🔀 Continue on Failure in Conditional Flows

When you migrate a Legacy Flow to a Conditional Flow using the Migration Guide, the Continue on Failure behavior does not carry over automatically. This page explains why and shows how to reproduce it with conditions.

Known issue

Conditional Flows have no Continue on Failure toggle. In Legacy Flows, enabling Continue on Failure on a task made the task finish with a warning status instead of failing the whole Flow, and could send a warning notification. Because Conditional Flows have no Continue on Failure toggle, this behavior is not migrated automatically — you have to rebuild it manually using conditions.

Workaround: reproduce continue-on-failure with conditions

In Conditional Flows, you recreate this behavior with a condition that branches on a task’s job status. The example below starts from a first phase named Extract Data that contains two tasks: a GitHub extractor (My GitHub) and a Google Sheets extractor (Vouchers and Global Bars).

1. Start with the Extract Data phase containing the GitHub and Google Sheets extractor tasks.

   

2. Add a condition to the GitHub extractor so that, when it finishes with an error, the Flow sends a notification and then continues. Set the condition to IF My GitHub Job Status Is Failed, THEN Continue To the Existing Phase Notification; in the ELSE branch, Continue To the Existing Phase Phase 2.

   

3. Add a Notification task to the Notification phase and configure it — for example, a name, a message, and an email destination.

   

4. Add the task where the Flow continues to Phase 2. In this example it is a Snowflake transformation named Set variable.

   

5. Connect the output of the Notification phase so it also continues to Phase 2. Click the plus icon (+) under the Notification phase and, under Manage Path, choose Connect Path. Then select Phase 2 from the drop-down.

   

6. The resulting Flow branches on the GitHub extractor failure into the Notification phase and rejoins Phase 2.

   

Caution

Any task that does not have a condition on the failed status — such as the Google Sheets extractor in this example — still fails the whole Flow if it finishes with an error. Add a failed-status condition to every task that should be allowed to continue on failure.