Use Lakona Observability

Lakona uses the standard .NET telemetry primitives: Meter, ActivitySource, and ILogger. The application owns the OpenTelemetry SDK, Collector, exporters, dashboards, and alerts. There is no Lakona-specific telemetry protocol to deploy or learn.

Connect OpenTelemetry

Subscribe to the instrumentation scopes published by LakonaGameServerTelemetry:

using Lakona.Game.Server.Observability;
using OpenTelemetry.Metrics;
using OpenTelemetry.Resources;
using OpenTelemetry.Trace;

server.AddServices(services =>
{
    services.AddOpenTelemetry()
        .ConfigureResource(resource => resource.AddService("game-gateway"))
        .WithMetrics(metrics => metrics
            .AddMeter(LakonaGameServerTelemetry.MeterNames.ToArray())
            .AddRuntimeInstrumentation()
            .AddProcessInstrumentation()
            .AddOtlpExporter())
        .WithTracing(tracing => tracing
            .AddSource(LakonaGameServerTelemetry.ActivitySourceNames.ToArray())
            .AddAspNetCoreInstrumentation()
            .AddOtlpExporter());
});

Point the exporter at the deployment’s Collector with standard OTEL_* environment variables. The Collector can then route the same signals to Prometheus, Grafana, Tempo, Jaeger, Loki, Application Insights, or a hosted OpenTelemetry backend.

What To Monitor

Start with host CPU, memory, GC, process restarts, and network throughput from standard runtime/process/host instrumentation. Add Lakona’s low-cardinality application metrics:

• actor activation population and mailbox queue length
• active, disconnected, and resumable sessions
• cluster route drops, expiry, and backpressure
• RPC request rate, response status, and dispatch duration
• timer capacity rejections
• reliable-push continuity loss

Keep actor ids, session ids, and request ids out of metric labels. Use sampled traces and structured logs when individual-request detail is needed.

Health Is A Separate Probe

Telemetry export can be delayed or unavailable without making a process dead. Lakona therefore keeps orchestration probes as HTTP endpoints:

GET /_lakona/health/live
GET /_lakona/health/ready

Configure those endpoints under Lakona:Health and the shared listener under Lakona:Management:Http. Hotfix admin access, if enabled, is configured under Lakona:Management:Admin.

Lakona no longer exposes /_lakona/diagnostics/* or owns a Prometheus/trace export switch. For the complete source catalog, setup example, and multi-node guidance, see the repository’s Observability guide.