StatefulActor now has built-in observability. Runtime logs, invocation logs, and metrics are available directly from the telnyx-edge CLI, with real-time streaming for logs, so you can debug actors without wiring up a separate monitoring stack.
What's new
- Runtime logs: The console.log and console.error output your actor methods produce, retrieved with telnyx-edge actors logs <type> --type runtime. Runtime logs are type-level output and are not correlated to a specific invocation ID.
- Invocation logs: One platform-generated record per method call, including the instance, method, outcome, and duration. Filter to a single actor instance with --instance <id>. Retrieved with telnyx-edge actors logs <type> --type invocations.
- Metrics: Request counts, success and error rates, latency, request and response bytes, CPU, memory, and stored state size. Retrieved with telnyx-edge actors metrics <type>.
- Live streaming: Pass --tail to telnyx-edge actors logs <type> to stay attached and stream new log lines as they arrive, for runtime and invocation logs alike, with --type filtering for both log types and --instance <id> filtering for invocation logs. For example, telnyx-edge actors logs my-actor --tail --type invocations --instance alice. --tail requires CLI v0.5.10 or newer.
By default, both commands read a bounded window of historical data, configurable with --since. History mode exits after printing, while --tail keeps the logs command attached and streams new lines as they arrive. Live sessions are severed roughly every 10 minutes and reconnect automatically, but lines emitted during the gap are lost. You need CLI v0.5.8 or newer for history-mode logs and metrics. --tail requires v0.5.10 or newer.
To debug an issue, find failed method calls in invocation logs, see which actor instance and method were involved, inspect nearby runtime output, and check metrics for latency spikes, memory pressure, or changes in stored state. To catch a problem as it happens, stream both log types live with --tail.
Why it matters
When a StatefulActor represents a checkout, session, game, device, or other stateful entity, you need to debug it in production. In a multi-vendor stack, observability is its own vendor problem. You run the actor on one platform, ship logs to another, pipe metrics to a third, and correlate across dashboards that were never designed to talk to each other. The monitoring stack becomes a thing you build and maintain alongside the application.
On Telnyx, the platform that runs the actor produces its logs and metrics because it is the thing executing the code. Runtime output, method-level invocation records, and resource metrics are available from the same CLI you use to manage your Stateful Actors. You get useful observability out of the box, without having to integrate a separate monitoring stack just to start debugging.
Getting started
- Update the CLI to v0.5.10 or newer: npm i -g @telnyx/edge.
- Pull runtime logs for a type: telnyx-edge actors logs my-actor --type runtime.
- Pull invocation logs to see method-level timing and outcomes: telnyx-edge actors logs my-actor --type invocations.
- Pull metrics for request counts, latency, CPU, memory, and stored state size: telnyx-edge actors metrics my-actor.
- Stream logs live as they arrive: telnyx-edge actors logs my-actor --tail.
Learn more in the actor observability docs, including logs and metrics.
