Installation administrators can use analytics and retained call records to identify failures, request volume and timing. Connection users should contact the operator when they need evidence that is not included in their client interface.
Investigate a failed workflow#
Narrow the time range, server, tool and result status where filters are available. Compare the failing call with the workflow's expected account and inputs. Distinguish a client-authentication failure from an upstream-service failure.
Use request counts, error patterns and timing to decide whether the issue is a bad input, revoked access, throttling or a slow dependency. A graph alone does not prove that a write succeeded; verify the resulting record.
Decide what to retain#
Call metadata and full response-body logging are different settings. For privacy-sensitive workflows, avoid retaining bodies unless they are necessary and authorised. Check the current logging options on the actual installation and define appropriate retention.
The documentation site does not display customer call records. A public help page is never the place to publish operational logs or private configuration.
Redaction and audit records#
The platform includes controls for redaction and audit-related evidence. Verify that the configured rules cover your real data shapes. A redaction option does not prove that every field, attachment or upstream error is safe to share.
Keep audit evidence and privacy retention requirements consistent. Do not erase logs or records simply to make a dashboard look clean. Handle necessary deletion through the agreed retention and recovery process.
Share a small diagnostic packet#
Provide the timestamp and timezone, connection reference, tool name, error code and expected result. Inspect any export for keys, personal information and unrelated customer records. Replay only when the operation is safe and authorised; see testing.