Queues

Queues on ThingsBoard Edge guarantee message delivery to the rule engine — messages are processed in order, none are silently dropped, and the rule chain receives each message exactly once. All queues are created on the server and propagated to each connected Edge instance automatically.

For a full reference on queue configuration, see Queues.

Two queue implementations are supported:

Type	Default	Use case
In-memory	Yes	Development and proof-of-concept environments. Not recommended for production — data is lost if the Edge process restarts.
Kafka	No	Production environments where durability and fault tolerance are required.

Queues are configured by the System Administrator on the server and automatically pushed to every connected Edge instance — no assignment or configuration on Edge is needed.

Once propagated, queues are available in Edge rule chains. The default queue is named Main and handles all messages unless a rule chain explicitly switches to another.

Use the checkpoint node to move a message to a different queue mid-chain. For example, you might route alarm messages to a high-priority queue while telemetry continues through Main — giving alarm processing its own delivery guarantee and preventing a telemetry spike from delaying alarms.

Analytics nodes — such as Aggregate stream — also rely on queue configuration to control message ordering and windowing behavior.