Section 8 · Process Automation · 20% of exam
8.3 Automation troubleshooting, testing, and fault handling
Exam objective
Given a scenario, identify appropriate tools and methods for troubleshooting declarative automation. Troubleshooting begins by reproducing the failure with a specific user, record, action, and timestamp. Capture the exact error text and determine whether the failure occurs in a screen flow, a record save, an approval, or a scheduled path. Check active flow versions, entry criteria, record type, field values, user permissions, validation rules, and related automation. A flow may not run because its conditions were false, because a different version is active, or because the transaction failed earlier in the save process.
Flow Builder debug tools can help test a flow with selected inputs and inspect decisions and assignments, but a debug run may not perfectly reproduce permissions, triggering context, or all transaction behavior. Use Flow error emails and error details to locate the failing element and fault. Debug logs can show which automation executed, while Setup Audit Trail can show when a flow or configuration changed. Select tools based on the question: execution path, failure, or setup history.
Create a test matrix that covers positive and negative paths, boundary values, user roles, and related-record states. For a record-triggered flow, test create and update events, when conditions are met and not met, and when fields change from one value to another. Test bulk changes and partial failures where relevant. For a screen flow, test navigation, required inputs, fault messages, accessibility, and data access. For approval automation, test each decision path and record editability while pending.
Fault handling should help the user recover without leaking sensitive details. Use fault connectors where appropriate to display a clear message or create a monitored error record. Decide who owns failed interviews and how they are retried. Avoid swallowing an error and presenting success when an update did not happen. A failed external action may need a retry queue or integration monitoring rather than a user repeating the entire transaction and causing duplicates. Check dependencies and limits when a flow behaves differently in bulk. Large collections, loops that perform repeated operations, record locking, CPU or query limits, and competing automation can cause failures. Use collection operations and bulk-friendly design. Keep decisions readable and move reusable logic into subflows when that improves maintainability, while recognizing that subflows add dependencies and complexity. Use descriptions and clear element labels so another administrator can diagnose the process later.
Before changing a production flow, preserve the current active version and understand the release process. Clone or create a new version, test in a sandbox with representative data, validate permissions and entry criteria, and deploy through the team's controlled method. After activation, monitor error rates and business outcomes. If a flow is deactivated to stop harm, document the records and work that may have been missed and how they will be reconciled. Exam focus: match diagnostic evidence to the likely cause. A missing run suggests entry criteria or activation; an element failure suggests permissions, data, limits, or a fault; a changed outcome after a deployment suggests version or configuration history. Know that a debug run is not a substitute for testing under the target user and transaction context. Choose fault handling that preserves a clear audit and recovery path.
Build tests around business state transitions, not only field values in isolation. The transition often determines whether automation should run.
Prioritize evidence before making changes. Save the active flow version, note the triggering record values, and capture the exact error, then change one variable at a time in a sandbox. When the error is intermittent, compare successful and failed records for data, user, timing, and transaction volume. Establish an alert threshold for repeated failures and a runbook for retrying safely. After repair, inspect records that failed while the process was unavailable and reconcile missed work. Test both the original defect and adjacent paths to avoid a narrow fix that creates a new failure. Practical scenario
A record-triggered flow fails only for a non-admin user when it creates a related record. Compare the user's object and field access, the flow's run context, required fields, and validation rules on the target object. Reproduce with a representative account, inspect the flow error details and debug log, and test the fix in a sandbox. If system context is necessary, limit the flow's exposed inputs and outputs and ensure it cannot be used to retrieve unauthorized data. Extra concept
Flow versions are configuration artifacts with activation state. A user can be testing a draft version while production uses another active version. Confirm the active version in the affected org and include activation checks in deployment validation.
Key takeaways
- Check active versions, entry criteria, user context, data, and related automation.
- Test positive, negative, boundary, bulk, and security paths.
- Provide monitored failures and a safe retry or reconciliation process.
- Official references: https://help.salesforce.com/s/articleView?id=005298969&language=en_US&type=1 https://help.salesforce.com/s/articleView?id=sf.flow.htm&type=5 https://help.salesforce.com/s/articleView?id=sf.code_debug_log.htm&type=5 https://trailhead.salesforce.com/content/learn/superbadges/superbadge_record_triggered_flow
