Explore ETCD

What is ETCD

https://etcd.io/

• A distributed key-value data store which stores the state of the system and objects
• Use Raft protocol to coordinate the data access in distributed setup
• You can influence the way the API Server is using etcd via a number of options at start-up time. –etcd-servers List of etcd servers to connect with (scheme://ip:port)
• Kubernetes stores its objects in etcd either as a JSON string or in Protocol Buffers(“protobuf” for short) format

Flow from kubectl apply -f pod.yaml to object in etcd

• Get the version from http path (e.g. v1 or v1beta1)
• Create the empty object using Scheme in the right version
• JSON or protobuf decoder converts the payload into object
• Write requests are all applied to ETCD leader node

Flow from kubectl get pods

• Get request is sent to kube-apiserver
• kube-apiserver uses the ETCD client lib to talk to etcd cluster. Implementation
• ETCD3 client performs a quorum read by default to make sure always the latest value gets returned. ETCD2 needs to pass in a flag quorum-read=true to achieve the quorum read

Lab

• Using footloose to build the vm-like clusters
• Using Ansible to configure the clusters
  • Install pre requisites
  • Install etcd
• Bring up a kind cluster using the external etcd cluster

References

• https://blog.scottlowe.org/2020/04/23/setting-up-etcd-with-etcdadm/
• https://github.com/weaveworks/footloose
• https://blog.scottlowe.org/2020/04/23/setting-up-etcd-with-etcdadm/