deploy-agents

To deploy agents to OpenShift and refresh MLflow tokens, you can automate the process using Helm-based rollout, which validates prerequisites, writes per-agent .env files, patches mlflow-tracking-token secrets, restarts deployments, and verifies tracing readiness via startup logs.

No, manual configuration is not needed. The deployment process auto-detects cluster configuration by extracting BASE_URL, model details, MLflow settings, and registry prefix directly from existing standard agent deployments in the namespace, avoiding redundant user input.

An OpenShift agent rollout requires the oc CLI, helm, and a container CLI to be installed and accessible. The deployment workflow validates these prerequisites before resolving target agents, building images, or initiating the Helm-based deployment.

MLflow token refresh updates the mlflow-tracking-token secret for all applicable agents in the namespace simultaneously. It then restarts affected deployments, verifies secretKeyRef wiring, and confirms MLflow connectivity and tracing readiness by checking startup logs.

OpenShift agent deployment health checks can fail if secretKeyRef wiring is incorrect or MLflow tracing is not ready. The deployment process verifies runtime health via the route /health endpoint and checks startup logs to confirm valid MLflow connectivity.

Yes, you can deploy all standard agents at once by specifying the target as 'all'. The workflow resolves target agents, filters out non-standard ones, writes per-agent .env files safely without logging secrets, and performs a validated Helm-based deployment.