Observability Cookbook Distributed debugging requires end-to-end visibility. This page links tracing, metrics, and logging into actionable steps. Structured logs Include correlation IDs to link logs to traces. log.LogInformation("{CorrelationId} handling {Message}", id, msg); Metrics to watch Mailbox depth per actor Message processing latency Restart counts and deadletters Sample dashboard graph LR subgraph Metrics m1[Mailbox depth] m2[Processing latency] end subgraph Logs l1[Correlation ID] end subgraph Traces t1[Spans per PID] end m1-->l1 m2-->t1 Start with these building blocks and iterate: good observability turns incidents into quick fixes rather than mysteries. Read more...

Supervisor Strategy Recipes Supervisors decide what to do when their children fail. The recipes below build on the core Supervision concepts with practical defaults. Restart with backoff Restarting too quickly can thrash the system. Apply exponential backoff and give up after repeated failures. .NET public override SupervisorStrategy SupervisorStrategy => new OneForOneStrategy( maxNrOfRetries: 5, withinTimeRange: TimeSpan.FromSeconds(10), decider: ex => SupervisorDirective.Restart); Go var strategy = actor.NewOneForOneStrategy(5, time.Second*10, func(reason interface{}) actor.Directive { return actor. Read more...