Skip to main content

Monitoring: logs and runs

Dataset log​

Datasets → View log (and View log on any dataset row) opens the Dataset log: every load, manual, scheduled, triggered or Gateway-pushed, with Queued, Dataset, System, Instance, Type (full, incremental, export, field rename, instance rename, AI import), Status, Records and Duration, and the error on failure. Each export to a destination is its own export-sync row under the load it belongs to, with the destination in the System column; a load's duration excludes its exports. Clicking a row opens the detail with the endpoint URLs that were called.

The grid stays live while it is unscrolled and polls while anything is running; once scrolled, a banner offers Refresh.

Job log and Schedule log​

Jobs → View log lists job runs with Started, Job, What ran, Status, Duration and the error, filtered by status and date; schedules have the same list with Trigger and Steps. A row opens the Run detail: status, trigger, timings, a phase strip sized by time that filters the grid when clicked, and the tree of what ran, with Kind, What ran, Where, Start, End, Duration and Records (subtotals on the phases and activities). Rows are tinted by outcome rather than carrying a status column; search also matches a fan-out value; Export to Excel exports the visible rows. Stop stops a running run, and Clear stuck steps tidies a finished run whose steps are still marked running.

Statuses: queued, running, succeeded, failed, timeout, cancelled (stopped by a person), skipped, aborted by restart (the service restarted mid-run) and completed with failures (shown as warning: steps set to continue on error failed and the job ran to the end).

Lineage​

Jobs and schedules have a lineage view, described in Lineage view, and datasets a lineage page from the Datasets list.

Stuck loads and jobs​

  • A cloud load or export that reports no progress for 90 minutes is failed automatically with a message saying so; rows already loaded are kept, and running the dataset again continues. A row that never reported progress at all is failed after 12 hours.
  • A Gateway job that reports no progress for the On-prem job stall limit (default 30 minutes, Setup → Runtime) is failed so it does not block further refreshes.
  • A job that passes the Long-running job warning sends an email and keeps running; one that passes its Time limit is stopped.
  • After a service restart, loads that were running are put back in the queue, and runs that were in flight end as aborted by restart.