Docker container management (start/stop/restart/logs)

[LiukScot]

Summary

Add the ability to start, stop, restart containers and view live container logs from the dashboard.

Problem

The Docker collector is currently read-only. Users must SSH into the server to manage containers or view their logs.

Implementation

Backend

• New API endpoints:
  • POST /api/v1/docker/containers/:id/start
  • POST /api/v1/docker/containers/:id/stop
  • POST /api/v1/docker/containers/:id/restart
  • GET /api/v1/docker/containers/:id — full container inspect (env, ports, volumes, resource limits)
  • GET /api/v1/docker/containers/:id/logs?tail=200&follow=true — SSE or WebSocket stream for live logs
• Extend internal/collectors/docker.go with action methods using Docker API (/containers/:id/start, /stop, /restart, /json, /logs)

Frontend

• Container row click opens a detail panel/drawer
• Detail panel shows: full config, env vars, port mappings, volumes, resource limits
• Action buttons: Start, Stop, Restart with confirmation dialog
• Live log viewer: scrollable, auto-follow, with search/filter
• Color-coded log output (stderr in red)

Security consideration

Container management is a privileged operation. Should require admin role (ties into #8 user management).

Files to create/modify

• internal/collectors/docker.go — action methods
• internal/server/server.go — new routes
• New frontend/src/components/ContainerDetail.svelte
• frontend/src/components/ContainerTable.svelte — make rows clickable

[claude]

Triage automatico (reopen)

Feature request: estendere il collector Docker da read-only ad amministrabile (start/stop/restart/inspect/logs streaming) con un pannello di dettaglio nel frontend.

Stato attuale del codebase: il collector internal/collectors/docker.go espone solo ListContainers e GetContainerStats; in internal/server/server.go è registrata solo la rotta GET /api/v1/docker/containers. Nessun endpoint di azione o di logs è presente, e ContainerDetail.svelte non esiste.

Categoria: enhancement · Priorità: medium.

Nota di scope: la sicurezza richiamata nell'issue (admin role) dipende da #8. Conviene chiarire se l'implementazione di questi endpoint debba attendere il sistema di ruoli o adottare un gating temporaneo (es. flag di config / token admin) per non bloccare il lavoro frontend/backend.