Skip to content

armgit / Pulse

Public Imported from GitHub
0

Real-time monitoring dashboard for Proxmox VE, PBS, Docker, Kubernetes, TrueNAS and vSphere. Self-hosted, with alerts for failed backups, full disks, unhealthy containers and offline nodes.

13.4 KB
Raw Blame
package monitoring

import (
	"context"
	"fmt"
	"strings"
	"sync"
	"time"

	"github.com/rcourtman/pulse-go-rewrite/internal/alerts"
	"github.com/rcourtman/pulse-go-rewrite/internal/config"
	"github.com/rcourtman/pulse-go-rewrite/internal/unifiedresources"
	"github.com/rcourtman/pulse-go-rewrite/internal/websocket"
	"github.com/rs/zerolog/log"
)

// InstallSnapshotCounts holds install-wide resource and alert counts aggregated
// across tenant monitors.
type InstallSnapshotCounts struct {
	PVENodes                             int
	PBSInstances                         int
	PMGInstances                         int
	VMs                                  int
	Containers                           int
	AgentHosts                           int
	DockerHosts                          int
	DockerContainers                     int
	KubernetesClusters                   int
	KubernetesNodes                      int
	KubernetesPods                       int
	KubernetesDeployments                int
	StoragePools                         int
	PhysicalDisks                        int
	CephClusters                         int
	NetworkShares                        int
	TrueNASSystems                       int
	TrueNASVMs                           int
	TrueNASApps                          int
	VMwareHosts                          int
	VMwareVMs                            int
	VMwareDatastores                     int
	AvailabilityTargets                  int
	ActiveAlerts                         int
	AlertsFired30d                       int
	AlertsAcknowledged30d                int
	AlertsResolved30d                    int
	NotificationAttempts7d               int
	NotificationDeliveries7d             int
	NotificationFailures7d               int
	NotificationFailuresAuthentication7d int
	NotificationFailuresRateLimited7d    int
	NotificationFailuresConnectivity7d   int
	NotificationFailuresTLS7d            int
	NotificationFailuresConfiguration7d  int
	NotificationFailuresRejected7d       int
	NotificationFailuresServerError7d    int
	NotificationFailuresUnknown7d        int
	AlertQuality                         alerts.AlertQualitySnapshot
}

// ReloadableMonitor wraps a Monitor with reload capability
type ReloadableMonitor struct {
	mu          sync.RWMutex
	mtMonitor   *MultiTenantMonitor
	persistence *config.MultiTenantPersistence
	config      *config.Config
	wsHub       *websocket.Hub
	ctx         context.Context
	cancel      context.CancelFunc
	parentCtx   context.Context
	reloadChan  chan chan error
}

// NewReloadableMonitor creates a new reloadable monitor
func NewReloadableMonitor(cfg *config.Config, persistence *config.MultiTenantPersistence, wsHub *websocket.Hub) (*ReloadableMonitor, error) {
	mtMonitor := NewMultiTenantMonitor(cfg, persistence, wsHub)
	// No error check needed for NewMultiTenantMonitor as it doesn't return error

	rm := &ReloadableMonitor{
		mtMonitor:   mtMonitor,
		config:      cfg,
		persistence: persistence,
		wsHub:       wsHub,
		reloadChan:  make(chan chan error, 1),
	}

	return rm, nil
}

// Start starts the monitor with reload capability
func (rm *ReloadableMonitor) Start(ctx context.Context) {
	rm.mu.Lock()
	rm.parentCtx = ctx
	rm.ctx, rm.cancel = context.WithCancel(ctx)
	rm.mu.Unlock()

	// Start the multi-tenant monitor manager
	// Note: It doesn't start individual monitors until requested via GetMonitor()
	// But we might want to start "default" monitor if it exists?
	// For now, lazy loading handles it.

