Every so often I see a timeout error when sending a message to a Service Bus from a logic app. Usually the Service Bus respond in milliseconds.
What strategies are there for assuring a message makes it into a Service Bus Queue/Topic to be processed in the case of intermittent outages of the Service Bus?
Is my only option to parallel this architecture in another region and have fallbacks between them?
You can enable and configure retries for the action that creates the Service Bus message in the producer Logic App. This will significantly reduce the potential for data loss.
By enabling message creation attempt retries, you need to consider scenarios where message creation succeeded but the Logic App action results in a failures. Those situations could include timeouts; gateway connection failures; and other transient errors. To prevent duplicate Service Bus messages, you can enable duplicate detection and include a workflow-run-unique MessageId. When considering your options for that value, don't forget about the possibility of workflow resubmits.
