- config: defaults -> JSON file -> NOEMA_* env with validation and redaction - observability: slog with secret-redacting handler; Prometheus registry with the metric set from the plan (events, workspace, memory, LLM, tools, queues) - scheduler: supervised Group and bounded Pool so no goroutine is unowned - storage: pgx pool with statement timeout, embedded migration runner (advisory lock, checksums, up/down/status), per-test schema helper - migrations/0001: append-only guard trigger, users, sessions, api tokens, recovery codes, audit log, settings - pkg/ids: UUIDv7 with monotonic counter
64 lines
4.1 KiB
Go
64 lines
4.1 KiB
Go
package observability
|
|
|
|
import (
|
|
"github.com/prometheus/client_golang/prometheus"
|
|
"github.com/prometheus/client_golang/prometheus/collectors"
|
|
)
|
|
|
|
// Metrics holds every Prometheus instrument Noema exposes. Instruments are
|
|
// created once here so that names are visible in one place and no package
|
|
// registers globals.
|
|
type Metrics struct {
|
|
Registry *prometheus.Registry
|
|
|
|
HTTPRequests *prometheus.CounterVec
|
|
HTTPDuration *prometheus.HistogramVec
|
|
AuthFailures prometheus.Counter
|
|
Events *prometheus.CounterVec
|
|
Candidates prometheus.Counter
|
|
Admissions prometheus.Counter
|
|
Retrievals prometheus.Counter
|
|
Beliefs prometheus.Gauge
|
|
Predictions prometheus.Counter
|
|
PredictionErr prometheus.Counter
|
|
LLMRequests *prometheus.CounterVec
|
|
LLMTokens *prometheus.CounterVec
|
|
LLMFailures *prometheus.CounterVec
|
|
ToolRequests *prometheus.CounterVec
|
|
ToolDenied *prometheus.CounterVec
|
|
QueueDepth *prometheus.GaugeVec
|
|
Workers *prometheus.GaugeVec
|
|
}
|
|
|
|
// NewMetrics creates and registers all instruments on a fresh registry.
|
|
func NewMetrics() *Metrics {
|
|
reg := prometheus.NewRegistry()
|
|
m := &Metrics{
|
|
Registry: reg,
|
|
HTTPRequests: prometheus.NewCounterVec(prometheus.CounterOpts{Name: "noema_http_requests_total", Help: "HTTP requests by route and status."}, []string{"route", "status"}),
|
|
HTTPDuration: prometheus.NewHistogramVec(prometheus.HistogramOpts{Name: "noema_http_request_duration_seconds", Help: "HTTP request latency.", Buckets: prometheus.DefBuckets}, []string{"route"}),
|
|
AuthFailures: prometheus.NewCounter(prometheus.CounterOpts{Name: "noema_auth_failures_total", Help: "Failed authentication attempts."}),
|
|
Events: prometheus.NewCounterVec(prometheus.CounterOpts{Name: "noema_events_total", Help: "Cognitive events appended by type."}, []string{"type"}),
|
|
Candidates: prometheus.NewCounter(prometheus.CounterOpts{Name: "noema_workspace_candidates_total", Help: "Thought candidates scored."}),
|
|
Admissions: prometheus.NewCounter(prometheus.CounterOpts{Name: "noema_workspace_admissions_total", Help: "Thoughts admitted to the workspace."}),
|
|
Retrievals: prometheus.NewCounter(prometheus.CounterOpts{Name: "noema_memory_retrievals_total", Help: "Memory retrievals."}),
|
|
Beliefs: prometheus.NewGauge(prometheus.GaugeOpts{Name: "noema_beliefs_total", Help: "Beliefs currently held across minds."}),
|
|
Predictions: prometheus.NewCounter(prometheus.CounterOpts{Name: "noema_predictions_total", Help: "Predictions created."}),
|
|
PredictionErr: prometheus.NewCounter(prometheus.CounterOpts{Name: "noema_prediction_errors_total", Help: "Prediction error events."}),
|
|
LLMRequests: prometheus.NewCounterVec(prometheus.CounterOpts{Name: "noema_llm_requests_total", Help: "LLM requests by provider and role."}, []string{"provider", "role"}),
|
|
LLMTokens: prometheus.NewCounterVec(prometheus.CounterOpts{Name: "noema_llm_tokens_total", Help: "LLM tokens by provider and direction."}, []string{"provider", "direction"}),
|
|
LLMFailures: prometheus.NewCounterVec(prometheus.CounterOpts{Name: "noema_llm_failures_total", Help: "LLM failures by provider."}, []string{"provider"}),
|
|
ToolRequests: prometheus.NewCounterVec(prometheus.CounterOpts{Name: "noema_tool_requests_total", Help: "Capability invocations by capability."}, []string{"capability"}),
|
|
ToolDenied: prometheus.NewCounterVec(prometheus.CounterOpts{Name: "noema_tool_denied_total", Help: "Denied capability invocations."}, []string{"capability"}),
|
|
QueueDepth: prometheus.NewGaugeVec(prometheus.GaugeOpts{Name: "noema_cognitive_queue_depth", Help: "Pending work per queue."}, []string{"queue"}),
|
|
Workers: prometheus.NewGaugeVec(prometheus.GaugeOpts{Name: "noema_worker_count", Help: "Busy workers per pool."}, []string{"pool"}),
|
|
}
|
|
reg.MustRegister(
|
|
collectors.NewGoCollector(), collectors.NewProcessCollector(collectors.ProcessCollectorOpts{}),
|
|
m.HTTPRequests, m.HTTPDuration, m.AuthFailures, m.Events, m.Candidates, m.Admissions, m.Retrievals,
|
|
m.Beliefs, m.Predictions, m.PredictionErr, m.LLMRequests, m.LLMTokens, m.LLMFailures, m.ToolRequests,
|
|
m.ToolDenied, m.QueueDepth, m.Workers,
|
|
)
|
|
return m
|
|
}
|