Track Kubernetes Changes¶

Let's track changes to Kubernetes objects using Robusta. Notifications will be sent to a Sinks, like Slack or MSTeams.

In this tutorial you will:

• Specify which Kubernetes object to track

• Filter out noisy changes and only track certain YAML fields

• Send a diff of exactly what changed

Defining a Playbook¶

Add the following YAML to the customPlaybooks Helm value:

customPlaybooks:
- triggers:
    - on_deployment_update: {}
  actions:
    - resource_babysitter:
        omitted_fields: []
        fields_to_monitor: ["spec.replicas"]

Then perform a Helm Upgrade.

Testing Your Playbook¶

Scale a deployment that exists in your cluster:

Run the following YAML files to simulate a deployment change

kubectl apply -f https://raw.githubusercontent.com/robusta-dev/kubernetes-demos/main/crashpod/healthy.yaml
kubectl apply -f https://raw.githubusercontent.com/robusta-dev/kubernetes-demos/main/crashpod/broken.yaml

A Robusta notification will arrive in your configured sinks, showing exactly what changed in the deployment:

How it Works¶

We configured a custom playbook with the trigger on_deployment_update. This trigger fires whenever Kubernetes Deployments are updated.

The trigger fires on all Deployment changes, even uninteresting changes to the Deployment's status performed by Kubernetes itself on static clusters.

The action is resource_babysitter action, which itself performs further filtering and ignores uninteresting changes. This action is a little unusual - most of the time triggers perform all the filtering and actions act on everything that reaches them.

In the future we're planning to improve the trigger mechanism. Filters like fields_to_monitor will move from the resource_babysitter into triggers like on_deployment_update <on_deployment_update>.

Adding Change Routing¶

To send change notifications to a specific sink instead of all sinks, you can choose between two methods:

1. Use Sink Matchers

2. Explicitly specify a sink in the playbook

Here is the latter method:

customPlaybooks:
- triggers:
  - on_deployment_update: {}
  actions:
  - resource_babysitter:
      omitted_fields: []
      fields_to_monitor: ["spec.replicas"]
  sinks:
  - some_sink_name

Cleanup¶

Remove this playbook from customPlaybooks and perform a Helm Upgrade.