Runs
The kinds of run, where each one's history lives, how long it is kept, and what you can cancel.
A run is one execution of something you set up: a sync, file processing, a webhook delivery, a pipeline, an automation, an app or a report. Each kind keeps its history next to the thing that ran.
| Run | What starts it | Where its history is |
|---|---|---|
| Sync run | Its schedule, or Run now | Runs and Logs tabs of the connection on Home |
| File processing | Adding, replacing or removing files in a File Set, or Process files | The Files tab of the file processor on Home |
| Webhook delivery | A POST to the connection's ingest URL | Query the connection's dataset |
| Pipeline run | Run, a deploy, or an update to a source with Trigger on update | Runs tab of the pipeline on Home |
| Automation run | Its schedule, or Run now | Run history on the automation's page |
| App run | Manual, its schedule, or an update to an input dataset with Trigger on update | Runs tab of the app on Home |
| Report run | Run now, chat, or its schedule | The report's page, with its documents |
Retries
When a sync run fails because of a problem on Beetl's side, Beetl retries it automatically, up to 7 attempts. A sync run that fails in the integration's own code, for example because the code throws an error, is not retried; fix the cause and run it again. Each sync attempt has 15 minutes and 1 GiB of staged output. A sync's checkpoint only moves forward after its write commits, so the next run starts from the same place and no records are skipped.
File processing runs in the background and is not listed on a Runs tab. The processor's Files tab shows each document as Waiting, Processed, Needs review or Failed, and a failed document shows the reason. Beetl does not repeat an AI request on its own when the earlier one may have been billed. See Documents and spreadsheets.
A webhook answers 202 once Beetl has accepted the delivery. To confirm the data landed, query the
connection's dataset. See Webhook.
Reading run history
Syncs. The Runs tab on a connection lists the runs of all its syncs, filterable by sync and
time range, with status, duration, the rows added by Append runs, and the failure reason for
a failed run. A run is queued, running, retrying, succeeded, skipped (another run was active, or the
settings changed) or failed. The Logs tab shows each run's log lines, including the messages the
integration writes with ctx.log.
Pipelines. The Runs tab on a pipeline shows each run's status, what started it (Manual, Event for a source update, or System for a deploy), its timings and any error. Pipelines do not write logs.
Automations. An automation's page lists its runs. A completed run is Healthy or Needs attention; open it to read the report and the queries behind it. A run that takes more than 7 minutes is marked failed. See Automations.
Apps. The Runs tab of an app lists its runs. A run is Pending, Processing, Retrying, AwaitingPublication, Paused, Succeeded or Failed. A Paused run stopped while publishing its output datasets, and the tab offers Resume publication or Abandon publication. See Apps.
Reports. A report run is running, completed or failed, and shows how many documents were issued and how many failed. See Reports.
MCP clients can read pipeline and automation runs with list_pipeline_runs, get_pipeline_run,
list_automation_runs and get_automation_run, and app runs with list_app_runs and
get_app_run.
How long history is kept
| History | Kept for |
|---|---|
| Sync runs and logs | 30 days |
| Pipeline runs | No limit |
| Automation runs and reports | No limit |
| App runs | No limit |
| Report documents | Until the report is deleted |
| Query results, including automation evidence | 7 days |
An automation report outlives the query results it links to, so evidence in a report older than a week no longer opens. See Query results.
What you can cancel
A pipeline stage preview can be cancelled while it runs. Cancel in the Query editor stops waiting for the result, but the query may still finish and appear under Recent results. Sync runs, file processing, pipeline runs, automation runs, app runs and report runs cannot be cancelled once started; pausing a sync or an automation stops future runs only.
Getting alerted
Beetl shows run status in the app and does not send email, Slack or webhook notifications. The
markers on Home and the Automations list show failures and findings at a glance. To raise alerts in
your own tools, have an MCP client poll list_pipeline_runs and list_automation_runs on a
schedule.
Related
Last updated on