Models not initializing

[dempo93]

What happened?

Ollama starts correctly but if I shell into the container I cannot see any of the downloaded/customized/running models.

Possible solutions:

• Support for initializing models using a Helm hook job instead of a lifecycle postStart command #200
• Improve observability of the lifecycle event by redirecting its output to docker log/stderr/observable file
• Consider pulling models inside a blocking initContainer

Chart version

1.3.2

Kubernetes version

1.32

Kubernetes distribution

Eks

Relevant log output

Not available

Your values.yaml

ollama:
  gpu:
    enabled: true
  models:
    pull:
      - gemma3:27b
    run:
      - gemma3:27b
  tolerations:
  - key: "dedicated"
    operator: "Equal"
    value: "nvidia"
    effect: "NoSchedule"
  nodeSelector:
    node.kubernetes.io/instance-type: g6e.xlarge
  extraEnv:
    - name: OLLAMA_KEEP_ALIVE
      value: "9999h"
    - name: OLLAMA_CONTEXT_LENGTH
      value: "12288"

Additional context