Message Queue Trigger

A Message Queue Trigger binds a message queue topic to a function so that each message on that topic invokes the function with the message as the request body.

The current architecture is built on KEDA (Kubernetes Event-Driven Autoscaling). For each MessageQueueTrigger of kind keda, Fission provisions a KEDA ScaledObject and a connector Deployment that consumes the queue, calls your function through the Router, and scales with the depth of the queue, including scaling to zero when the topic is idle.

Message Queue Trigger is an optional component. It runs as the mqtrigger service inside fission-bundle, and the KEDA scaler manager runs alongside it to reconcile ScaledObject resources. Using a KEDA-based trigger requires KEDA (v2.20) installed in the cluster.

How it works

flowchart LR
  queue["Message Queue Topic"] -->|"<b>3.</b> reports lag"| scaler["KEDA Scaler"]
  subgraph k8s["Kubernetes Cluster"]
    mqt["MQT Scaler Manager"] -->|"<b>1.</b> creates"| scaledobject["KEDA ScaledObject"]
    scaledobject -->|"<b>2.</b> drives"| scaler
    scaler -->|"<b>4.</b> scales 0..N"| connector["Connector Deployment"]
    queue -->|"<b>5.</b> delivers message"| connector
    connector -->|"<b>6.</b> HTTP POST message"| router["Router"]
    router -->|"<b>7.</b> forwards request"| fnPod["Function Pod"]
    connector -->|"<b>8.</b> on success / error"| respTopics["Response / Error Topic"]
  end

  class queue,respTopics user
  class mqt,scaledobject,scaler,router fission
  class connector,fnPod pod
  classDef user fill:#ffffff,stroke:#94a3b8,color:#1f2a43
  classDef fission fill:#e8f0fe,stroke:#2d70de,color:#1f2a43
  classDef pod fill:#e6f7f1,stroke:#11a37f,color:#1f2a43,stroke-dasharray:5 3

1. You create a MessageQueueTrigger CRD with mqtkind: keda, naming the queue type, topic, target function, and scaling settings.
2. The MQT scaler manager reconciles the trigger and creates a KEDA ScaledObject (and, when a secret is referenced, a TriggerAuthentication) plus a connector Deployment for that queue type.
3. KEDA polls the queue, and based on its depth scales the connector Deployment between the configured minimum and maximum replica counts, scaling to zero during idle periods.
4. A connector replica consumes messages from the topic and sends each message as an HTTP POST to the Router, which routes it to the function.
5. If a response topic is configured, the function’s successful output is published to it; if an error topic is configured, failed invocations are published there.

The connector Deployment is pointed at the Router’s internal URL through the HTTP_ENDPOINT environment variable, so messages reach functions over the cluster-internal listener.

Supported queue connectors

The KEDA-based trigger validates the following queue types (spec.messageQueueType):

See the Message Queue Trigger (KEDA) usage guides for connector-specific setup.

Scaling and behavior knobs

These fields on the MessageQueueTrigger spec map to KEDA ScaledObject settings:

• pollingInterval - how often KEDA checks the queue and adjusts the connector Deployment.
• cooldownPeriod - how long to wait after the last active reading before scaling back to zero.
• minReplicaCount / maxReplicaCount - the bounds KEDA scales the connector Deployment between.
• metadata - scaler-specific trigger metadata passed through to KEDA.
• secret - the name of a secret used to build a KEDA TriggerAuthentication.
• respTopic / errorTopic - topics for function output and error responses.
• maxRetries - how many times the connector retries a failed invocation.

Legacy built-in trigger

The original built-in (non-KEDA) message queue trigger is legacy. New triggers should use mqtkind: keda.

Before KEDA integration, Fission ran message queue consumers directly inside the mqtrigger component. That built-in path now retains Kafka in-tree only, and it does not provide event-driven autoscaling or scale-to-zero. The error topic in the legacy path is supported by a limited set of queues (NATS, Kafka, and Redis lists). Prefer the KEDA-based trigger for all supported queues, broader connector coverage, and queue-depth autoscaling.

Related

• Router - the component connectors post messages to.
• Executor - how the invoked function’s pods are managed.
• Message Queue Trigger (KEDA) usage guides - per-connector setup.