What is the Infra console?
The Infra console is the second entry on the DevOpsPad navigation rail, next to 中间件 (Middleware). It groups a Kubernetes or K3s cluster by environment and application instead of by raw resource type, so you can see a service’s artifacts, releases, Deployments, Pods, logs, metrics, events and configuration in one place.
Database and cluster browsing elsewhere in DevOpsPad is read-only. The Infra console is the one place that can change a cluster, and every change goes through a reviewed plan: you see the current state and the target, DevOpsPad runs a server-side dry-run, and nothing is written until you confirm.
How it is organized
Pick an environment and an application in the top-left selectors. The sidebar then shows four groups (the app shows the Chinese labels):
| Group | Pages |
|---|---|
| 集群 (Cluster) | 节点 / 机器 (Nodes / Machines) |
| 部署 (Deploy) | AI 执行记录 (AI run records), 制品库 (Artifacts), 制品发布 (Releases), Deployment, Pod 实例 (Pods) |
| 可观测 (Observe) | 监控 (Monitoring), 日志 (Logs), 应用事件 (Application events) |
| 配置 (Configure) | 应用配置 (Application config), Nginx 配置 (Nginx config), 项目映射 (Project mapping), 共享与同步 (Sharing & sync) |
The header also has 集群资源 (Cluster resources), which opens the regular Kubernetes/K3s connection list and resource queries, and 接入 K3s (Connect K3s), the wizard that maps workloads into the project. A theme button switches the console between dark and light.
Where the data comes from
- Cluster state (nodes, Deployments, Pods, logs, events, metrics, ConfigMaps) is read live from the Kubernetes API through a saved connection. See kubeconfig contexts and SSH tunnels.
- Project data (environments, applications, artifacts, release records and AI run records) lives in a local project directory as YAML and JSON files. You can share it with a small team through Git. See project.yaml.
- Credentials never go into the project directory. Each person keeps their own kubeconfig and keychain secrets; the project only stores connection names.
DevOpsPad remembers the last workspace and the Infra environment, application and page you used.
What it can change
| Action | Target | Help |
|---|---|---|
| Release an image | One container of an existing Deployment | Release an image |
| Scale | Deployment replicas, 0–1,000 | Scale and restart |
| Restart | Rolling restart of a Deployment | Scale and restart |
| Roll back | A container image, to a previous successful release | Roll back a release |
| Edit a ConfigMap | Text keys of a referenced ConfigMap | Application config |
| Edit host Nginx | One .conf file under /etc/nginx/conf.d/ over SSH |
Application config |
StatefulSets and DaemonSets are listed and can be inspected, but they cannot be modified from the Infra console.
Limits
- Lists are capped at 500 resources; the page tells you to narrow the selector if the cap is reached.
- Monitoring shows current CPU and memory snapshots from metrics-server. There is no history.
- Logs are bounded snapshots (up to 2,000 lines / 256 KiB), not a live stream.
- Image builds and registry pushes are out of scope. DevOpsPad releases images that already exist in your registry.
Frequently asked questions
Does the Infra console need a server or an agent in my cluster?
No. It talks to the Kubernetes API with the kubeconfig and optional SSH tunnel of a saved connection. Project files live in a local directory on your machine.
Can the Infra console create or delete workloads?
No. It can release a new image to an existing Deployment, scale it, restart it and roll back an image. Creating and deleting workloads stays with your existing cluster tooling.