# Metrics (/docs/reference/metrics) > The 47 Prometheus metric families better-compute exposes on /metrics, with their labels. Every family carries the `better_compute_` prefix and none has a tenant label (series stay bounded by `maxSeries`). Histograms also expose `_bucket`, `_sum` and `_count` series. Families are grouped by the function that registers them: `standardMetrics` on every process, the others where a host wires them. ## standardMetrics [#standardmetrics] | Metric | Type | Labels | Description | | ------------------------------------------- | --------- | --------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | `better_compute_gateway_sessions` | gauge | `role` | Open daemon sessions on this gateway | | `better_compute_gateway_handshakes_total` | counter | `outcome` | Session handshakes | | `better_compute_command_latency_seconds` | histogram | `kind`, `lane` | Command enqueue to terminal result. Buckets (s): 0.05, 0.1, 0.25, 0.5, 1, 1.5, 2.5, 5, 10, 30, 60 | | `better_compute_commands_swept_total` | counter | `to` | Commands the CommandSweeper moved | | `better_compute_work_queue_depth` | gauge | `kind` | Due work\_queue rows | | `better_compute_reconcile_duration_seconds` | histogram | `reconciler` | One reconciler run. Buckets (s): 0.005, 0.01, 0.025, 0.05, 0.1, 0.25, 0.5, 1, 2.5, 5, 10, 30 | | `better_compute_generation_lag_nodes` | gauge | | Nodes whose observed generation is behind | | `better_compute_placements_total` | counter | `outcome` | Placement attempts | | `better_compute_booking_conflicts_total` | counter | | Booking conflicts that forced a replan | | `better_compute_pool_gpus` | gauge | `state` | GPUs by state across pools | | `better_compute_node_connectivity` | gauge | `state` | Nodes by connectivity | | `better_compute_backlog` | gauge | `queue` | Rows waiting in a queue table | | `better_compute_usage_dead_letters_total` | counter | | Usage events dead-lettered | | `better_compute_iam_inbox_spill_total` | counter | | IAM events spilled because the inbox was down | | `better_compute_iam_calls_total` | counter | `op`, `outcome` | IamPort calls | | `better_compute_iam_call_duration_seconds` | histogram | `op` | IamPort call latency (each op takes ≥1 IAM lock acquisition). Buckets (s): 0.005, 0.01, 0.025, 0.05, 0.1, 0.25, 0.5, 1, 2.5, 5, 10, 30 | | `better_compute_resolver_lookups_total` | counter | `result` | Authz resolver lookups | | `better_compute_tenant_states` | gauge | `state` | IAM tenants compute refuses (§12.8), by state: suspended, deleted | | `better_compute_tenant_teardowns_pending` | gauge | | Deleted tenants whose teardown has not finished (a module still releasing) | | `better_compute_sandbox_ready_seconds` | histogram | `source` | Sandbox create to its instance first running (§3.7), by source: a warm claim or a cold create; the counts give the warm-pool hit ratio. Buckets (s): 0.5, 1, 2, 5, 10, 20, 30, 60, 120, 300 | | `better_compute_sandbox_expiry_lag_seconds` | histogram | | How late a sandbox was expired after its expiresAt. Buckets (s): 1, 5, 15, 30, 60, 120, 300, 900 | | `better_compute_sandbox_warm_instances` | gauge | `state` | Unclaimed warm-pool instances, by instance state | | `better_compute_sandbox_warm_deficit` | gauge | | Warm instances missing below the templates’ warm.min, summed over live templates | | `better_compute_sandbox_warm_refills_total` | counter | | Warm-pool instances created to refill templates | | `better_compute_cert_expiry_seconds` | gauge | `kind` | Seconds until the soonest expiry | | `better_compute_clock_offset_seconds` | gauge | | Process clock minus compute\_now() | | `better_compute_security_events_total` | counter | `type` | Security events recorded | ## createObserver [#createobserver] | Metric | Type | Labels | Description | | -------------------------------------- | --------- | --------------------------------- | ---------------------------------------------------------------------------------------------------------- | | `better_compute_spans_total` | counter | `kind`, `name`, `outcome`, `code` | Units of work by kind, name, outcome and code | | `better_compute_span_duration_seconds` | histogram | `kind` | Latency of units of work by