Files
neuron-web/dist/k3s-soul-demo.yaml
will.anderson c6ee45a374
Dev — Build & local smoke test / build-smoke (pull_request) Failing after 3m54s
fix: run k3s as root, bump HPA CPU threshold to 80%
k3s needs CAP_SYS_ADMIN to create network namespaces and mount cgroups.
USER landing was preventing this. Cloud Run gen2 is the security boundary.

60% CPU was too conservative for soul-demo — it is I/O-bound (LLM API calls),
not CPU-bound. 80% gives correct headroom before scaling kicks in.
2026-05-09 12:40:27 -05:00

91 lines
1.7 KiB
YAML

---
apiVersion: apps/v1
kind: Deployment
metadata:
name: soul-demo
namespace: default
labels:
app: soul-demo
spec:
replicas: 1
selector:
matchLabels:
app: soul-demo
template:
metadata:
labels:
app: soul-demo
spec:
containers:
- name: soul-demo
image: soul-demo:local
imagePullPolicy: Never
ports:
- containerPort: 7772
env:
- name: NEURON_HOME
value: /srv/soul/engram-demo
- name: NEURON_PORT
value: "7772"
resources:
requests:
cpu: 250m
memory: 256Mi
limits:
cpu: 1000m
memory: 512Mi
livenessProbe:
httpGet:
path: /healthz
port: 7772
initialDelaySeconds: 10
periodSeconds: 15
failureThreshold: 3
readinessProbe:
httpGet:
path: /healthz
port: 7772
initialDelaySeconds: 5
periodSeconds: 10
volumeMounts:
- name: engram-data
mountPath: /srv/soul/engram-demo
volumes:
- name: engram-data
emptyDir: {}
---
apiVersion: v1
kind: Service
metadata:
name: soul-demo
namespace: default
spec:
type: NodePort
selector:
app: soul-demo
ports:
- port: 7772
targetPort: 7772
nodePort: 7772
protocol: TCP
---
apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
name: soul-demo
namespace: default
spec:
scaleTargetRef:
apiVersion: apps/v1
kind: Deployment
name: soul-demo
minReplicas: 1
maxReplicas: 8
metrics:
- type: Resource
resource:
name: cpu
target:
type: Utilization
averageUtilization: 80