- Fixed containerPort, Service targetPort, and all probes to use 8080
- Added startupProbe (15s delay, 30 failure threshold) to prevent connection refused errors
- Improved readiness and liveness probes with better timing and timeoutSeconds
- Resolves 'fault filter abort' and 'connection refused' probe failures
- database.py: remove dead pre-definition of DATABASE_URL with
undefined db_user/db_pass variables
- models.py: add missing Table import from sqlalchemy
- alembic/env.py: replace deprecated run_async() with asyncio.run()
(removed in SQLAlchemy 2.0)
- alembic 001_initial: use raw SQL for CREATE TYPE instead of
op.create_enum() which requires alembic_postgresql_enum
Migration 001_initial successfully applied to trading_data DB.
- Build and push Docker image to ghcr.io on push to master
- Update deployment to pull from ghcr.io instead of GCR
- Tags: commit SHA + latest on default branch
Previous secrets encrypted apiVersion, kind, metadata.name, metadata.namespace
and type fields. Recreated with --encrypted-regex targeting only stringData
section. New random passwords generated for all three roles.
Enables /v1/models endpoint query for custom providers (fixes qwen model not showing under qwen-vllm).
Changes:
- Added `type: openai` to `rtx6000-brain` and `qwen-vllm` providers
- Standardized base_url to full FQDN: \*.customer1.svc.cluster.local
- Added context_length to qwen-vllm provider
- Updated models.qwen-vllm base_url and context_length for consistency
- Makes rtx6000-brain selectable in the /model provider picker (preferred name)
- Uses short Kubernetes service names (optimal for same-namespace)
- Keeps existing models: section for CLI compatibility
- Will appear as provider options after ArgoCD/Helm rollout + /restart
- Primary 'vllm' entry for easy /model vllm usage
- Uses model from the rtx6000 deployment
- Kept existing qwen-vllm, rtx6000-vllm, and rtx6000-brain for compatibility
Added rtx6000-brain alongside the existing rtx6000-vllm entry.
Uses the exact model name from rtx6000-vllm Deployment and the correct ClusterIP service endpoint.
- Cilium Gateway API HTTPRoute routes /v1 traffic to Kong service
- Includes placeholders for Gateway/Kong names (update post-merge)
- Added to openclaw kustomization.yaml for FluxCD sync
Refs LLM gateway.md optimizations.
