Ensuring applications are alive and healthy
By default, Kubernetes ensures that the desired state of applications in a cluster is reached. It will restart and recreate failed containers when a process exits or a node fails. However, that might not be enough to tell if the application running inside the pod is healthy. In order to ensure that the workloads are alive and healthy, Kubernetes implements the concept of probes.
Probe
A probe is a diagnostic that is performed by a Kubernetes kubelet on a container. A diagnostic can be an arbitrary command executed inside a container or TCP probe, or an HTTP request.
Kubernetes offers three types of probes, as shown in the following list:
- Liveness: Ensures that a process in a container is alive and, if not, restarts the container. For the case when the application catches a deadlock, restarting the container usually helps to make the application more available despite bugs.
- Readiness: Ensures that the application is ready...
