From e05f9bfd83fba0dfaa82079746b30bc8bebd5d4b Mon Sep 17 00:00:00 2001
From: Ben Vincent <ben@unkin.net>
Date: Sat, 23 May 2026 17:59:43 +1000
Subject: [PATCH] feat: increase litellm resources (#144)

finding litellm performance has dropped, crashed in multiple cases, and
then it had scaled to the maximum level using the majority of memory in
cluster.

- reduce the rate at which litellm autoscales
- increase the requests/limits to match usage

Reviewed-on: https://git.unkin.net/unkin/argocd-apps/pulls/144
---
 apps/base/litellm/cnpg_cluster.yaml | 6 +++---
 apps/base/litellm/deployment.yaml   | 4 ++--
 apps/base/litellm/hpa.yaml          | 8 ++++----
 3 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/apps/base/litellm/cnpg_cluster.yaml b/apps/base/litellm/cnpg_cluster.yaml
index ea95389..5c09d29 100644
--- a/apps/base/litellm/cnpg_cluster.yaml
+++ b/apps/base/litellm/cnpg_cluster.yaml
@@ -76,11 +76,11 @@ spec:
     updateInterval: 30
   resources:
     limits:
-      cpu: 500m
-      memory: 512Mi
+      cpu: 1
+      memory: 1024Mi
     requests:
       cpu: 250m
-      memory: 256Mi
+      memory: 512Mi
   smartShutdownTimeout: 180
   startDelay: 3600
   stopDelay: 1800
diff --git a/apps/base/litellm/deployment.yaml b/apps/base/litellm/deployment.yaml
index 0f3b1fc..6039b10 100644
--- a/apps/base/litellm/deployment.yaml
+++ b/apps/base/litellm/deployment.yaml
@@ -56,10 +56,10 @@ spec:
           resources:
             limits:
               cpu: "2"
-              memory: 6Gi
+              memory: 8Gi
             requests:
               cpu: 250m
-              memory: 2Gi
+              memory: 6Gi
           volumeMounts:
             - mountPath: /app/config.yaml
               name: config
diff --git a/apps/base/litellm/hpa.yaml b/apps/base/litellm/hpa.yaml
index b954aa8..746626c 100644
--- a/apps/base/litellm/hpa.yaml
+++ b/apps/base/litellm/hpa.yaml
@@ -10,14 +10,14 @@ spec:
     kind: Deployment
     name: litellm
   minReplicas: 2
-  maxReplicas: 10
+  maxReplicas: 4
   metrics:
     - type: Resource
       resource:
         name: cpu
         target:
           type: Utilization
-          averageUtilization: 60
+          averageUtilization: 80
   behavior:
     scaleUp:
       stabilizationWindowSeconds: 0
@@ -25,7 +25,7 @@ spec:
       policies:
         - type: Percent
           value: 100
-          periodSeconds: 30
+          periodSeconds: 60
         - type: Pods
           value: 4
           periodSeconds: 30
@@ -34,7 +34,7 @@ spec:
       selectPolicy: Min
       policies:
         - type: Percent
-          value: 10
+          value: 30
           periodSeconds: 60
         - type: Pods
           value: 2