Skip to content
DocsOpen Hub
Test & operate

Read logs and execution history

On this page

Cloudgate records gateway requests, outbound HTTP calls, and workflow execution details. Choose the log that matches the question, then narrow the environment and time range.

Choose a view#

View What it explains
API Action Logs Incoming gateway request, action, status, and overall execution
HTTP Requests Calls the workflow made to external services
Workflow/session detail Node sequence, durations, results, and errors
Schedule history/state Background job timing and execution context
Warning System Operational signals that need attention

An action is an API definition. An Action Log is one call to that definition.

Filter before comparing#

Set the environment and reporting period first. Use the action/controller, response status, and available search fields to reduce noise. Include the timezone when sharing a timestamp with another operator.

Correlate an outbound failure with its parent session. A failed provider call and the overall failed action can be two records for the same business operation, so avoid treating all rows as independent incidents.

Understand redaction#

New logs redact recognized credentials and secret-selector values before persistence. Those values cannot be recovered through an export or an unmask operation. The workflow itself still receives its actual execution values.

This is not general personal-data classification. Ordinary names, emails, or custom opaque fields may still be present. Inspect evidence before sharing it outside its normal audience.

Retention and export#

Availability depends on logging configuration, retention, and archived evidence. Session keys are temporary execution history, not durable application storage. Older logs may have different detail and redaction behavior from new runs.

Use the available export controls for a targeted set of records. Include the associated version and environment in an investigation. Missing or expired evidence cannot be recreated simply by opening the latest graph.

For a live failure, follow debugging; for a longer investigation, attach selected ticket evidence.

Need to investigate a problem? Start with troubleshooting