Customer Runbooks

These runbooks cover customer actions in the hosted service. Gnok handles cluster changes, worker restarts, and infrastructure incidents.

A query is stuck

Find the query in Studio and request Cancel. Record its query ID, SQL, and start time. After cancellation or a lost connection, inspect the final status. For a write, verify the affected data before resubmitting it.

Service access is interrupted

Record the failing operation and error reference. Preserve drafts and avoid repeatedly submitting writes. Email support@gnok.io with the timestamp and your organization's sign-in name. When access returns, check history and target state before retrying incomplete work.

A team member needs different access

An authorized administrator should review the user's roles and object grants, make the scoped change, and verify the intended action using the user's permissions. See RBAC and user lifecycle.

Usage is higher than expected

Review usage and query history for expensive or repeated workloads, scheduled work, and idle compute where applicable to your billing mode. Use authorized warehouse and resource-monitor controls; confirm account rates and limits separately. See billing.

An AI or scheduled workflow stops

Check its run history, input availability, the organization's AI budget and rollout policy (Studio Operations → AI Governance), and execution identity grants. Restore only the access it needs. Verify the output and progress after recovery; registration or scheduling alone does not prove processing completed.

For more detailed diagnosis, see troubleshooting.