Runtime Health and Delivery

Runtime health and delivery

BlokeBot reports whether the host bot and each hosted channel are ready rather than hiding missing authorization or disconnected transports.

Runtime recovery

IRC and EventSub sessions use bounded retry policies. Hosted EventSub channels have their own recovery policy so a channel fault does not have to erase installation-wide state.

Public-chat delivery

Outgoing public chat is persisted before delivery. The queue tracks retryable, terminal, expired, and ambiguous outcomes and retains terminal records for the configured period.

This is durable delivery, not an exactly-once guarantee: a transport can become ambiguous after Twitch may already have accepted a message. BlokeBot avoids inventing certainty in that state.

Alerts

Operational problems can create durable alerts. The dashboard separates active alerts from acknowledged history and records who handled an alert and when.

Useful checks during an incident:

1. Confirm host-bot identity and authorization.
2. Check the selected channel's readiness and authorization state.
3. Inspect active alerts and recent delivery outcomes.
4. Review service logs for transport or persistence failures.
5. Correct the underlying configuration before repeatedly restarting the service.

Policy values are documented in Retry and delivery policies.