Support for initializing models using a Helm hook job instead of a lifecycle postStart command #200
Description
Is your feature request related to a problem? Please describe
Currently, models are initialized using a pod lifecycle postStart event. This has two downsides: 1) the pod may appear unhealthy if downloading the models takes a long time, 2) it's difficult to debug since one can't access the log output of the event.
Describe the solution you'd like
An alternative implementation could us a Kubernetes job to initialize the ollama host "remotely": this job could be configured to run as a Helm post-install, post-upgrade hook, using the OLLAMA_HOST setting to connect to the running ollama server and should be able to run the same commands as in https://github.com/otwld/ollama-helm/blob/main/templates/deployment.yaml#L174-L229
Describe alternatives you've considered
No response
Additional context
I'd be happy to create a PR for this if it seems useful!