	// Watch for reload signals
	go rm.watchReload(ctx)
}

// Reload triggers a monitor reload
func (rm *ReloadableMonitor) Reload() error {
	rm.mu.RLock()
	ctx := rm.parentCtx
	rm.mu.RUnlock()
	if ctx == nil {
		return fmt.Errorf("reloadable monitor has not been started")
	}
	if err := ctx.Err(); err != nil {
		return err
	}

	// The watcher exits with the server context. Neither a full request
	// channel nor an abandoned reply may keep a caller waiting after that.
	// Keep the reply buffered so an in-flight reload can still finish.
	done := make(chan error, 1)
	select {
	case rm.reloadChan <- done:
	case <-ctx.Done():
		return ctx.Err()
	}
	select {
	case err := <-done:
		return err
	case <-ctx.Done():
		return ctx.Err()
	}
}

// watchReload watches for reload signals
func (rm *ReloadableMonitor) watchReload(ctx context.Context) {
	for {
		select {
		case <-ctx.Done():
			return
		case done := <-rm.reloadChan:
			log.Info().Msg("reloading monitor configuration")
			if err := rm.doReload(); err != nil {
				log.Error().Err(err).Msg("failed to reload monitor")
				done <- err
			} else {
				log.Info().Msg("monitor reloaded successfully")
				done <- nil
			}
		}
	}
}

// doReload performs the actual reload
func (rm *ReloadableMonitor) doReload() error {
	rm.mu.Lock()
	defer rm.mu.Unlock()

	// Load fresh configuration
	cfg, err := config.LoadWithoutLoggingInit()
	if err != nil {
		return fmt.Errorf("reloadable monitor reload config: %w", err)
	}

	// Polling interval changes and other settings require a full reload
	log.Info().Msg("performing full monitor reload")

	// Cancel current monitor
	if rm.cancel != nil {
		rm.cancel()
	}

	// Stop the underlying multi-tenant monitor to ensure its global context is canceled
	// and all child monitors are stopped.
	if rm.mtMonitor != nil {
		rm.mtMonitor.Stop()
	}

	// Wait a moment for cleanup
	time.Sleep(1 * time.Second)

	// Authentication middleware and API handlers retain the original runtime
	// configuration. Refresh that object so the replacement default monitor also
	// sees later token creation, rotation and revocation through the same owner.
	config.Mu.Lock()
	if rm.config == nil {
		rm.config = cfg
	} else {
		*rm.config = *cfg
	}
	config.Mu.Unlock()

	// Create new multi-tenant monitor
	// Note: We lose existing instances state here, which is expected on full reload.
	newMTMonitor := NewMultiTenantMonitor(rm.config, rm.persistence, rm.wsHub)

	// Replace monitor
	rm.mtMonitor = newMTMonitor

	// Start new monitor context (individual monitors are lazy loaded/started)
	rm.ctx, rm.cancel = context.WithCancel(rm.parentCtx)

	return nil
}

// GetMultiTenantMonitor returns the current multi-tenant monitor instance
func (rm *ReloadableMonitor) GetMultiTenantMonitor() *MultiTenantMonitor {
	rm.mu.RLock()
	defer rm.mu.RUnlock()
	return rm.mtMonitor
}

// GetMonitor returns the default monitor instance (compatibility shim)
// It ensures the "default" tenant is initialized.
func (rm *ReloadableMonitor) GetMonitor() *Monitor {
	rm.mu.RLock()
	defer rm.mu.RUnlock()
	if rm.mtMonitor == nil {
		return nil
	}
	monitor, err := rm.mtMonitor.GetMonitor("default")
	if err != nil {
		log.Debug().Err(err).Msg("Default monitor unavailable")
		return nil
	}
	return monitor
}

// GetConfig returns the current configuration used by the monitor.
func (rm *ReloadableMonitor) GetConfig() *config.Config {
	rm.mu.RLock()
	defer rm.mu.RUnlock()
	if rm.config == nil {
		return nil
	}
	return rm.config
}

