felhom-controller/controller/internal/web/backup_handlers.go

package web

import (
	"context"
	"errors"
	"net/http"
	"strings"
	"time"

	"gitea.dooplex.hu/admin/felhom-controller/internal/agentapi"
	"gitea.dooplex.hu/admin/felhom-controller/internal/quiesce"
)

// Whole-guest backup visibility + manual trigger (spec Part 2). The agent owns whole-guest
// vzdump/PBS backup; the controller is a read-only window onto it (GET /backup/{status,due},
// /restore-test/status) plus a "Mentés most" trigger that goes through the quiesce loop (the
// CONTROLLER owns quiescing — stop stacks → POST /backup → resume — so the captured state is
// app-consistent, not the agent's crash-consistent default). Cadence/retention CONFIG is NOT here
// (hub-served policy, slice 10).

// guestBackupView is the template payload for the "Rendszermentés" section. Times are time.Time so
// the existing fmtTime/timeAgo funcmap helpers format them; size is int64 for fmtBytes.
type guestBackupView struct {
	Available bool   // agent reachable + a status read succeeded
	Note      string // shown when not Available (unprovisioned / unreachable)

	Phase   string // idle | running | snapshotted | done | failed
	Running bool   // a backup job is in progress now

	HasBackup bool
	Success   bool
	StartedAt time.Time
	SizeBytes int64
	Target    string // human label: "Biztonsági szerver – külön hardver (PBS)" / "Helyi tároló (local)"
	Offsite   bool   // the whole-guest backup landed on the PBS offsite tier (separate hardware)
	Archive   string
	Mode      string // snapshot | stop
	StopMode  bool   // mode == stop → full app downtime during the backup (warn)

	Due       bool
	DueReason string
	AgeHours  int64 // age of the newest successful backup, hours (for "X órája")

	HasRestoreTest bool
	RestorePass    bool
	RestoreVerified string
	RestoreTestedAt time.Time

	CanTrigger bool // a backup trigger (quiesce loop) is wired
}

// loadGuestBackup fetches the agent's whole-guest backup view (best-effort). Returns a view with
// Available=false (+ a note) when the agent isn't configured/reachable — the page still renders.
func (s *Server) loadGuestBackup(ctx context.Context) *guestBackupView {
	v := &guestBackupView{CanTrigger: s.backupTrigger != nil}
	client, err := s.agentClient()
	if err != nil {
		v.Note = "A host-ügynök nincs konfigurálva ezen a gépen."
		return v
	}
	st, err := client.BackupStatus(ctx)
	if err != nil {
		v.Note = "A host-ügynök jelenleg nem elérhető."
		return v
	}
	v.Available = true
	v.Phase = st.Phase
	v.Running = st.Phase == agentapi.PhaseRunning || st.Phase == "snapshotted"
	if st.Backup != nil {
		v.HasBackup = true
		v.Success = st.Backup.Success
		v.SizeBytes = st.Backup.SizeBytes
		v.Archive = st.Backup.Archive
		v.Mode = st.Backup.Mode
		v.StopMode = st.Backup.Mode == "stop"
		v.Target = backupTargetLabel(st.Backup)
		v.Offsite = backupIsPBS(st.Backup)
		if t, perr := time.Parse(time.RFC3339, st.Backup.StartedAt); perr == nil {
			v.StartedAt = t
		}
	}
	// Due window (best-effort; a failure just leaves the due fields zero).
	if due, derr := client.BackupDue(ctx); derr == nil {
		v.Due = due.Due
		v.DueReason = due.Reason
		if due.AgeSecs != nil {
			v.AgeHours = *due.AgeSecs / 3600
		}
	}
	// Restore-test (the "verified restorable" trust signal; nil until one runs).
	if rt, rerr := client.RestoreTestStatus(ctx); rerr == nil && rt != nil {
		v.HasRestoreTest = true
		v.RestorePass = rt.Pass
		v.RestoreVerified = rt.Verified
		if t, perr := time.Parse(time.RFC3339, rt.TestedAt); perr == nil {
			v.RestoreTestedAt = t
		}
	}
	return v
}

// backupIsPBS reports whether a whole-guest backup landed on the PBS offsite tier (separate
// hardware), inferred from the target id / archive volid ("felhom-pbs"/"pbs:" ⇒ PBS).
func backupIsPBS(b *agentapi.BackupRecord) bool {
	id := strings.ToLower(b.TargetID)
	arc := strings.ToLower(b.Archive)
	return strings.Contains(id, "pbs") || strings.HasPrefix(arc, "felhom-pbs") || strings.Contains(arc, "pbs:")
}

// backupTargetLabel maps the agent's backup target to a customer-facing Hungarian label. The PBS
// case calls out that the backup is on SEPARATE HARDWARE (real disaster recovery — survives a host
// disk/hardware failure), which is the whole point of re-pointing the backup offsite.
func backupTargetLabel(b *agentapi.BackupRecord) string {
	if backupIsPBS(b) {
		return "Biztonsági szerver – külön hardver (PBS)"
	}
	if b.TargetID != "" {
		return "Helyi tároló (" + b.TargetID + ")"
	}
	return "Helyi tároló"
}

// ServeBackupAPI dispatches /api/guest-backup/* (whole-guest manual trigger + status poll). A
// distinct prefix from apiRouter's app-data /api/backup/{run,status}. Wired behind RequireAuth +
// CsrfProtect in main.go.
func (s *Server) ServeBackupAPI(w http.ResponseWriter, r *http.Request) {
	switch {
	case r.URL.Path == "/api/guest-backup/trigger" && r.Method == http.MethodPost:
		s.handleBackupTriggerAPI(w, r)
	case r.URL.Path == "/api/guest-backup/status" && r.Method == http.MethodGet:
		s.handleBackupStatusAPI(w, r)
	default:
		http.NotFound(w, r)
	}
}

// handleBackupTriggerAPI starts an app-consistent whole-guest backup NOW via the quiesce loop. It
// returns immediately (the backup runs async, minutes); the page polls /api/backup/status.
func (s *Server) handleBackupTriggerAPI(w http.ResponseWriter, r *http.Request) {
	if s.backupTrigger == nil {
		writeDiskJSON(w, http.StatusServiceUnavailable, false, "a rendszermentés nem érhető el ezen a gépen", nil)
		return
	}
	if err := s.backupTrigger.TriggerNow(); err != nil {
		if errors.Is(err, quiesce.ErrBackupInProgress) {
			writeDiskJSON(w, http.StatusConflict, false, "mentés már folyamatban van", nil)
			return
		}
		s.logger.Printf("[ERROR] [web] backup trigger failed: %v", err)
		writeDiskJSON(w, http.StatusBadGateway, false, err.Error(), nil)
		return
	}
	s.logger.Printf("[INFO] [web] manual whole-guest backup triggered (quiesce loop)")
	writeDiskJSON(w, http.StatusOK, true, "", map[string]any{"started": true})
}

// handleBackupStatusAPI proxies the agent's GET /backup/status for the page's progress poll.
func (s *Server) handleBackupStatusAPI(w http.ResponseWriter, r *http.Request) {
	client, err := s.agentClient()
	if err != nil {
		writeDiskJSON(w, http.StatusServiceUnavailable, false, err.Error(), nil)
		return
	}
	st, err := client.BackupStatus(r.Context())
	if err != nil {
		writeDiskJSON(w, http.StatusBadGateway, false, err.Error(), nil)
		return
	}
	writeDiskJSON(w, http.StatusOK, true, "", map[string]any{
		"phase": st.Phase, "job_id": st.JobID, "error": st.Error, "backup": st.Backup,
	})
}