Monitoring and Troubleshooting
Understand run status, find output, and stop scripts safely.
Follow a run in Furge
Historical runs stay in the editor's Backtests panel. Live deployments open a job page and remain available under Scripts → Jobs. Run details update automatically while active; opening a details tab loads its current data.
| Status | What it means | What to do |
|---|---|---|
| Queued / checking | Preparing the saved script and running deployment checks. | Open Logs; do not repeatedly deploy the same script. |
| Backtesting | Processing the chosen historical range with simulated execution. | Follow Logs, then inspect Results. |
| Starting | The live job is starting. | Wait for the runtime to report running. |
| Running | The runtime reports the job active. It is not proof that an order filled or a fresh record arrived. | Inspect Events and Logs. |
| Stopping | Stop was requested; shutdown is in progress. | Wait for stopped/completed status. |
| Completed / stopped | The job has ended. | Inspect its final output and artifacts. |
| Failed | Preparation or execution failed. | Read the error and Logs before trying again. |
| Unknown | The current job state could not be confirmed. | Check Runtime and account activity before resubmitting. |
Each run keeps its saved source revision. Editing a script does not change an already-running job; deploy the new revision separately and stop the old job if it should no longer run.
Where output goes
| Script output | Where to see it |
|---|---|
print(...) | Logs. |
return {"metrics": {...}} from on_data | Activity for live jobs. Results shows the latest metrics for backtests. |
| Execution events | Live Results / Events and on_execution(ctx). An accepted local reference is not a fill. |
Files saved with ctx.artifact_path(...) | Artifacts after export; see Artifacts. |
Common problems
| Problem | Check |
|---|---|
| No output yet | A source must update before on_data runs. Candles require a complete interval with trades. Onchain markets may be quiet. |
None / empty history | Guard indexed reads before using fields. No index returns [] until records exist. |
| History is shorter than expected | lookback bounds retained records; it does not fetch past ticks when a live job starts. |
| Backtest rejected | Check source availability. Raw trades and direct orderbooks are live-only. Historical intervals are provider-specific. |
| Source not configured | Declare a complete selector as a literal string or module-level constant. A dynamic string cannot introduce an undeclared subscription. |
| Key missing from a record | Check the source's returned fields. Optional and provider-specific fields are not guaranteed; use .get(...). |
| Runtime unavailable | Open Runtime, check its status/version, and start or upgrade it when appropriate. Saving a script does not start a runtime. |
| No trades in Results | Read-only scripts do not trade. For trading scripts, inspect execution events, conditions, accounts, and venue minimums. |
| Submission timed out | Inspect account activity or operation status first. A timeout does not prove the exchange rejected it. |
Never put private keys, API secrets, or bot tokens into source code or logs. Use configured accounts and supported notification settings.
Stop safely
Click Stop on the active job or backtest. Closing the page does not stop execution. Stopping does not automatically cancel resting orders, close positions, or reverse a broadcast contract transaction. Inspect the affected exchange account after stopping a trading script.
CLI equivalent
mlab script jobs
mlab script status <JOB_ID>
mlab script logs <JOB_ID> --follow
mlab script stop <JOB_ID>For runtime limits and available packages, see Runtime Safety.