// ReadSnapshot returns the current state for a specific tenant.
func (rm *ReloadableMonitor) ReadSnapshot(orgID string) interface{} {
	if orgID == "" {
		orgID = "default"
	}
	mtMonitor := rm.GetMultiTenantMonitor()
	if mtMonitor == nil {
		log.Debug().Str("orgID", orgID).Msg("ReadSnapshot requested with no active multi-tenant monitor")
		return nil
	}

	monitor, err := mtMonitor.GetMonitor(orgID)
	if err != nil {
		log.Debug().Err(err).Str("orgID", orgID).Msg("ReadSnapshot monitor unavailable")
		return nil
	}
	return monitor.ReadSnapshot()
}

// AggregateInstallSnapshotCounts returns install-wide resource and alert counts
// across all provisioned organizations.
func (rm *ReloadableMonitor) AggregateInstallSnapshotCounts() InstallSnapshotCounts {
	rm.mu.RLock()
	mtMonitor := rm.mtMonitor
	persistence := rm.persistence
	rm.mu.RUnlock()

	if mtMonitor == nil {
		return InstallSnapshotCounts{}
	}

	orgIDs := []string{"default"}
	if persistence != nil {
		orgs, err := persistence.ListOrganizations()
		if err != nil {
			log.Warn().Err(err).Msg("Telemetry snapshot falling back to default organization after tenant listing failed")
		} else {
			seen := make(map[string]struct{}, len(orgs))
			orgIDs = orgIDs[:0]
			for _, org := range orgs {
				if org == nil {
					continue
				}
				orgID := strings.TrimSpace(org.ID)
				if orgID == "" {
					continue
				}
				if _, exists := seen[orgID]; exists {
					continue
				}
				seen[orgID] = struct{}{}
				orgIDs = append(orgIDs, orgID)
			}
			if len(orgIDs) == 0 {
				orgIDs = []string{"default"}
			}
		}
	}

	var counts InstallSnapshotCounts
	now := time.Now().UTC()
	for _, orgID := range orgIDs {
		monitor, err := mtMonitor.GetMonitor(orgID)
		if err != nil || monitor == nil {
			log.Debug().Err(err).Str("org_id", orgID).Msg("Telemetry snapshot could not load tenant monitor")
			continue
		}
		accumulateInstallSnapshotCounts(&counts, monitor)
		accumulateInstallOutcomeCounts(&counts, monitor, now)
	}
	return counts
}

func accumulateInstallOutcomeCounts(counts *InstallSnapshotCounts, monitor *Monitor, now time.Time) {
	if counts == nil || monitor == nil {
		return
	}
	if now.IsZero() {
		now = time.Now().UTC()
	}
	if alertManager := monitor.GetAlertManager(); alertManager != nil {
		quality := alertManager.AlertQualityTelemetrySnapshot(now)
		counts.AlertQuality.Add(quality)
		counts.AlertsFired30d += quality.Fired30d
		counts.AlertsAcknowledged30d += quality.Acknowledged30d
		counts.AlertsResolved30d += quality.Resolved30d
	}
	if notificationManager := monitor.GetNotificationManager(); notificationManager != nil {
		stats, err := notificationManager.GetTelemetryStats(now.Add(-7 * 24 * time.Hour))
		if err != nil {
			log.Debug().Err(err).Msg("Telemetry snapshot could not read notification delivery aggregates")
			return
		}
		counts.NotificationAttempts7d += stats.Attempts
		counts.NotificationDeliveries7d += stats.Deliveries
		counts.NotificationFailures7d += stats.Failures
		counts.NotificationFailuresAuthentication7d += stats.FailureClasses.Authentication
		counts.NotificationFailuresRateLimited7d += stats.FailureClasses.RateLimited
		counts.NotificationFailuresConnectivity7d += stats.FailureClasses.Connectivity
		counts.NotificationFailuresTLS7d += stats.FailureClasses.TLS
		counts.NotificationFailuresConfiguration7d += stats.FailureClasses.Configuration
		counts.NotificationFailuresRejected7d += stats.FailureClasses.Rejected
		counts.NotificationFailuresServerError7d += stats.FailureClasses.ServerError
		counts.NotificationFailuresUnknown7d += stats.FailureClasses.Unknown
	}
}

