Logs

Edge Compute exposes two log types through the same CLI command and API endpoint:

Type	Source	What it contains
`runtime`	Your function	stdout/stderr such as `console.log`, `console.error`, and uncaught exceptions
`invocations`	Edge Compute platform	One record for each HTTP request served, even when your function writes no logs

Runtime logs

Runtime logs are the default:

telnyx-edge logs my-func
telnyx-edge logs my-func --type runtime
telnyx-edge logs my-func --since 10m
telnyx-edge logs my-func --last 200
telnyx-edge logs my-func --json

Each line prints as [timestamp] [level] message, oldest first. The level is best-effort and can be inaccurate on stack traces; a multi-line error arrives as several separate entries.

Invocation logs

Use --type invocations for platform-generated request records:

telnyx-edge logs my-func --type invocations
telnyx-edge logs my-func --type invocations --since 10m
telnyx-edge logs my-func --type invocations --last 200
telnyx-edge logs my-func --type invocations --json

Invocation logs report traffic even when your code never calls console.log. Each record includes request metadata such as timestamp, HTTP method, path, status code, duration, request and response size, and serving region when available. --since, --last, and --json work the same way for both log types. --since defaults to 1h and is capped at 24h; --last defaults to 50 and is capped at 250. Results are returned oldest first. The API exposes both streams through GET /v2/compute/funcs/{id}/logs with type=runtime or type=invocations. See the API Reference.

Tail logs live

Requires CLI v0.5.3 or newer.

telnyx-edge logs my-func --tail
telnyx-edge logs my-func --tail --type invocations
telnyx-edge logs my-func --tail --json

--tail switches from the default read-a-window-and-exit behavior to a live stream: the CLI stays attached and prints new log lines as the platform receives them, until you press Ctrl-C. The stream is delivered over a WebSocket direct from the ingest pipeline — there is no polling and no history: only lines written while you are attached arrive.

Flag	Description
`--tail`	Stream new log lines live instead of reading history
`--type`	In tail mode, `runtime` or `invocations`; omitted, tail shows both interleaved

In tail mode, --since and --last are meaningless (there is no history to page through) and are silently ignored. --type means something different here than in the one-shot read: an unset --type shows both runtime and invocation lines, interleaved as they arrive; pass --type runtime or --type invocations to filter to one kind. With --json, each line is one JSON object in the shape of the record that arrived. If the connection drops, the CLI reconnects with capped backoff and keeps streaming.

Emit custom events to a sink you run

Runtime and invocation logs cover platform-retained logs. If you also need application-specific structured events in your own system, send them over HTTPS to a collector you control. Store the collector credential as a secret, never in code. Declare the secret in func.toml:

[[secrets]]
binding = "LOG_SINK_KEY"
name    = "LOG_SINK_KEY"

telnyx-edge secrets add LOG_SINK_KEY "<your collector's API key>"
telnyx-edge types

Keep telemetry off the response’s critical path, apply an outbound timeout, and do not let a failed telemetry request fail the function response. Log metadata rather than sensitive request payloads. See the CLI reference for the full command reference.

Overview

Get Started

Best Practices

Guides

Local Development

Configuration

CI/CD

Custom Domains

Runtime APIs

Observability

CLI

Platform

For AI Agents

Runtime logs

Invocation logs

Tail logs live

Emit custom events to a sink you run

​Runtime logs

​Invocation logs

​Tail logs live

​Emit custom events to a sink you run

Runtime logs

Invocation logs

Tail logs live

Emit custom events to a sink you run