blob: 87349aa5c3bb1e842c372e97f00457a9904d1887 [file] [log] [blame]
= Affinity Trait
// Start of autogenerated code - DO NOT EDIT! (description)
Allows constraining which nodes the integration pod(s) are eligible to be scheduled on, based on labels on the node,
or with inter-pod affinity and anti-affinity, based on labels on pods that are already running on the nodes.
It's disabled by default.
This trait is available in the following profiles: **Kubernetes, Knative, OpenShift**.
// End of autogenerated code - DO NOT EDIT! (description)
// Start of autogenerated code - DO NOT EDIT! (configuration)
== Configuration
Trait properties can be specified when running any integration with the CLI:
```
kamel run --trait affinity.[key]=[value] --trait affinity.[key2]=[value2] integration.groovy
```
The following configuration options are available:
[cols="2,1,5a"]
|===
|Property | Type | Description
| affinity.enabled
| bool
| Can be used to enable or disable a trait. All traits share this common property.
| affinity.pod-affinity
| bool
| Always co-locates multiple replicas of the integration in the same node (default *false*).
| affinity.pod-anti-affinity
| bool
| Never co-locates multiple replicas of the integration in the same node (default *false*).
| affinity.node-affinity-labels
| []string
| Defines a set of nodes the integration pod(s) are eligible to be scheduled on, based on labels on the node.
| affinity.pod-affinity-labels
| []string
| Defines a set of pods (namely those matching the label selector, relative to the given namespace) that the
integration pod(s) should be co-located with.
| affinity.pod-anti-affinity-labels
| []string
| Defines a set of pods (namely those matching the label selector, relative to the given namespace) that the
integration pod(s) should not be co-located with.
|===
// End of autogenerated code - DO NOT EDIT! (configuration)
== Examples
* To schedule the integration pod(s) on a specific node using the https://kubernetes.io/docs/concepts/configuration/assign-pod-node/#interlude-built-in-node-labels[built-in node label] `kubernetes.io/hostname`:
[source,shell]
$ kamel run -t affinity.node-affinity-labels="kubernetes.io/hostname in(node-66-50.hosted.k8s.tld)" ...
* To schedule a single integration pod per node (using the `Exists` operator):
[source,shell]
$ kamel run -t affinity.pod-anti-affinity-labels="camel.apache.org/integration" ...
* To co-locate the integration pod(s) with other integration pod(s):
[source,shell]
$ kamel run -t affinity.pod-affinity-labels="camel.apache.org/integration in(it1, it2)" ...
The labels options follow the requirements from https://kubernetes.io/docs/concepts/overview/working-with-objects/labels/#label-selectors[Label selectors]. They can be multi-valuated, then the requirements list is ANDed, e.g., to schedule a single integration pod per node AND not co-located with the Camel K operator pod(s):
[source,shell]
$ kamel run -t affinity.pod-anti-affinity-labels="camel.apache.org/integration" -t affinity.pod-anti-affinity-labels="camel.apache.org/component=operator" ...
More information can be found in the official Kubernetes documentation about https://kubernetes.io/docs/concepts/configuration/assign-pod-node/[Assigning Pods to Nodes].