// accumulateAlertOutcomeCounts remains the focused history-folding seam used
// by monitoring tests. Production collection uses the manager snapshot so the
// same calculation also includes current age, adoption, and persistence health.
func accumulateAlertOutcomeCounts(counts *InstallSnapshotCounts, history []alerts.Alert, cutoff time.Time) {
	if counts == nil {
		return
	}
	quality := alerts.CalculateAlertQualitySnapshot(history, nil, cutoff, cutoff.Add(30*24*time.Hour))
	counts.AlertQuality.Add(quality)
	counts.AlertsFired30d += quality.Fired30d
	counts.AlertsAcknowledged30d += quality.Acknowledged30d
	counts.AlertsResolved30d += quality.Resolved30d
}

func accumulateInstallSnapshotCounts(counts *InstallSnapshotCounts, monitor *Monitor) {
	if counts == nil || monitor == nil {
		return
	}

	readState := monitor.GetUnifiedReadStateOrSnapshot()
	if readState != nil {
		counts.PVENodes += len(readState.Nodes())
		counts.PBSInstances += len(readState.PBSInstances())
		counts.PMGInstances += len(readState.PMGInstances())
		counts.VMs += len(readState.VMs())
		counts.Containers += len(readState.Containers())
		counts.AgentHosts += len(readState.Hosts())
		counts.DockerHosts += len(readState.DockerHosts())
		counts.DockerContainers += len(readState.DockerContainers())
		counts.KubernetesClusters += len(readState.K8sClusters())
		counts.KubernetesNodes += len(readState.K8sNodes())
		counts.KubernetesPods += len(readState.Pods())
		counts.KubernetesDeployments += len(readState.K8sDeployments())
		counts.StoragePools += len(readState.StoragePools())
		counts.PhysicalDisks += len(readState.PhysicalDisks())
	}
	resources, _ := monitor.UnifiedResourceSnapshot()
	accumulateInstallSnapshotUnifiedResourceCounts(counts, resources)
	counts.ActiveAlerts += len(monitor.ActiveAlertsSnapshot())
}

func accumulateInstallSnapshotUnifiedResourceCounts(counts *InstallSnapshotCounts, resources []unifiedresources.Resource) {
	if counts == nil {
		return
	}
	for _, resource := range resources {
		resourceType := unifiedresources.CanonicalResourceType(resource.Type)
		switch resourceType {
		case unifiedresources.ResourceTypeCeph:
			counts.CephClusters++
		case unifiedresources.ResourceTypeNetworkShare:
			counts.NetworkShares++
		}

		if resource.Availability != nil && resource.Availability.Enabled {
			counts.AvailabilityTargets++
		}

		if resource.TrueNAS != nil {
			switch resourceType {
			case unifiedresources.ResourceTypeAgent:
				counts.TrueNASSystems++
			case unifiedresources.ResourceTypeVM:
				counts.TrueNASVMs++
			case unifiedresources.ResourceTypeAppContainer:
				counts.TrueNASApps++
			}
		}

		if resource.VMware != nil {
			switch resourceType {
			case unifiedresources.ResourceTypeAgent:
				counts.VMwareHosts++
			case unifiedresources.ResourceTypeVM:
				counts.VMwareVMs++
			case unifiedresources.ResourceTypeStorage:
				counts.VMwareDatastores++
			}
		}
	}
}

// Stop stops the monitor
func (rm *ReloadableMonitor) Stop() {
	rm.mu.Lock()
	defer rm.mu.Unlock()

	if rm.cancel != nil {
		rm.cancel()
	}

	if rm.mtMonitor != nil {
		rm.mtMonitor.Stop()
	}
}