Skip to content

This is the Fluentd input plugin which queries Kubernetes API to collect Kubernetes objects (like Nodes, Namespaces, Pods, etc.)

License

Notifications You must be signed in to change notification settings

peter-moffatt/fluent-plugin-kubernetes-objects

 
 

Repository files navigation

Codeship Status for splunk/fluent-plugin-kubernetes-objects

fluent-plugin-kubernetes-objects

Fluentd input plugin to collect objects in a kubernetes cluster.

This input plugin can collects data in two ways:

  • pull: it collects all available objects at a interval by calling the list APIs.
  • watch: it collects only the new objects when they show up by calling the watch APIs.

Installation

See also: Plugin Management.

RubyGems

$ gem install fluent-plugin-kubernetes-objects

Bundler

Add following line to your Gemfile:

gem "fluent-plugin-kubernetes-objects"

And then execute:

$ bundle

Configuration

Example

<source>
  @type kubernetes_objects
  tag k8s.*

  <pull>
    resource_name nodes
  </pull>

  <pull>
    resource_name pods
    namespace default
  </pull>

  <watch>
    resource_name events
  </watch>

In this example, it will pull all the nodes , and all the pods in default namespace, and also watch events. And all those data will be tagged with prefix "k8s.".

Parameters

@type (string) (required)

This must be kubernetes_objects.

tag (string) (optional)

The tag of the event.

* can be used as a placeholder that expands to the actual resource name (watch objects will have .watch suffix). For example, if you set

tag k8s.*
<pull>
  resource_name pods
</pull>
<watch>
  resource_name events
</watch>

The tag for pods data will be k8s.pods, while the tag for events will be k8s.events.watch.

Default value: kubernetes.*.

kubernetes_url (string) (optional)

The URL to the kubernetes API endpoint. By default, it will read from environment variables KUBERNETES_SERVICE_HOST and KUBERNETES_SERVICE_PORT. If those environment variables are not available, and this parameter is not set, error will be raised.

api_version (string) (optional)

Kubernetes API version, it includes the API group name and the version, just like the apiVersion in a Kubernetes manifests YAML file. E.g. v1, rbac.authorization.k8s.io/v1, etc.

Default value: v1.

client_cert (string) (optional)

Path to the certificate file for this client.

client_key (string) (optional)

Path to the private key file for this client.

ca_file (string) (optional)

Path to the CA file.

insecure_ssl (bool) (optional)

When set to true, it will ignore inscure HTTPS connections (i.e. it ignores server SSL certificate verification errors).

Default value: false.

secret_dir (string) (optional)

Path of the location where pod's service account's credentials are stored.th of the location where pod's service account's credentials are stored.

Default value: /var/run/secrets/kubernetes.io/serviceaccount.

bearer_token_file (string) (optional)

Path to the file contains the API token. By default it reads from the file "token" in the secret_dir.

<pull> Section (optional) (multiple)

Ths <pull> section is used to define which object to pull from the cluster. One section defines one object.

resource_name (string) (required)

The name of the object to pull. E.g. pods, nodes. This must match api_version. The resource_name is not exactly the same as the name in Kubernetes API. If a name contains multiple words, e.g. daemonsets and replicasets, you will need use underscore _ to separate the words. So daemonsets becomes daemon_sets, and replicasets becomes replica_sets, and so on.

namespace (string) (optional)

Only the objects belong to the namespace specified in this parameter will be pulled. When it's not set, it will pull from all name spaces.

label_selector (string) (optional)

A selector to restrict the list of returned objects by labels.

field_selector (string) (optional)

A selector to restrict the list of returned objects by fields.

interval (time) (optional)

The interval at which the objects will be pulled.

Default value: 15m.

<watch> Section (optional) (multiple)

Ths <watch> section is used to define which object to watch from the cluster. One section defines one object.

This section has exactly the same parameters except interval as the <pull> section does. Check the <pull> section above for details.

<storage> Section (optional) (single)

Defines where to storage checkpoints for watch. Each object the plugin watches, it will record the latest resoruce_version of that object. And when the fluentd restarts, the plugin will send the previously recorded resource_version to the kubernetes watch API.

By default, it uses the memory storage.

See also Storage section configurations and Storage Plugin Overview.

License

See License.

About

This is the Fluentd input plugin which queries Kubernetes API to collect Kubernetes objects (like Nodes, Namespaces, Pods, etc.)

Resources

License

Code of conduct

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Ruby 98.8%
  • Other 1.2%