Pods
- Pod ensures containers are in running state, if container is exited, pod restarts the container
- Two errors:
- crashloop backoff
- OOM killed
- Application run is a container but wrapped as pod.
- IF we want to scale application, we scale pods not containers.
- To control pods we have resources such as
- replicasets (ReplicationController)
- Daemonsets
- Deployments
Labels
- Labels are key value pairs
- Labels are used as query mechanism in k8s
Playing with labels
-
Watch classroom recording
- Labels can be queried using selectors
- selectors are of two types
- Writing labels in manifest. Labels are part of metadata
Replica sets
- Replica set ensure the number of pods is maintained
- In Replica set we specify
- pod
- number of pods
- Replica sets identify pods with labels with both equality and set based selectors
Example 1: Lets create a replicaset with 4 nginx pods
- Spec
---
apiVersion: apps/v1
kind: ReplicaSet
metadata:
name: example-1
labels:
purpose: learning
spec:
replicas: 4
minReadySeconds: 2
selector:
matchLabels:
app: web
template:
metadata:
labels:
app: web
env: dev
spec:
containers:
- name: nginx
image: nginx:1.14.2
resources:
limits:
memory: "128Mi"
cpu: "500m"
ports:
- containerPort: 80
- replicasets creates pods mentioned in the template section if the desired count vs actual count is not same.
- Replica sets calcualtes desired count on the basis of matching labels
- Exercise:
- Try finding a command to run the repliaset with 2 httpd pods
Exercise – docker
- write a multistaged docker file using distroless java for spring petclinic with minimal size possible and no major vulnerabilities.
