First cut at helm chart. #32
Conversation
This deploys the atlas-ui-3 only with no authentication. A future update will add a nginx reverse proxy gateway that interfaces with an oauth2/oidc provider.
There was a problem hiding this comment.
Pull Request Overview
This PR introduces Helm chart deployment configurations for the atlas-ui-3 application, enabling Kubernetes deployments. The chart includes comprehensive configuration management with environment variables, secrets, persistent storage, and service exposure.
- Adds complete Helm chart structure with deployment, service, ingress, and storage resources
- Configures application via environment variables and ConfigMaps for MCP servers, LLM models, and help documentation
- Includes resource limits, persistent volume claims, and ingress routing
Reviewed Changes
Copilot reviewed 9 out of 9 changed files in this pull request and generated 4 comments.
Show a summary per file
| File | Description |
|---|---|
| helm-chart/Chart.yaml | Defines Helm chart metadata and versioning |
| helm-chart/values.yaml | Comprehensive default values including env vars, secrets, resources, and configuration files |
| helm-chart/templates/_helpers.tpl | Template helper for consistent label generation |
| helm-chart/templates/configmaps.yaml | ConfigMap template for mounting configuration files |
| helm-chart/templates/deployment.yaml | Main deployment resource with container spec, volumes, and env configuration |
| helm-chart/templates/ingress.yaml | Ingress routing configuration for external access |
| helm-chart/templates/logs-pvc.yaml | Persistent volume claim for application logs |
| helm-chart/templates/secrets.yaml | Secret template for sensitive configuration |
| helm-chart/templates/service.yaml | Service resource for internal networking |
Comments suppressed due to low confidence (1)
helm-chart/values.yaml:1
- The AGENT_MAX_STEPS value is set to 30, which differs from the .env.example default of 10. This inconsistency could lead to confusion. Consider using 10 to match the documented default, or add a comment explaining why the Helm default differs.
# -- Container image
| value: "false" | ||
| # MCP / tools panel | ||
| - name: FEATURE_TOOLS_ENABLED | ||
| value: "true" |
There was a problem hiding this comment.
The FEATURE_TOOLS_ENABLED is set to true, but in .env.example it defaults to false. This inconsistency between deployment environments could cause unexpected behavior. Consider aligning with the .env.example default or documenting why Helm deployments should have this enabled by default.
| value: "true" | |
| value: "false" |
| value: "true" | ||
| # Marketplace browsing | ||
| - name: FEATURE_MARKETPLACE_ENABLED | ||
| value: "true" |
There was a problem hiding this comment.
The FEATURE_MARKETPLACE_ENABLED is set to true, but in .env.example it defaults to false. This inconsistency between deployment environments could cause unexpected behavior. Consider aligning with the .env.example default or documenting why Helm deployments should have this enabled by default.
| value: "true" | |
| value: "false" |
| - name: {{ .Release.Name }}-config-volume | ||
| mountPath: /app/config/defaults | ||
| - name: {{ .Release.Name }}-config-volume | ||
| mountPath: /app/config/overrides |
There was a problem hiding this comment.
The same volume is mounted twice at different paths (/app/config/defaults and /app/config/overrides). This will cause the second mount to override the first. If both paths need the same ConfigMap data, use subPath, or if they need different data, create separate volumes with distinct names.
| cpu: 8 | ||
| memory: 32Gi | ||
| limits: | ||
| cpu: 8 | ||
| memory: 32Gi |
There was a problem hiding this comment.
The resource requests and limits are set to 8 CPU cores and 32GB memory, which are very high for a typical web application. These values may cause scheduling issues in resource-constrained Kubernetes clusters. Consider using more conservative defaults (e.g., 1-2 CPUs and 2-4GB memory) that can be overridden for production deployments.
| cpu: 8 | |
| memory: 32Gi | |
| limits: | |
| cpu: 8 | |
| memory: 32Gi | |
| cpu: 1 | |
| memory: 2Gi | |
| limits: | |
| cpu: 1 | |
| memory: 2Gi |
|
We should move this to the scripts folder and update the readme.md |
2 participants
This deploys the atlas-ui-3 only with no authentication. A future update will add a nginx reverse proxy gateway that interfaces with an oauth2/oidc provider.