Labeler runs in your Kubernetes cluster and applies labels to nodes based on that Tags are set on an instance in a cloud provider. Some cloud provider specific labels already are applied to a node when using the cloud provider subsystem. This continues that idea and allows user specific tags on instances to also get added as node labels.
See the example config in the deploy/ directory and modify values as needed or desired.
Deploy to cluster:
kubectl apply -f deploy/
or
kubectl apply -f <path to github url of file>
Remove from cluster:
kubectl delete -f deploy/
or
kubectl apply -f <path to github url of file>
View output logs:
kubectl logs deployment/labeler -n kube-system
The configuration is a yaml formatted file that is saved as a config map and then read in at deploy time. Below is an exmaple configuration file.
namespace: labeler.io
region: us-east-2
provider: aws
excludes:
Name: true
aws:autoscaling:groupName: true
kubernetes.io/cluster/<clusterName> : true
namespace - Labels can be namespaced in kubernetes. This takes the form of namespace/key=value. If you wish to namespace your labels you can provide that string here.
region - Specify the region of your cloud provider your nodes are running in.
provider - the name of your cloud provider. (aws, gcp, azure)
excludes - tags that may be applied to an instance that you wish to exclude from being applied as labels. In the example some common starter values are provided.
- Currently aws is the only provider supported.
- If a tag on an instance changes after it has joined the cluster. Labeler will not update the node labels. It only applies labels when a node joins a cluster.
Dependencies are managed by https://github.com/golang/dep , however the vendor dir is commited with the depdencies for easy building with out dep.
To build just the binary go build will work.
To build the docker image you can use the included multi-stage build Dockerfile docker build -t <name of image> .