Labels in Kubernetes are the key-value pair that does not have any predefined meaning and that can be attached to any objects or any resources such as Pod, Deployment, or Node which is mainly used to identify and organize Kubernetes resources. We could add an 'environment' label to a set of pods with the value 'production' and 'development' to other sets of pods. We could then use label selectors to select only the pods in the 'production' environment.

The label is defined under the metadata field with 'labels' as a field. Multiple labels can be added to a single object.

Let us take an example of creating a label.

vi label-pod.yml

kind: Pod
apiVersion: v1
metadata:
  name: label-first-pod
  labels:
    env: developments
    class: pods
spec:
  containers:
    - name: label-container
      image: ubuntu
      command: ["/bin/bash","-c","while true; do echo this-is-lable-pod; sleep 5; done"]

Now apply the above configuration

kubectl apply -f label-pod.yml

Get labels on all the pods

kubectl get pods --show-lables

If we want to add a label to an existing pod

kubectl label pods <pod-name> <key=value>

In our case, it would be something like

kubectl label pods label-first-pod myname=sandeep

Here myname will be added as a label with the value of sandeep in the pod label-first-pod

To see the labels of all the pod

kubectl get pods --show-labels

Now list pods matching a label

kubectl get pods -l env=developments

List pods where the 'developments' label is not present

kubectl get pods -l env!=developments

Now delete the pod

kubeclt delete -f label-pod.yml

Labels Selectors

The label selectors are the filter to narrow down resources based on their labels. As the name suggests, it is used to select specific pods to perform the action. It may be to specify pods to be included in a service, to be managed by deployment or rollout, or to be scaled by a horizontal pod scaler. The API currently supports two types of selectors

1. Equality Based: They are the kind of selectors that matches resources that have a label with a specific key and value. For example, the selector “env=development” would match any resource with the label “env=development”. We can also specify multiple selectors by separating them with a comma. For example, “env=development, myname=sandeep”.

2. Set Based: The set selectors match multiple values based on a set of values for a key. There are mainly three types of set-based selectors:

   1. {key} in {values}. For example ‘environment in (production, development). It matches with a label ‘environment’ that has either production or development. It matches either production or development.

   2. {key} notin {values}. For example ‘environment notion (production, development). It matches neither production nor development.

   3. {key} For example, ‘environment’, matches resources with a label ‘environment’ that has any value.

Example of Equality Based selector

kubectl get pods -l class=pods, myname=sandeeep

Example of a Set Based selector

kubectl get pods -l 'env in (development, testing)'

Node selector

When we create a pod, the master node randomly assigns a node to it. If we want to assign a specific node to a pod then we use 'nodeSelector'. One use case for selecting labels is to constrain the set of nodes onto which a pod can schedule i.e. you can tell apod to only be able to run on a particular node.

Generally, such constraints are unnecessary as the scheduler will always automatically do a reasonable placement, but in certain circumstances, we might need it.

We can use labels to tag nodes and use label selectors to specify the pods to run on specific nodes.

For this purpose, first, we give a label to the node, and then we use a node selector for the pod configuration.

For labeling the nodes

kubectl get nodes

kubectl label nodes ip-xxx-xxx-xxx-xxx <label>

For example, we can do like

kubectl label nodes ip-172-31-34-55 hardware=t2-medium

Now we have labeled the node with 'hardware=t2-medium'. To create the pod in the specified node

vi node-pod.yml

kind: Pod
apiVersion: v1
metdata:
  name: node-pod
  label:
    env: development
    class: pods
spec:
  containers:
    - name: container-node
      image: ubuntu
      command: ["/bin/bash","-c","while true; do echo this-is-lable-pod; sleep 5; done"]
  nodeSelector:
    hardware: t2-medium

kubectl apply -f node-pod.yml

To see where exactly the pod is running

kubectl get pods -o wide

We can also use multiple node selector criteria. For example

Label the node first

kubectl label nodes ip-xxx-xxx-xxx-xxx size=small region=us-west

vi node-pod-second.yml

kind: Pod
apiVersion: v1
metdata:
  name: node-pod-second
  label:
    env: development
    class: pods
spec:
  containers:
    - name: container-node
      image: ubuntu
      command: ["/bin/bash","-c","while true; do echo this-is-lable-pod; sleep 5; done"]
  nodeSelector:
    hardware: t2-medium
    size: small
    region: us-west

Here we have applied multiple node selectors like size and region.

kubectl apply -f node-pod-second.yml