Erich Blume
a32c99a252
Prevents OOM when switching between models — only one 14B model fits in 16GB VRAM at a time with KV cache for context. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
88 lines
2.3 KiB
YAML
88 lines
2.3 KiB
YAML
---
|
|
apiVersion: apps/v1
|
|
kind: Deployment
|
|
metadata:
|
|
name: ollama
|
|
namespace: ollama
|
|
spec:
|
|
replicas: 1
|
|
strategy:
|
|
type: Recreate
|
|
selector:
|
|
matchLabels:
|
|
app: ollama
|
|
template:
|
|
metadata:
|
|
labels:
|
|
app: ollama
|
|
spec:
|
|
runtimeClassName: nvidia
|
|
containers:
|
|
- name: ollama
|
|
image: ollama/ollama
|
|
ports:
|
|
- containerPort: 11434
|
|
name: http
|
|
env:
|
|
- name: OLLAMA_MODELS
|
|
value: /models
|
|
- name: OLLAMA_HOST
|
|
value: "0.0.0.0:11434"
|
|
- name: OLLAMA_MAX_LOADED_MODELS
|
|
value: "1"
|
|
- name: OLLAMA_NUM_PARALLEL
|
|
value: "1"
|
|
volumeMounts:
|
|
- name: models
|
|
mountPath: /models
|
|
resources:
|
|
requests:
|
|
memory: "512Mi"
|
|
cpu: "500m"
|
|
limits:
|
|
memory: "16Gi"
|
|
cpu: "4000m"
|
|
nvidia.com/gpu: "1"
|
|
livenessProbe:
|
|
httpGet:
|
|
path: /api/tags
|
|
port: 11434
|
|
initialDelaySeconds: 30
|
|
periodSeconds: 30
|
|
readinessProbe:
|
|
httpGet:
|
|
path: /api/tags
|
|
port: 11434
|
|
initialDelaySeconds: 10
|
|
periodSeconds: 10
|
|
- name: model-sync
|
|
image: ollama/ollama
|
|
command: ["/bin/bash", "/scripts/sync-models.sh"]
|
|
env:
|
|
- name: MODEL_LIST
|
|
value: /config/models.txt
|
|
- name: OLLAMA_HOST
|
|
value: "http://localhost:11434"
|
|
volumeMounts:
|
|
- name: models-config
|
|
mountPath: /config
|
|
- name: sync-script
|
|
mountPath: /scripts
|
|
resources:
|
|
requests:
|
|
memory: "64Mi"
|
|
cpu: "50m"
|
|
limits:
|
|
memory: "256Mi"
|
|
cpu: "200m"
|
|
volumes:
|
|
- name: models
|
|
persistentVolumeClaim:
|
|
claimName: ollama-models
|
|
- name: models-config
|
|
configMap:
|
|
name: ollama-models
|
|
- name: sync-script
|
|
configMap:
|
|
name: ollama-sync-script
|
|
defaultMode: 0755 # yamllint disable-line rule:octal-values
|