From cfd5a96331174d5f9ed8eb8ab203d8a62dd4fae0 Mon Sep 17 00:00:00 2001
From: Story Crater Bot <19826264+Riotpiaole@users.noreply.github.com>
Date: Fri, 21 Aug 2026 16:23:07 -0700
Subject: [PATCH] fix: retire one reasoning-predictor replica, run PP=2 across
 both V100s (Qwen3.5 MoE swap abandoned, moving to Ollama)

---
 k8s/apps/llm-serving/reasoning.yaml | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/k8s/apps/llm-serving/reasoning.yaml b/k8s/apps/llm-serving/reasoning.yaml
index 18ffdf1..19aae96 100644
--- a/k8s/apps/llm-serving/reasoning.yaml
+++ b/k8s/apps/llm-serving/reasoning.yaml
@@ -29,6 +29,7 @@ spec:
       - --dtype=float16
       - --kv-cache-dtype=auto
       - --tensor-parallel-size=1
+      - --pipeline-parallel-size=2
       - --max-model-len=16384
       - --gpu-memory-utilization=0.90
       - --max-num-seqs=4
@@ -85,11 +86,11 @@ spec:
         limits:
           cpu: '16'
           memory: 36Gi
-          nvidia.com/gpu: '1'
+          nvidia.com/gpu: '2'
         requests:
           cpu: '8'
           memory: 12Gi
-          nvidia.com/gpu: '1'
+          nvidia.com/gpu: '2'
       startupProbe:
         failureThreshold: 80
         httpGet:
@@ -103,8 +104,8 @@ spec:
         name: shm
     deploymentStrategy:
       type: Recreate
-    maxReplicas: 2
-    minReplicas: 2
+    maxReplicas: 1
+    minReplicas: 1
     nodeSelector:
       kubernetes.io/hostname: worker-1
     runtimeClassName: nvidia