Events | Logbun

Pass onEvent on LogbunConfig. Listener throws are swallowed (safeEmit) and never break the pipeline.

type LogbunEventType =
  | 'enqueue'
  | 'flush_ok'
  | 'flush_fail'
  | 'dlq'
  | 'drop'
  | 'truncated'
  | 'wal_fail'
  | 'bootstrap_fail'
  | 'prune_fail'
  | 'poison'
  | 'degraded'
  | 'stats'
  | 'limit';

interface LogbunEvent {
  type: LogbunEventType;
  tenantId?: string | null;
  error?: string;
  count?: number;
  detail?: string;
}

`detail` values used by the engine

`detail`	Typical `type`	Meaning
`unsafe_default_volatile`	`limit`	`mode` is volatile (including explicit)
`unsafe_default_require_tenant`	`limit`	`requireTenantId` left false and not DPT
`pre_ready_volatile`	`limit`	Durable-mode accept into the pre-ready buffer
`pre_ready_buffer_full`	`drop`	Pre-ready buffer at `maxPreReadyBuffer`
`max_active_tenants`	`dlq` or `drop`	New tenant key beyond cap
`max_tenant_id_bytes`	`drop`	Oversize `tenantId`
`require_tenant_id`	`drop`	Missing tenant when required
`get_tenant_id`	`drop`	`resolveTenantId` threw (`fire` path)
`wal_full`	`wal_fail` / `dlq`	Journal at cap; single-log DLQ fallback may still admit
`dlq_full`	—	DLQ at pending+processing cap
`recovery`	—	Recovery path
`durable_admission_scheduling`	`flush_fail`	Journal/DLQ committed; alarm scheduling failed
`maintenance_flush`	`flush_fail`	Flush phase of `runMaintenance` failed
`maintenance_dlq_scan`	`flush_fail`	DLQ scan phase failed
`maintenance_rearm`	`flush_fail`	`requestMaintenance` after a pass failed
`shutdown`	`drop`	`fire` after `shutdown()`
`shutdown_flush`	`flush_fail`	Flush during shutdown
`shutdown_maintenance`	`flush_fail`	In-flight maintenance during shutdown
`shutdown_deadline`	`limit`	`inflightFlushes > 0` after shutdown flush
`degraded`	`drop`	`fire` while degraded
`fire`	`drop`	Inner `fire` path threw
`reliability_get_stats`	`stats`	`getStatsDetailed` disk path threw
`prune_incomplete`	prune failure	SQLite/Turso batched prune hit the safety cap

onEvent is the observability seam. It is not a delivery callback.