim-pd/deployments.yaml index 3568b9a76..4f6781edb 100644 --- a/deploy/components/vllm-sim-pd/deployments.yaml +++ b/deploy/components/vllm-sim-pd/deployments.yaml @@ -69,7 +69,7 @@ spec: app: ${POOL_NAME} llm-d.ai/role: decode spec: - initContainers: + containers: - name: routing-sidecar image: ${SIDECAR_IMAGE} imagePullPolicy: IfNotPresent @@ -110,7 +110,6 @@ spec: valueFrom: fieldRef: fieldPath: status.podIP - containers: - name: vllm image: ${VLLM_SIMULATOR_IMAGE} imagePullPolicy: IfNotPresent diff --git a/deploy/components/vllm-sim/deployments.yaml b/deploy/components/vllm-sim/deployments.yaml index 423b55a9e..0fe50819a 100644 --- a/deploy/components/vllm-sim/deployments.yaml +++ b/deploy/components/vllm-sim/deployments.yaml @@ -14,7 +14,7 @@ spec: labels: app: ${POOL_NAME} spec: - initContainers: + containers: - name: routing-sidecar image: ${SIDECAR_IMAGE} imagePullPolicy: IfNotPresent @@ -55,7 +55,6 @@ spec: valueFrom: fieldRef: fieldPath: status.podIP - containers: - name: vllm image: ${VLLM_SIMULATOR_IMAGE} imagePullPolicy: IfNotPresent diff --git a/test/e2e/yaml/vllm-sim-dp.yaml b/test/e2e/yaml/vllm-sim-dp.yaml index b0452d91e..38838a39e 100644 --- a/test/e2e/yaml/vllm-sim-dp.yaml +++ b/test/e2e/yaml/vllm-sim-dp.yaml @@ -15,7 +15,7 @@ spec: app: ${POOL_NAME} llm-d.ai/role: decode spec: - initContainers: + containers: - name: routing-sidecar image: ${SIDECAR_IMAGE} imagePullPolicy: IfNotPresent @@ -39,7 +39,6 @@ spec: valueFrom: fieldRef: fieldPath: status.podIP - containers: - name: vllm image: ${VLLM_SIMULATOR_IMAGE} imagePullPolicy: IfNotPresent diff --git a/test/e2e/yaml/vllm-sim-pd.yaml b/test/e2e/yaml/vllm-sim-pd.yaml index 7a8abf14c..21dffac20 100644 --- a/test/e2e/yaml/vllm-sim-pd.yaml +++ b/test/e2e/yaml/vllm-sim-pd.yaml @@ -58,7 +58,7 @@ spec: app: ${POOL_NAME} llm-d.ai/role: decode spec: - initContainers: + containers: - name: routing-sidecar image: ${SIDECAR_IMAGE} imagePullPolicy: IfNotPresent @@ -72,7 +72,6 @@ spec: - containerPort: 8000 protocol: TCP restartPolicy: Always - containers: - name: vllm image: ${VLLM_SIMULATOR_IMAGE} imagePullPolicy: IfNotPresent diff --git a/test/sidecar/config/base/pod-qwen.yaml b/test/sidecar/config/base/pod-qwen.yaml index 78cc91f21..c4fcf58d2 100644 --- a/test/sidecar/config/base/pod-qwen.yaml +++ b/test/sidecar/config/base/pod-qwen.yaml @@ -3,7 +3,7 @@ kind: Pod metadata: name: qwen2-0--5b spec: - initContainers: + containers: - name: routing-proxy image: routing-proxy args: @@ -22,7 +22,6 @@ spec: # - "true" # initialDelaySeconds: 1 # periodSeconds: 360 - containers: - name: vllm image: vllm/vllm-openai:v0.8.2 args: diff --git a/test/sidecar/config/nixl/qwen-decoder-pod.yaml b/test/sidecar/config/nixl/qwen-decoder-pod.yaml index 6e24e5a85..993148096 100644 --- a/test/sidecar/config/nixl/qwen-decoder-pod.yaml +++ b/test/sidecar/config/nixl/qwen-decoder-pod.yaml @@ -17,7 +17,7 @@ spec: operator: In values: - NVIDIA-A100-SXM4-80GB - initContainers: + containers: - name: routing-proxy image: routing-proxy securityContext: @@ -33,7 +33,6 @@ spec: protocol: TCP restartPolicy: Always imagePullPolicy: Always - containers: - name: vllm image: vllm-openai securityContext: From c33eee4f43a78ee0fb715a565f4e14f15a282b2a Mon Sep 17 00:00:00 2001 From: roytman Date: Fri, 27 Feb 2026 17:02:43 +0200 Subject: [PATCH 2/2] remove restartPolicy from routing-proxy Signed-off-by: roytman --- test/sidecar/config/base/pod-qwen.yaml | 1 - test/sidecar/config/nixl/qwen-decoder-pod.yaml | 1 - 2 files changed, 2 deletions(-) diff --git a/test/sidecar/config/base/pod-qwen.yaml b/test/sidecar/config/base/pod-qwen.yaml index c4fcf58d2..48fc84d61 100644 --- a/test/sidecar/config/base/pod-qwen.yaml +++ b/test/sidecar/config/base/pod-qwen.yaml @@ -9,7 +9,6 @@ spec: args: - "--port=8000" - "--vllm-port=8001" - restartPolicy: Always # readinessProbe: # exec: # command: diff --git a/test/sidecar/config/nixl/qwen-decoder-pod.yaml b/test/sidecar/config/nixl/qwen-decoder-pod.yaml index 993148096..d7ca4770a 100644 --- a/test/sidecar/config/nixl/qwen-decoder-pod.yaml +++ b/test/sidecar/config/nixl/qwen-decoder-pod.yaml @@ -31,7 +31,6 @@ spec: ports: - containerPort: 8000 protocol: TCP - restartPolicy: Always imagePullPolicy: Always - name: vllm image: vllm-openai