DataFlow Failed trigger

The DataFlow Failed trigger is raised when the configured error count threshold is reached within the specified time window for the selected dataflows. It does not fire on every failure, only when the threshold condition is met.

A dataflow is considered failed when one or more nodes throw an exception, regardless of whether the error is handled later in the flow. Handling failed rows or exceptions inside the dataflow does not prevent the dataflow from being marked as failed for the purpose of this trigger.

Configuration

Setting	Description
Dataflows	The dataflow(s) this trigger should respond to. One or more can be selected. If no dataflow is selected, the trigger will never be raised.
Error count	How many times a dataflow must fail before the trigger is raised.
Time window	The time period in which the error count must be reached.
Window mode	Determines how failures are counted within the time window: Fixed or Rolling.
Cooldown	A period after the trigger fires during which it cannot be triggered again.

Window mode

Fixed

• Counting starts when the first error occurs.
• All failures are counted until the time window expires.
• As soon as the error count threshold is reached, the trigger is raised.
• After firing, a new time window starts.

Rolling

• On each failure, the system checks how many failures occurred in the past time window.
• If the number of failures in that window exceeds the threshold, the trigger is raised.
• If another failure occurs later and the count is still above the threshold, the trigger is raised again.
• Unlike fixed mode, the count does not reset automatically.

Both window modes respect the configured cooldown.

When to use it

This trigger is typically used to:

• Notify administrators of failing dataflows
• Start reporting or alerting dataflows
• Trigger recovery or cleanup processes

Tips

• In complex environments, have at least one dataflow with a DataFlow Failed trigger to ensure administrators are notified when issues occur.
• Use error count and cooldown together to prevent alert fatigue when a dataflow fails repeatedly.
• Prefer rolling window mode when detecting recurring instability, and fixed window mode when monitoring bursts of failures.