Restart Dataplane
The Restart action sends a graceful shutdown signal to the Dataplane, which triggers an immediate restart. Use it when the engine is unhealthy or wedged and refreshing the status has not brought it back.
Permissions
Only users with the Admin role can restart the Dataplane. The button is not shown to Managers. See the Permissions page for the full matrix.
Before You Restart
The confirmation dialog summarizes the impact in three areas.
| AREA | IMPACT |
|---|---|
| Immediate Failures | Connection tests, schema discovery, dry runs, source record views, and computed asset validation fail until the engine is back. |
| In-Flight Operations | Running operations are interrupted, then automatically restarted once the engine is back: Profile, Scan, and Promote continue with the work that had not been processed yet, while Sync, Export, and Materialize start over from the beginning. External Scan is aborted instead. See the Runs FAQ for details, including the automatic restart limit. |
| Unaffected | Flows, user management, teams, tags, quality check and anomaly configuration, scheduling, and integrations remain fully available. |
Two more things are worth knowing before you confirm.
Scheduled operations are not lost
Scheduling configuration stays available throughout. An operation that triggers while the engine is restarting is queued and picked up automatically once it is back.
Capture diagnostics first
A restart discards the engine's live state, and the platform keeps no diagnostics reports. If you want to review what led to the restart, capture and download a report before confirming. See Download a Diagnostics Report.
Steps
Step 1: Click the Restart button in the top-right corner of the Dataplane card.
Step 2: A confirmation dialog appears, listing what happens during a restart.
Step 3: Click Restart to confirm.
Step 4: A success message appears and the page shows loading indicators while the engine comes back.
Note
While the engine restarts, the Components card reports the Dataplane unhealthy and the fields on its Configuration subtab display --. The page re-checks the status automatically about 20 seconds after you confirm. How long the engine actually takes depends on your deployment, so if it is still unhealthy, use Refresh Status Summary to check again.