Skip to content

Understand delivery and recover failures

Open the campaign to inspect recipient delivery state and attempt history. SMTP acceptance means the configured server accepted the message; it does not confirm final delivery to a person’s inbox.

StateMeaning and next action
Ready/queuedWaiting for its due time, worker capacity, or domain readiness
Preparing/sendingAn attempt is being composed or submitted
AcceptedSMTP acceptance has been recorded
RetryA retryable failure has a persisted next attempt time
BlockedProfile or policy needs attention; correct it before retrying
FailedPermanent rejection or exhausted automatic retries; inspect the cause
Unknown/uncertainAcceptance may have happened; inspect the receiving system before deciding
CancelledWork is no longer scheduled for submission

Retryable failures allow eight retries after the initial attempt. Delays grow exponentially from two minutes with bounded positive jitter. Permanent rejection is not automatically retried. Profile connection/TLS/authentication problems can also trigger cooldown behavior.

Expired domain evidence pauses new work without consuming SMTP retries. Correct the domain verification problem, then inspect whether delivery resumes.

A connection can fail after the receiving SMTP server has accepted a message, or persistence can fail after submission. An unknown outcome is not proof of failure. Ruphish does not automatically resend it.

Inspect the attempt’s message identifier, protocol stage, SMTP status if present, and the receiving service’s evidence. Confirm acceptance only when you have evidence. Explicitly choosing to resend an unknown attempt accepts the possibility of duplicate mail. Failed/blocked deliveries have a separate retry action.

Actions check the current attempt generation. If the row changed, refresh before deciding again; do not replay a stale recovery action blindly.

Abandoned preparation can be reclaimed after its lease expires. Interrupted submission can become uncertain instead. Completion prevents further ordinary queued work, but cannot recall an external submission already in progress. A late confirmed acceptance may still settle that attempt.

For installation-wide outages, see operational troubleshooting. For automation, see campaign delivery actions.