Files
tiny-forge/internal/staticsite/healthcheck.go
T
alexei.dolgolyov 8d2c5a063b feat: static sites feature with Gitea/GitHub/GitLab support and Deno backend
Deploy static content from Git repository folders with optional server-side
API endpoints. Supports Gitea/Forgejo/Gogs, GitHub, and GitLab with provider
autodetection.

- New Sites entity with CRUD, encrypted secrets, and manual/push/tag sync triggers
- Pluggable GitProvider interface with three implementations
- Deno container mode: auto-generates router from API_{method}_{name} exports
- Static container mode: nginx serving files with optional markdown rendering
- Wizard UI with provider selector, repo picker, branch/folder tree pickers
- Deploy pipeline builds fresh image, starts container, configures NPM proxy
- Stop/Start buttons, force redeploy on manual trigger
- Periodic health checker detects crashed containers
- Proxy route existence check during auto-sync
2026-04-11 03:35:57 +03:00

112 lines
2.8 KiB
Go

package staticsite
import (
"context"
"fmt"
"log/slog"
"sync"
"time"
"github.com/alexei/docker-watcher/internal/docker"
"github.com/alexei/docker-watcher/internal/store"
"github.com/robfig/cron/v3"
)
// HealthChecker periodically checks that deployed static site containers
// are still running. If a container has crashed, it updates the site status
// to "failed" and optionally triggers a redeploy.
type HealthChecker struct {
store *store.Store
docker *docker.Client
manager *Manager
cron *cron.Cron
mu sync.Mutex
entryID cron.EntryID
running bool
}
// NewHealthChecker creates a new static site health checker.
func NewHealthChecker(st *store.Store, dockerClient *docker.Client, mgr *Manager) *HealthChecker {
return &HealthChecker{
store: st,
docker: dockerClient,
manager: mgr,
cron: cron.New(),
}
}
// Start begins the periodic health check with the given interval (e.g., "5m", "1m").
func (h *HealthChecker) Start(interval string) error {
h.mu.Lock()
defer h.mu.Unlock()
duration, err := time.ParseDuration(interval)
if err != nil {
return fmt.Errorf("parse interval %q: %w", interval, err)
}
if h.running {
h.cron.Remove(h.entryID)
}
spec := fmt.Sprintf("@every %s", duration)
id, err := h.cron.AddFunc(spec, h.check)
if err != nil {
return fmt.Errorf("schedule health check: %w", err)
}
h.entryID = id
h.running = true
h.cron.Start()
slog.Info("static site health checker started", "interval", interval)
return nil
}
// Stop stops the periodic health checker.
func (h *HealthChecker) Stop() {
h.mu.Lock()
defer h.mu.Unlock()
if h.running {
h.cron.Stop()
h.running = false
slog.Info("static site health checker stopped")
}
}
// check runs a single health check pass over all deployed static sites.
func (h *HealthChecker) check() {
sites, err := h.store.GetAllStaticSites()
if err != nil {
slog.Error("static site health check: failed to list sites", "error", err)
return
}
ctx := context.Background()
for _, site := range sites {
if site.Status != "deployed" || site.ContainerID == "" {
continue
}
running, err := h.docker.IsContainerRunning(ctx, site.ContainerID)
if err != nil {
// Container might have been removed externally.
slog.Warn("static site health check: container inspect failed",
"site", site.Name, "container", site.ContainerID[:12], "error", err)
h.manager.updateStatus(site.ID, "failed", site.LastCommitSHA, "container not found")
h.manager.publishEvent(site.ID, site.Name, "failed: container not found")
continue
}
if !running {
slog.Warn("static site health check: container not running",
"site", site.Name, "container", site.ContainerID[:12])
h.manager.updateStatus(site.ID, "failed", site.LastCommitSHA, "container stopped unexpectedly")
h.manager.publishEvent(site.ID, site.Name, "failed: container stopped unexpectedly")
}
}
}