fine grind the engine args

Author: YouNeedCryDear

config/runtimes/srt/Qwen/qwen-text-tp1-fp8-rt.yaml

@@ -138,7 +138,7 @@ spec:
         - --
       args:
         - |
-          python3 -m smg.launch_server \
+          smg serve \
           --enable-metrics \
           --host 0.0.0.0 \
           --log-requests \
@@ -149,6 +149,8 @@ spec:
           --tp-size 1 \
           --trust-remote-code \
           --grpc-mode \
+          --served-model-name vllm-model \
+          --router-log-json \
           --router-request-id-headers "opc-request-id"
       volumeMounts:
         - mountPath: /dev/shm

config/runtimes/srt/Qwen/qwen-text-tp1-rt.yaml

@@ -150,7 +150,7 @@ spec:
         - --
       args:
         - |
-          python3 -m smg.launch_server \
+          smg serve \
           --enable-metrics \
           --host 0.0.0.0 \
           --log-requests \
@@ -161,6 +161,8 @@ spec:
           --tp-size 1 \
           --trust-remote-code \
           --grpc-mode \
+          --served-model-name vllm-model \
+          --router-log-json \
           --router-request-id-headers "opc-request-id"
       volumeMounts:
         - mountPath: /dev/shm

config/runtimes/srt/Qwen/qwen-text-tp2-fp8-rt.yaml

@@ -138,7 +138,7 @@ spec:
         - --
       args:
         - |
-          python3 -m smg.launch_server \
+          smg serve \
           --enable-metrics \
           --host 0.0.0.0 \
           --log-requests \
@@ -149,6 +149,8 @@ spec:
           --tp-size 2 \
           --trust-remote-code \
           --grpc-mode \
+          --served-model-name vllm-model \
+          --router-log-json \
           --router-request-id-headers "opc-request-id"
       volumeMounts:
         - mountPath: /dev/shm

config/runtimes/srt/Qwen/qwen-text-tp2-rt.yaml

@@ -130,7 +130,7 @@ spec:
         - --
       args:
         - |
-          python3 -m smg.launch_server \
+          smg serve \
           --enable-metrics \
           --host 0.0.0.0 \
           --log-requests \
@@ -141,6 +141,8 @@ spec:
           --tp-size 2 \
           --trust-remote-code \
           --grpc-mode \
+          --served-model-name vllm-model \
+          --router-log-json \
           --router-request-id-headers "opc-request-id"
       volumeMounts:
         - mountPath: /dev/shm

config/runtimes/srt/Qwen/qwen-text-tp4-fp8-rt.yaml

@@ -135,7 +135,7 @@ spec:
         - --
       args:
         - |
-          python3 -m smg.launch_server \
+          smg serve \
           --enable-metrics \
           --host 0.0.0.0 \
           --log-requests \
@@ -146,6 +146,8 @@ spec:
           --tp-size 4 \
           --trust-remote-code \
           --grpc-mode \
+          --served-model-name vllm-model \
+          --router-log-json \
           --router-request-id-headers "opc-request-id"
       volumeMounts:
         - mountPath: /dev/shm

config/runtimes/srt/Qwen/qwen-text-tp4-rt.yaml

@@ -150,7 +150,7 @@ spec:
         - --
       args:
         - |
-          python3 -m smg.launch_server \
+          smg serve \
           --enable-metrics \
           --host 0.0.0.0 \
           --log-requests \
@@ -161,6 +161,8 @@ spec:
           --tp-size 4 \
           --trust-remote-code \
           --grpc-mode \
+          --served-model-name vllm-model \
+          --router-log-json \
           --router-request-id-headers "opc-request-id"
       volumeMounts:
         - mountPath: /dev/shm

config/runtimes/srt/Qwen/qwen-text-tp8-rt.yaml

@@ -127,7 +127,7 @@ spec:
         - --
       args:
         - |
-          python3 -m smg.launch_server \
+          smg serve \
           --enable-metrics \
           --host 0.0.0.0 \
           --log-requests \
@@ -138,6 +138,8 @@ spec:
           --tp-size 8 \
           --trust-remote-code \
           --grpc-mode \
+          --served-model-name vllm-model \
+          --router-log-json \
           --router-request-id-headers "opc-request-id"
       volumeMounts:
         - mountPath: /dev/shm

config/runtimes/srt/Qwen/qwen-vl-tp1-fp8-rt.yaml

@@ -127,7 +127,7 @@ spec:
         - --
       args:
         - |
-          python3 -m smg.launch_server \
+          smg serve \
           --enable-metrics \
           --enable-multimodal \
           --host 0.0.0.0 \
@@ -139,6 +139,8 @@ spec:
           --tp-size 1 \
           --trust-remote-code \
           --grpc-mode \
+          --served-model-name vllm-model \
+          --router-log-json \
           --router-request-id-headers "opc-request-id"
       volumeMounts:
         - mountPath: /dev/shm

config/runtimes/srt/Qwen/qwen-vl-tp1-rt.yaml

@@ -120,7 +120,7 @@ spec:
         - --
       args:
         - |
-          python3 -m smg.launch_server \
+          smg serve \
           --enable-metrics \
           --enable-multimodal \
           --host 0.0.0.0 \
@@ -132,6 +132,8 @@ spec:
           --tp-size 1 \
           --trust-remote-code \
           --grpc-mode \
+          --served-model-name vllm-model \
+          --router-log-json \
           --router-request-id-headers "opc-request-id"
       volumeMounts:
         - mountPath: /dev/shm

config/runtimes/srt/Qwen/qwen-vl-tp2-fp8-rt.yaml

@@ -127,7 +127,7 @@ spec:
         - --
       args:
         - |
-          python3 -m smg.launch_server \
+          smg serve \
           --enable-metrics \
           --enable-multimodal \
           --host 0.0.0.0 \
@@ -139,6 +139,8 @@ spec:
           --tp-size 2 \
           --trust-remote-code \
           --grpc-mode \
+          --served-model-name vllm-model \
+          --router-log-json \
           --router-request-id-headers "opc-request-id"
       volumeMounts:
         - mountPath: /dev/shm