Kubernetes Cluster Service
JVM since3.10.0 Native since3.10.0
Provides a Kubernetes implementation of the Camel Cluster Service SPI
Maven coordinates
Or add the coordinates to your existing project:
<dependency>
<groupId>org.apache.camel.quarkus</groupId>
<artifactId>camel-quarkus-kubernetes-cluster-service</artifactId>
</dependency>Check the User guide for more information about writing Camel Quarkus applications.
Additional Camel Quarkus configuration
Having only a single consumer in a cluster consuming from a given endpoint
When the same route is deployed on multiple pods, it could be interesting to use this extension in conjunction with the Master one. In such a setup, a single consumer will be active at a time across the whole camel master namespace.
For instance, having the route below deployed on multiple pods:
from("master:ns:timer:test?period=100").log("Timer invoked on a single pod at a time");As a result, a single consumer will be active across the ns camel master namespace. It means that, at a given time, only a single timer will generate exchanges across the whole cluster. In other words, messages will be logged every 100ms on a single pod at a time.
The kubernetes cluster service could further be tuned by tweaking quarkus.camel.cluster.kubernetes.* properties.
| Configuration property | Type | Default |
|---|---|---|
Whether the camel master namespace leaders should be distributed evenly across all the camel contexts in the cluster. |
|
|
The cluster service ID (defaults to null). |
| |
The URL of the Kubernetes master (read from Kubernetes client properties by default). |
| |
The connection timeout in milliseconds to use when making requests to the Kubernetes API server. |
| |
The name of the Kubernetes namespace containing the pods and the configmap (autodetected by default). |
| |
The name of the current pod (autodetected from container host name by default). |
| |
The jitter factor to apply in order to prevent all pods to call Kubernetes APIs in the same instant (defaults to 1.2). |
| |
The default duration of the lease for the current leader (defaults to 15000). |
| |
The deadline after which the leader must stop its services because it may have lost the leadership (defaults to 10000). |
| |
The time between two subsequent attempts to check and acquire the leadership. It is randomized using the jitter factor (defaults to 2000). |
| |
Service lookup order/priority (defaults to 2147482647). |
| |
The name of the lease resource used to do optimistic locking (defaults to 'leaders'). The resource name is used as prefix when the underlying Kubernetes resource can manage a single lock. |
| |
The lease resource type used in Kubernetes, either 'config-map' or 'lease' (defaults to 'lease'). |
| |
The labels key/value used to identify the pods composing the cluster, defaults to empty map. |
|
Configuration property fixed at build time. All other configuration properties are overridable at runtime.