kind. Buckets (s): 0.005, 0.01, 0.025, 0.05, 0.1, 0.25, 0.5, 1, 2.5, 5, 10, 30 | ## meteredAttestor [#meteredattestor] | Metric | Type | Labels | Description | | ---------------------------------------------- | --------- | --------------------- | ---------------------------------------------------------------------------------------------------- | | `better_compute_attestor_verifications_total` | counter | `attestor`, `outcome` | Attestor verifications by outcome | | `better_compute_attestor_verification_seconds` | histogram | `attestor` | Attestor verification time. Buckets (s): 0.005, 0.01, 0.025, 0.05, 0.1, 0.25, 0.5, 1, 2.5, 5, 10, 30 | ## meteredGatewayServices [#meteredgatewayservices] | Metric | Type | Labels | Description | | -------------------------------------------- | --------- | ------------------- | ----------------------------------------------------------------------------------------------------- | | `better_compute_gateway_service_calls_total` | counter | `method`, `outcome` | Join, Renew and Secret calls by outcome | | `better_compute_gateway_service_seconds` | histogram | `method` | Join, Renew and Secret time. Buckets (s): 0.005, 0.01, 0.025, 0.05, 0.1, 0.25, 0.5, 1, 2.5, 5, 10, 30 | ## meteredProvider [#meteredprovider] | Metric | Type | Labels | Description | | -------------------------------------- | --------- | ------------------------------- | -------------------------------------------------------------------------------- | | `better_compute_provider_calls_total` | counter | `provider`, `method`, `outcome` | Provider API calls by outcome | | `better_compute_provider_call_seconds` | histogram | `provider`, `method` | Provider API call time. Buckets (s): 0.05, 0.1, 0.25, 0.5, 1, 2.5, 5, 10, 30, 60 | ## placementObserver [#placementobserver] | Metric | Type | Labels | Description | | ---------------------------------- | --------- | -------- | ---------------------------------------------------------------------------------------------------------- | | `better_compute_placement_seconds` | histogram | `placer` | One placement attempt, by placer. Buckets (s): 0.005, 0.01, 0.025, 0.05, 0.1, 0.25, 0.5, 1, 2.5, 5, 10, 30 | ## gatewayCollector [#gatewaycollector] | Metric | Type | Labels | Description | | -------------------------------------------------- | ----- | ----------------- | ------------------------------------------------------------------------------ | | `better_compute_gateway_open_sessions` | gauge | `gateway`, `role` | Open agent sessions by role | | `better_compute_gateway_stream_headroom` | gauge | `gateway` | Streams this gateway can still admit | | `better_compute_gateway_draining` | gauge | `gateway` | Whether the gateway is draining (1) or not (0) | | `better_compute_gateway_lease_held` | gauge | `gateway` | Whether the gateway holds its lease (1) or is fenced (0) | | `better_compute_gateway_commands_in_flight` | gauge | `gateway` | Commands dispatched and not yet settled | | `better_compute_gateway_sessions_upgrade_required` | gauge | `gateway` | Sessions told to upgrade | | `better_compute_gateway_loop_consecutive_failures` | gauge | `gateway`, `loop` | Failed runs of a gateway loop since its last clean one (alert on a stuck loop) | | `better_compute_gateway_loop_failures` | gauge | `gateway`, `loop` | Failed runs of a gateway loop since start | | `better_compute_gateway_loop_last_ok_age_seconds` | gauge | `gateway`, `loop` | Seconds since a gateway loop last ran cleanly (-1 before its first clean run) | ## revocationFeedCollector [#revocationfeedcollector] | Metric | Type | Labels | Description | | ----------------------------------------------------- | ----- | ------ | ---------------------------------------------------------------------------------------------- | | `better_compute_revocation_feed_consecutive_failures` | gauge | `feed` | Revocation feed polls failed in a row: revoked certificates keep their sessions while it fails | ## eventCounter [#eventcounter] | Metric | Type | Labels | Description | | ----------------------------- | ------- | ------------------ | ------------------------------- | | `better_compute_events_total` | counter | `family`, `module` | Outbox events by catalog family |