Pulse/pkg/reporting/reporting.go
rcourtman 686c2e8716 feat(reporting): availability section computed from the resource state timeline
Performance reports answered 'what were the averages' but never 'was my
infrastructure up' - the question a managed-service client reads a
monthly report for. Reports now carry an Availability summary derived
from the recorded resource change timeline (state_transition entries
keyed by the canonical unified ID):

- uptime percent over the observed portion of the window, outage count,
  total downtime, and longest outage, rendered in the executive summary
  with an explicit semantics note; fleet summaries gain a per-resource
  Uptime column and CSV exports gain availability header lines
- absent/unknown spans are unobserved time: excluded from the uptime
  math entirely and disclosed as coverage, never counted as downtime.
  The journal records a registry absence for every monitor restart, so
  treating gaps as outages would invent fleet-wide downtime every time
  the operator restarts Pulse
- warning states count as up (the resource is reachable and serving);
  the uptime label clamps rounding so any real downtime can never
  display as a clean 100%
- resources with no timeline render no availability section at all
  rather than a fabricated number

Verified live against a real 7-day window: uptime/outage/downtime
figures reconcile with the raw resource_changes journal.
2026-06-10 17:48:11 +01:00

221 lines
7.5 KiB
Go

package reporting
import (
"time"
)
// ReportFormat represents the output format of a report
type ReportFormat string
const (
FormatCSV ReportFormat = "csv"
FormatPDF ReportFormat = "pdf"
)
// MetricReportRequest defines the parameters for generating a report
type MetricReportRequest struct {
ResourceType string
ResourceID string
MetricType string // Optional, if empty all metrics for the resource are included
Start time.Time
End time.Time
Format ReportFormat
Title string
Branding ReportBranding
// MetricsResourceID overrides ResourceID for metrics-store queries.
// The v6 API addresses resources by canonical unified ID while the
// metrics store is keyed by each platform's native source ID (the
// resource's metricsTarget); recovery points and Patrol findings stay
// keyed by the unified ID. Handlers set this from the resource's
// metricsTarget so the store query can match without changing the
// identity the rest of the report pipeline sees.
MetricsResourceID string
// Optional enrichment data (populated by handler from monitor state)
Resource *ResourceInfo // Details about the resource being reported on
Alerts []AlertInfo // Active and recently resolved alerts for this resource
Backups []BackupInfo // Backup information for VMs/containers
Storage []StorageInfo // Storage pools (for nodes)
Disks []DiskInfo // Physical disk health (for nodes)
Availability *AvailabilityInfo // Observed availability over the window (from the state timeline)
// Optional narrative interpretation. When Narrator is non-nil the
// engine builds a NarrativeInput from the queried report data and asks
// it to produce the executive summary; on error or nil it falls back to
// the heuristic narrator. Findings are passed through to NarrativeInput
// so a narrator can reference Patrol activity in the period.
Narrator Narrator
FindingsProvider FindingsProvider
}
// AvailabilityInfo summarizes a resource's observed availability over the
// report window, derived from the recorded resource state timeline.
//
// Time the resource was absent from the registry or in an unknown state
// (for example while the monitor itself was restarting) is treated as
// unobserved: it is excluded from the uptime calculation entirely rather
// than counted as downtime, and disclosed through ObservedPercent. A
// monitoring gap is not an outage, and a client-facing stability report
// must not present one as such.
type AvailabilityInfo struct {
// UptimePercent is up / (up + down) over the observed portion of the
// window. Online and warning states count as up; offline counts as
// down. Zero when the resource was never observed in the window.
UptimePercent float64
// ObservedPercent is the share of the report window during which the
// resource state was actually being recorded.
ObservedPercent float64
