Docs
Configuration
Sources
kubernetes_logs

kubernetes_logs

logs
 
state
 

Kubernetes version >= 1.22 is required.

Collects Pod logs from Kubernetes Nodes, automatically enriching data with metadata via the Kubernetes API.

This source requires read access to the /var/log/pods directory. When running in a Kubernetes cluster, user should provided a hostPath volume for tracking state of logs.

Providers

There are two provider kubernetes and kubelet, both of them can meet our requirement.

kubernetes

Connecting to kubernetes’s API server and list-watch pods on this node.

type: kubernetes

# The `name` of the Kubernetes `Node` that Vertex runs at.
# Required to filter the `Pod`s to only include the ones with the
# log files accessible locally.
#
# Optional
node_name: null

# Specifies the label selector to filter `Pod`s with, to be used in
# addition to the built-in `vertex.io/exclude` filter
# 
# See: https://kubernetes.io/docs/concepts/overview/working-with-objects/labels/#label-selectors
#
# Optional
label_selector: null

# Specifies the field selector to filter `Pod`s with, to be used in
# addition to the built-in `Node` filter.
# 
# See: https://kubernetes.io/docs/concepts/overview/working-with-objects/field-selectors/#list-of-supported-fields
#
# Optional
field_selector: null

kubelet

Retrieving pod list from kubelet every interval. This variant do not need to connect to API server, so this could mitigate the API server heavy traffic issue for large cluster.

type: kubelet

# HTTP endpoint of the Kubelet's API
#
# Optional
endpoint: http://${NODE_NAME}:10250

# The interval of between each scan
#
# Optional
interval: 10s

Example 

provider: 
  type: kubernetes

  # The `name` of the Kubernetes `Node` that Vertex runs at.
  # Required to filter the `Pod`s to only include the ones with the
  # log files accessible locally.
  #
  # Optional
  node_name: null

  # Specifies the label selector to filter `Pod`s with, to be used in
  # addition to the built-in `vertex.io/exclude` filter
  # 
  # See: https://kubernetes.io/docs/concepts/overview/working-with-objects/labels/#label-selectors
  #
  # Optional
  label_selector: null

  # Specifies the field selector to filter `Pod`s with, to be used in
  # addition to the built-in `Node` filter.
  # 
  # See: https://kubernetes.io/docs/concepts/overview/working-with-objects/field-selectors/#list-of-supported-fields
  #
  # Optional
  field_selector: null

# Configuration for how the events are enriched with Pod metadata.
#
# Optional
fields: 
  pod: 
    name: foo.bar

    namespace: foo.bar

    uid: foo.bar

    ip: foo.bar

    ips: foo.bar

    labels: foo.bar

    annotations: foo.bar

    node_name: foo.bar

  container: 
    name: foo.bar

    image: foo.bar

# Reads from the specified streams
#
# Optional
stream: all # all | stdout | stderr
kubernetes_eventslibvirt