Pods are temporary, or we can say it’s disposable entities. Therefore, they do not self-heal themselves.

Do not get confused with restarting a container in a Pod with restarting a Pod. If a Pod gets terminated for any reason, we cannot bring it back but create an identical one. We cannot reschedule the same Pod on another node.

Pods pass through the defined lifecycle. For example, a Pod can be in one of the following phases:

Pending — A Pod will be in this phase when it gets created and accepted by the cluster but is waiting for all its containers to get ready and running. This will also include the waiting time to get scheduled onto a node and the time spent downloading the container images.

Running — All its containers have been created when placed onto a Node. Moreover, when it’s one or more container is still getting started, running, or restarting.

Succeeded — When it’s all containers have been successfully terminated and will not be restarted.

Failed — When all its containers are terminated, and at least one has failed.

Unknown — When a Pod state cannot be derived. This could happen when having trouble communicating with a Node. 

I'm sharing more