Control Failure: failed_when & ignore_errors

ignore_errors does not hide the failure; output still reads FAILED. It only stops the failure from halting later tasks.

Use failed_when to set your own failure condition. The task fails only when your expression is true.

failed_when usually inspects a registered result: its rc, stdout or stderr. Capture output first, then judge it.

Set failed_when to false and the task can never fail, even on a nonzero code. Cleaner than ignore_errors for known-noisy commands.

failed_when can take a list of conditions joined with and, or a compound expression with or for richer rules.

The related changed_when controls the changed status instead of failure, useful to mark a command task as never-changed.

Reach for failed_when when you can describe success precisely; use ignore_errors only for truly optional, best-effort steps.

Instead of ignoring errors, let a real failure trigger a rescue block. You get recovery instead of a silent skip. 🤝

Both failed_when and ignore_errors live on a single task, not a whole play. Each task decides its own failure behavior.

failed_when defines exactly when a task fails, while ignore_errors lets the play continue past a failure. You own the rules. ✅