KUBERNETES: Understand and use probes effectively

Introduction

Let's talk about the Kubernetes probes and why they matter in their implementations. By administering production container applications, even small optimizations can have enormous benefits.

With the aim of reducing implementation times, making their applications react better to the event scale and manage the health of the capsules in operation requires adjusting the management of the life cycle of its container. This is exactly the reason why the appropriate configuration, and implementation, of the Kubernetes probes is vital for any critical implementation. They help their cluster to make intelligent decisions about traffic routing, reset and the allocation of resources.

Correctly configured probes drastically improve the reliability of their application, reduce the inactivity time of implementation and handle unexpected errors with grace. In this article, we will explore the three types of probes available in kubernettes and how to use them together helps configure more resistant systems.

Fast fast

Understanding exactly what each probe and some common configuration patterns do is essential. Each of them has a specific purpose in the life cycle of the container and when used together, they create a solid rock frame to maintain the availability and performance of its application.

Home: Optimization of starting times

Starting probes are evaluated once when a new POD is turned due to an extension event or a new implementation. It serves as a guardian of the rest of the container verifications and the fine adjustment will help its applications to better manage the greatest load or degradation of the service.

Configuration example:

startupProbe:
  httpGet:
    path: /health
    port: 80
  failureThreshold: 30
  periodSeconds: 10

Key control:

• Keep periodSeconds Low, so that the probe is triggered often, quickly detecting a successful deployment.
• Increase failureThreshold at a high enough value to accommodate the worst start time.

The start probe will verify if your container has begun by consulting the configured route. He will also do it Stop the activation of life probes and preparation Until you succeed.

Livials: Detection of dead containers

Your advantage probes answer a very simple question: “Does this capsule continue to work properly?” If not, K8S will restart it.

Configuration example:

livenessProbe:
  httpGet:
    path: /health
    port: 80
  periodSeconds: 10
  failureThreshold: 3

Key control:

• Since K8S will completely restart your container and turn a new one, add a failureThreshold to combat intermittent abnormalities.
• Avoid using initialDelaySeconds As is too restrictive, use a starting probe.

Keep in mind that a failure probe will break down your capsule currently in operation and turn a new one, so prevent it from being too aggressive, that is for the next one.

Preparation: Unexpected Errors Management

The preparation probe determines whether to start, or continue, receive traffic. It is extremely useful in situations in which its container lost the connection to the database or is too used and should not receive new applications.

Configuration example:

readinessProbe:
  httpGet:
    path: /health
    port: 80
  periodSeconds: 3
  failureThreshold: 1
  timeoutSeconds: 1

Key control:

• Since this is your first guard to stop traffic to unhealthy objectives, make the probe aggressive and reduce the periodSeconds .
• Keep failureThreshold At least, you want to fail quickly.
• The waiting time period must also be maintained to handle slower Containers
• Give the readinessProbe Enough time to recover having a longer race livenessProbe .

Additional reading:
Living, preparation and starting probes
Configure life, preparation and start probes