// TotalDowntime is the cumulative time spent in a down state.
TotalDowntime time.Duration
// LongestOutage is the longest contiguous stretch of down time.
LongestOutage time.Duration
// DownIncidents counts distinct transitions into a down state.
DownIncidents int
}
// Observed reports whether the resource was observed at all in the window.
func (a *AvailabilityInfo) Observed() bool {
return a != nil && a.ObservedPercent > 0
}
// ResourceInfo contains details about the resource being reported on
type ResourceInfo struct {
Name string
DisplayName string
Status string
Host string // URL for nodes
Node string // Parent node for VMs/containers
Instance string // Proxmox instance name
Uptime int64
KernelVersion string
PVEVersion string
OSName string
OSVersion string
IPAddresses []string
CPUModel string
CPUCores int
CPUSockets int
MemoryTotal int64
DiskTotal int64
LoadAverage []float64
Temperature *float64 // CPU temp if available
Tags []string
ClusterName string
IsCluster bool
}
// AlertInfo contains alert information for the report
type AlertInfo struct {
Type string
Level string // warning, critical
Message string
Value float64
Threshold float64
StartTime time.Time
ResolvedTime *time.Time // nil if still active
Acknowledged bool
}
// BackupInfo contains backup information for VMs/containers
type BackupInfo struct {
Type string // vzdump, pbs
Storage string
Timestamp time.Time
Size int64
Verified bool
Protected bool
VolID string
NextBackup *time.Time
}
// StorageInfo contains storage pool information
type StorageInfo struct {
Name string
Type string // lvm, zfs, dir, nfs, etc.
Status string
Total int64
Used int64
Available int64
UsagePerc float64
Content string // images, rootdir, backup, etc.
ZFSHealth string // For ZFS pools
ZFSErrors int // Checksum/read/write errors
}
// DiskInfo contains physical disk health information
type DiskInfo struct {
Device string
Model string
Serial string
Type string // nvme, ssd, hdd
Size int64
Health string // PASSED, FAILED, UNKNOWN
Temperature int // Celsius
WearLevel int // 0-100, percentage of life REMAINING (100 = healthy, 0 = end of life, -1 = unknown)
}
// MultiReportRequest defines the parameters for generating a multi-resource report.
type MultiReportRequest struct {
Resources []MetricReportRequest // One per resource, each with enrichment
Format ReportFormat
Start time.Time
End time.Time
Title string
MetricType string
Branding ReportBranding
// Optional fleet-level narrative interpretation. When FleetNarrator is
// non-nil the engine builds a FleetNarrativeInput from the queried
// per-resource report data and asks it to produce the cross-resource
// summary; on error or nil it falls back to the heuristic fleet
// narrator. FindingsProvider, when set, is consulted per-resource so
// patrol findings can flow into per-resource narratives.
FleetNarrator FleetNarrator
Narrator Narrator
FindingsProvider FindingsProvider
}
// MultiReportData holds the data for multi-resource report generation.
type MultiReportData struct {
Title string
Start time.Time
End time.Time
GeneratedAt time.Time
Resources []*ReportData // Reuse existing ReportData per resource
TotalPoints int
Brand *ReportBrand
// Fleet-level narrative interpretation, populated by the engine when
// the request supplies a FleetNarrator (or always populated with the
// heuristic fallback). The renderer prefers this over recomputing
// observations inline.
FleetNarrative *FleetNarrative
}
// Engine defines the interface for report generation.
// This allows the enterprise version to provide PDF/CSV generation.
//
// NarrativeFor and FleetNarrativeFor return the structured narrative
// without rendering, for callers that want the synthesis layer in a
// non-PDF form (Pulse Assistant tool calls, programmatic consumers).
type Engine interface {
Generate(req MetricReportRequest) (data []byte, contentType string, err error)
GenerateMulti(req MultiReportRequest) (data []byte, contentType string, err error)
NarrativeFor(req MetricReportRequest) (*Narrative, error)
FleetNarrativeFor(req MultiReportRequest) (*FleetNarrative, error)
}
var (
globalEngine Engine
)
// SetEngine sets the global report engine.
func SetEngine(e Engine) {
globalEngine = e
}
// GetEngine returns the current global report engine.
func GetEngine() Engine {
return globalEngine
}