Logs & Telemetry

Use client logs to diagnose a problem on your computer, or the website's Logs panel to inspect recent activity. Neither a connected status nor a log entry alone proves a model is ready to answer.

Read client logs#

Read recent output once:

actual logs --tail 100 --once

Follow new output:

actual logs

Filter recent errors:

actual logs --since 5m --level error --once

Use actual logs --help for available filters, including component selection. Use actual stats or actual stats --format json to inspect current operating statistics.

Find the log file#

Run actual status and read Log file. Current installations normally keep rotating logs in the user's local application-data directory:

Settings and environment overrides can change the location. Older builds may not report a destination, and file logging can be disabled. The path reported by your running instance is the one to inspect.

Collect a diagnostic excerpt#

actual logs --tail 200 --once > actual-diagnostic.log

Review the excerpt before sharing it and remove credentials or private content. Include the time and timezone, affected computer or cluster, model, client version, and the action that failed.

If you need a custom log destination, deliberately restart Actual with the setting. Restarting interrupts current work:

actual restart --log-file /tmp/actual.log

Passing a startup log flag while Actual is already running does not change that running instance's log destination. Use a suitable writable path on your platform.

Inspect website logs#

Open Computers, select the relevant computer or cluster, and open Logs. The panel shows up to 50 recent rows and refreshes while open. Updates can be delayed, and offline or unavailable computers may have no recent entries.

The panel has no dedicated export button. Copy a small relevant excerpt, or collect client logs for a longer diagnostic record.

See Machine Installation, Model Discovery, and API Quickstart for recovery steps.