Install custom artifact bundles

Prerequisites

Ensure the model bundle .tar file is available on the server (the support team can provide this file).
Ensure the gv-model-deployer chart is available on the server (included in the gv-platform-<version>.tar installation package).

Steps:

Connect to server:
```
# ssh.exe airgap
```

Execute the below commands:

# BUNDLE=region.customer_name-3.17.3
# gunzip models_$BUNDLE.tar.gz
# sudo /usr/local/bin/ctr -n=k8s.io images import models_$BUNDLE.tar

Output:

unpacking images.master.k3s.getvisibility.com/models:region.customer_name-3.17.3 
(sha256:181f5622f978443355a8d8c7090e38e41f2b1c0057ba12ba20e843d4c33d1ee9)...done

Execute the below commands:

# helm upgrade \
--debug --install gv-model-deployer gv-model-deployer-0.0.1.tgz \
--wait --timeout=10m0s --kubeconfig /etc/rancher/k3s/k3s.yaml \
--set models.version="$BUNDLE"
# kubectl logs $(kubectl get pods -A | grep ml-model | awk '{print $2}')

Note:

Make sure to wipe all detectors and import the new ones if your platform version has not done this automatically.
If you also need structured data models, the air‑gapped gv-model-deployer does not support them; manually upload them to MinIO via a temporary pod.