Eero Tamminen

> @mythi `no-install-recommends` makes building fail. For making it build without fail, we need to add more. Is it what we want? They are anyway in `builder` image that will...

@poussa can comment more on certificate use on K8s side. As to other items... * Neither host networking nor NICE capability are needed for Kubernetes, so I would assume them...

> set return_full_text=False to only get generated text (and not the input + generated text). Would that break things if used also with e.g. Qwen?

@Serverfrog Could you paste here: * pod spec, at least following sections: `nodename`/`nodeSelector`, `securityContext` (both for pod & container), `resources`? * node k8s GPU info: `kubectl describe node YOUR_NODE_NAME |...

> @tkatila GAS was enabled as a test from me, after it was already not working, that it maybe that. For example it worked shortly for one pod with using...

> Which also explained why, in privileged mode where always on the iGPU (but it could also be that i could not configured ffmpeg through that interface correctly to use...

I don't think documentation is enough. Currently model downloading is done by each container separately when they start, and those services having write access to that volume. Meaning that user/admin...

Helm charts are already using HF downloader in initContainers: https://github.com/opea-project/GenAIInfra/blob/main/helm-charts/common/vllm/templates/deployment.yaml#L53 There could be a separate script / container using that, which would download all specified models to a location expected...

> [@Yu-amd](https://github.com/Yu-amd) [@eero-t](https://github.com/eero-t) [@mkbhanda](https://github.com/mkbhanda) thank a lot for raising the problem. Just remind if there is any update OIM could help in pre-downloading models, but its RFC does not specifically...

> Besides the model, there are many other OPEA microservices are downloading data from internet during runtime. We need to fix them 1 by 1: [opea-project/GenAIComps#1480](https://github.com/opea-project/GenAIComps/issues/1480) @lianhao Thanks for investigating...